Journal of Experimental Botany, Vol. 64, No. 14, pp. 4271–4287, 2013 doi:10.1093/jxb/ert240 Advance Access publication 7 September, 2013 This paper is available online free of all access charges (see http://jxb.oxfordjournals.org/open_access.html for further details)

Research paper Genome-wide identification and characterization of cadmium-responsive and their target in radish (Raphanus sativus L.) roots

Liang Xu1, Yan Wang1, Lulu Zhai1, Yuanyuan Xu1, Liangju Wang1, Xianwen Zhu2, Yiqin Gong1, Rugang Yu1, Cecilia Limera1 and Liwang Liu1,* 1 National Key Laboratory of Crop Genetics and Germplasm Enhancement, College of Horticulture, Nanjing Agricultural University, Nanjing 210095, PR China 2 Department of Plant Sciences, North Dakota State University, Fargo, ND 58108, USA

* To whom correspondence should be addressed. E-mail: [email protected]

Received 16 April 2013; Revised 26 June 2013; Accepted 28 June 2013

Abstract MicroRNAs (miRNAs) are endogenous non-coding small RNAs that play vital regulatory roles in plant growth, devel- opment, and environmental stress responses. Cadmium (Cd) is a non-essential heavy metal that is highly toxic to living organisms. To date, a number of conserved and non-conserved miRNAs have been identified to be involved in response to Cd stress in some plant species. However, the miRNA-mediated regulatory networks responsive to Cd stress in radish (Raphanus sativus L.) remain largely unexplored. To dissect Cd-responsive miRNAs and their targets systematically at the global level, two small RNA libraries were constructed from Cd-treated and Cd-free roots of radish seedlings. Using Solexa sequencing technology, 93 conserved and 16 non-conserved miRNAs (represent- ing 26 miRNA families) and 28 novel miRNAs (representing 22 miRNA families) were identified. In all, 15 known and eight novel miRNA families were significantly differently regulated under Cd stress. The expression patterns of a set of Cd-responsive miRNAs were validated by quantitative real-time PCR. Based on the radish mRNA transcriptome, 18 and 71 targets for novel and known miRNA families, respectively, were identified by the degradome sequencing approach. Furthermore, a few target transcripts including phytochelatin synthase 1 (PCS1), iron transporter protein, and ABC transporter protein were involved in plant response to Cd stress. This study represents the first transcrip- tome-based analysis of miRNAs and their targets responsive to Cd stress in radish roots. These findings could provide valuable information for functional characterization of miRNAs and their targets in regulatory networks responsive to Cd stress in radish.

Key words: Cadmium stress, degradome, high-throughput sequencing, microRNAs, Raphanus sativus, transcriptome.

Introduction Cadmium (Cd) is a widespread heavy metal pollutant that and development mainly through alterations in photosyn- is highly toxic to living organisms. Vast areas of agricultural thesis, respiration, and nitrogen metabolism, as well as a soils are contaminated with Cd via atmospheric deposi- decrease in water and basic mineral nutrient uptake (Besson- tion and direct application of phosphate fertilizers, animal Bard et al., 2009). Cd can accumulate in the human body manure, sewage sludge, and irrigation water (Uraguchi et al., over time via the food chain, resulting in a risk of chronic 2009; Wiebe et al., 2010). Due to its high mobility and water toxicity to kidney tubules , bone, lungs and some other solubility, Cd2+ is readily taken up by plant roots and can be organs. (Grant et al., 2008; Ishikawaa et al., 2012). Thus, translocated into aerial organs. Cd2+ inhibits plant growth the elucidation of the regulatory mechanisms underlying Cd

© The Author 2013. Published by Oxford University Press on behalf of the Society for Experimental Biology. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. 4272 | Xu et al. uptake, accumulation, translocation, and detoxification is or real-time PCR (qRT-PCR.)-based analysis, researchers becoming an urgent goal. have found that miR171 and miR393 are engaged in response MicroRNAs (miRNAs) are a class of endogenous non- to Cd stress through negatively regulating their targets in coding small RNAs (sRNAs) that regulate gene expression at O. sativa (Huang et al., 2009), Medicago truncatula (Zhou the transcriptional and post-transcriptional levels by guiding et al., 2008), and B. napus (Huang et al., 2010). In rice, a total target mRNA cleavage or translational inhibition (Voinnet, of 19 Cd-regulated miRNAs were identified using miRNA 2009). In plants, the primary miRNA transcripts (pri-miR- microarray analysis (Ding et al., 2011). More recently, high- NAs) are translated from nuclear-encoded MIR genes and throughput sequencing technology has become a reliable cleaved by -like1 (DCL1), leading to the generation of and efficient tool for functional genomics studies such as stem–loop miRNA:miRNA* duplexes known as pre-miRNA genome-wide transcriptome analysis, sRNA sequencing, and (Ruiz-Ferrer and Voinnet, 2009). Thereafter, mature miR- gene expression pattern analysis at single- resolution NAs are bound by the argonaute (AGO) protein to assemble (Morozova and Marra, 2008; McCormick et al., 2011). Using the RNA-induced silencing complex (RISC) (Jones-Rhoades this approach, Zhou et al. (2012a) isolated eight up-regulated et al., 2006). Mature miRNAs guide the RISC to bind to cog- and 10 down-regulated miRNA families in response to Cd nate target genes through either cleaving target mRNAs with stress in B. napus. Additionally, Al3+-mediated miRNAs near perfect complementarity or repressing their translation (four up-regulated and 24 down-regulated) were identified in with lower complementarity (Bartel et al., 2004; Brodersen M. truncatula (Chen et al., 2012b). These findings indicated et al., 2008). Besides regulating a range of essential cellular that a number of miRNAs could play key roles in the regula- and biological processes, a significant fraction of miRNAs tion of plant responses to heavy metal stress. have been shown to play crucial roles in plant responses to Radish (Raphanus sativus L., 2n=2x=18), a major member a variety of abiotic and biotic stresses, such as nutritional of the Brassicaceae family, is an important annual or bien- deficiency (Pant et al., 2009; Liang et al., 2010), drought nial root vegetable crop worldwide (Wang and He, 2005). (Zhou et al., 2010; Li et al., 2011; Wang et al., 2011), salin- With a comparative analysis approach, 48 conserved miR- ity (Macovei and Tuteja, 2012; Li et al., 2013), cold (Zhang NAs belonging to nine miRNA families were isolated from et al., 2009; Barakat et al., 2012), heat (Yu et al., 2011; Chen the expressed sequence tag (EST) databases of R. sativus et al., 2012a), oxidative stress (Sunkar et al., 2006), and heavy (Muvva et al., 2012). Moreover, Xu et al. (2013) first con- metal stress (Ding et al., 2011; Chen et al., 2012b; Zhou et al., structed an sRNA library from radish roots and identified 2012a, b; Zeng et al., 2012). 546 conserved miRNA families as well as 15 novel miRNAs Target validation is a prerequisite to characterize function- using high-throughput sequencing. Nevertheless, there is no ally the biological roles of miRNAs in plants. Modified ′5 - report on systematic identification of Cd-regulated miRNAs rapid amplification of cDNA ends (RACE) has been widely and their target genes at the global level in radish. To investi- employed for cleavage site mapping and target validation gate the roles of miRNAs thoroughly in regulatory networks in some species (Jones-Rhoades et al., 2006). Nevertheless, responsive to Cd stress in radish, two sRNA libraries were this approach is only applicable for target confirmation on constructed from Cd-free and Cd-treated radish roots and a small scale due to it being laborious, low efficiency, and sequenced by the Solexa/Illumina system. The aims of this time-consuming (Li et al., 2013). Recently, high-throughput study were to identify Cd-regulated miRNAs from radish degradome sequencing, a method known as parallel analysis roots and to validate the conserved and new target transcripts of RNA ends (PARE), has been successfully established and for Cd-responsive miRNAs by transcriptome-based degra- adapted to validate miRNA splicing targets in a variety of dome analysis. The outcomes of this study could enhance our plant species, such as Arabidopsis (Addo-Quaye et al., 2008), understanding of the miRNA-mediated regulatory networks Oryza sativa (Li et al., 2010), soybean (Shamimuzzaman responsive to Cd stress in radish, and provide new insights et al., 2012; Zeng et al., 2012), and Brassica napus (Zhou into elucidating the molecular genetic mechanisms underly- et al., 2012a; M.Y. Xu et al., 2012). This technology provides ing plant response to Cd stress. a new efficient strategy to confirm predicted miRNA targets on a large scale in plants (J.H. Yang et al., 2013; M.Y. Xu et al., 2012). Materials and methods Understanding heavy metal-regulated gene expression and regulatory networks is a first critical step to eluci- Plant culture and Cd treatment date the genetic molecular mechanism of metal accumula- Seeds of radish advanced inbred line ‘NAU-RG’ were surface steri- tion and homeostasis (Khraiwesh et al., 2012). Regulation lized in 1.2% NaOCl and germinated at 25 °C for 3 d in the dark. Germinated seeds were grown in plastic pots and cultured in a of gene expression by sRNAs at the transcriptional and/ growth chamber under 14 h light (25 °C)/10 h dark (18 °C). After or post-transcriptional level is a newly discovered mecha- 3 weeks, seedlings were transplanted into a plastic container with nism for plant growth and development and environmental modified half-strength Hoagland nutrient solution as previously described (L. Xu et al., 2012). The seedlings were treated with stress responses (Ruiz-Ferrer and Voinnet, 2009; Li et al., –1 2013). Increasing evidence has revealed that miRNA-medi- 200 mg l CdCl2·2.5H2O for 1, 6, 12, 24, and 48 h, respectively. Seedlings grown in Cd-free solution were treated as controls. Roots ated gene regulation plays a significant role in heavy metal were harvested and immediately frozen in liquid nitrogen and stored regulatory networks (Ding and Zhu, 2009; Khraiwesh et al., at −80 °C. The roots exposed to Cd stress at 12 h were further used 2012; Zhou et al., 2012a, b). Using a direct cloning strategy for Cd-treated sRNA library construction. Identification of Cd-responsive microRNAs and their targets in radish | 4273

Transcriptome and small RNA sequencing The raw data were pre-processed to remove adaptor sequences Total RNA was isolated from Cd-free and Cd-treated roots of rad- and low-quality sequencing reads, and only 20–21 nt sequences ish using Trizol reagent (Invitrogen) according to the manufacturer’s with high quality scores were retained for further subsequent protocols. Equal amounts of RNA from Cd-free and Cd-treated roots analysis. The degradome reads were mapped to the radish refer- (12 h) were pooled for transcriptome and degradome library construc- ence sequences as described above. Perfect matching sequences tion. The transcriptome library was prepared using an Illumina TruSeq were used to identify potentially sliced miRNA targets by the RNA Sample PrepKit following the manufacturer’s instructions. After CleaveLand pipeline (Addo-Quaye et al., 2008, 2009). Alignments removing reads containing only 3′-RNA adaptors and low-quality with no more than five mismatches and no mismatches at the cleav- reads, mRNA transcriptome de novo assembly was performed using age site (between the 10th and 11th nucleotides) were retained and the SOAP2 program (Li et al., 2009). The two sRNA libraries from scored by previously described methods (Allen et al., 2005). The Cd-free (CK) and Cd-treated (Cd200) roots were prepared based on identified targets were grouped into three categories based on the a previously described procedure (Hafner et al., 2008). Briefly, sRNA relative abundance of the degradome signatures at the miRNA tar- fragments ranging from 18 to 30 nucleotides (nt) were separated and get sites (Addo-Quaye et al., 2008). purified by polyacrylamide gel electrophoresis, and ligated to ′5 - and 3′-RNA adaptors by T4 RNA ligase (TaKaRa). The adaptor-ligated sRNAs were subsequently transcribed to single-stranded cDNA using qRT-PCR validation SuperScript II Reverse Transcriptase (Invitrogen). Both sRNA and Total RNAs were isolated from six radish samples (0, 1, 6, 12, 24, and transcriptome sequencing were performed on a Genome Analyzer II 48 h) as described above using Trizol (Invitrogen) following the manu- (Illumina, San Diego, CA, USA). facturer’s protocols. Quantitative RT-PCR (qRT-PCR) for miRNAs were performed using the One Step Primer Script® miRNA cDNA Synthesis Kit (Takara) and SuperScript® III Reverse Transcriptase Analysis of small RNA sequencing data (Invitrogen), respectively. PCRs were carried out in a 20 µl reaction Clean reads were screened from raw sequencing reads by removing mixture consisting of 2 µl of diluted cDNA, 0.2 µM forward and contaminated reads including sequences with 5′-primer contaminants, reverse primer, and 10 µl of 2× SYBR Green PCR Master Mix. The without the inserted tag, with poly(A) tails, either shorter than 15 nt reactions were carried out on an iCycler iQ real-time PCR detection or longer than 30 nt. The remaining unique RNAs were mapped to system (BIO-RAD) at 95 °C for 30 s, and 45 cycles of 95 °C for 5 s, the radish reference sequences containing genomic survey sequences 58 °C for 15 s, and 72 °C for 20 s. The threshold cycle (Ct) was deter- (GSS), EST sequences, and the mRNA transcriptome sequences using mined as the cycle number at which the fluorescence intensity passed the SOAP2 program (Li et al., 2009). Sequences with a perfect match a pre-determined threshold. All reactions were assayed in triplicate, were retained for further analysis. Sequences matching non-coding and 5.8S rRNA was used as the reference gene. The data were statis- RNAs included rRNAs, tRNAs, small nuclear RNAs (snRNAs), and tically analysed with SAS Version 9.0 software (SAS Institute, Cary, small nucleolar RNAs (snoRNAs) in the Rfam (http://www.sanger. NC, USA) using Duncan’s multiple range test at the P<0.05 level of ac.uk/Software/Rfam) and NCBI GenBank (http://www.ncbi.nih. significance. The primers for the miRNA qRT–PCR are shown in gov/GenBank/) databases (Benson et al., 2011) were removed. The Supplementary Table S1 available at JXB online. remaining unique sequences were aligned with known miRNAs from miRBase 19.0 (http://www.mirbase.org/index.shtml) with a maximum of two mismatches allowed. Mireap software (https://sourceforge.net/ projects/mireap/) was used to predict novel miRNAs from the remain- Results ing unknown sRNAs. Basic criteria (Meyers et al., 2008) were used for screening the potential novel miRNAs. The stem–loop structures of Overview of transcriptome and small RNA sequencing pre-miRNAs were constructed by Mfold (Zuker, 2003). In order to obtain global mRNAs from radish, an mRNA library constructed from total RNAs of radish root was Differential expression analysis of miRNAs under Cd stress sequenced by the Illumina/Solexa system, resulting in the The frequency of miRNAs from two libraries was normalized to generation of ~71.95 million raw reads. After the removal of 1 million by total clean reads of miRNAs in each sample (normal- poly(A) tails, short and low-quality tags, and adaptor contam- ized expression=actual miRNA count/total count of clean reads×1 ination, 66 110 340 clean reads were obtained. Furthermore, a 000 000). If the normalized read count of a given miRNA is zero, the expression value was modified to 0.001 for further analysis. The total of 150 455 contigs with an average length of 299 nt were fold change between the Cd200 and CK library was calculated as: produced, which were further assembled into 73 084 unigenes fold change=log2 (Cd200/CK). The miRNAs with fold changes >2 with pair-end annotation. These assembled unigenes varied or <0.5 and with P ≤ 0.05 were considered to be up-regulated or from 200 nt to 8482 nt in length, with an average length of down-regulated in response to Cd stress, respectively. The P-value 763 nt. This mRNA transcriptome database, combined with was calculated according to previously established methods (Man et al., 2000; Li et al., 2011). the available GSS and EST sequences released into NCBI databases, make up the radish reference sequences for the prediction of known and novel miRNAs in radish. Construction and analysis of degradome libraries To identify miRNAs responsive to Cd in radish, two sRNA Poly(A) RNA was isolated from 200 µg of total RNA using the libraries were constructed from Cd-free (CK) and Cd-treated Oligotex mRNA mini kit (Qiagen). The degradome library was constructed following a previously described method (German (Cd200) roots of radish seedlings, and sequenced by a Solexa/ et al., 2008, 2009). In brief, polyadenylated transcripts possessing Illumina analyzer. In total, 29.27 million raw reads represent- 5′-monophosphates were ligated to an RNA oligonucleotide adap- ing 7.7 million unique sequences were generated in two sRNA tor containing an MmeI recognition site using T4 DNA ligase. libraries (Table 1; Supplementary Fig. S1 at JXB online). Subsequently, first-strand cDNA was generated and amplified with After removing low-quality tags and adaptor contaminations, six PCR cycles. Thereafter, the PCR product was digested with MmeI and ligated to a 3′ double-stranded DNA adaptor. Finally, 15 779 290 (representing 4 615 663 unique sequences) and 13 the ligated products were amplified with 20 PCR cycles, gel purified, 495 250 (representing 4 071 113 unique sequences) clean reads and sequenced on an Illumina Genome Analyzer II. ranging from 15 nt to 30 nt were obtained for CK and Cd200 4274 | Xu et al.

Table 1. Summary of common and specific sequences between CK and Cd200 sRNA libraries

Class Unique sRNAs Percentage Total sRNAs Percentage

Total_sRNAs 7 703 916 100.00% 29 274 540 100.00% CK&Cd200 982 860 12.76% 20 458 468 69.88% CK_specific 3 632 803 47.16% 5 042 275 17.22% Cd200_specific 3 088 253 40.09% 3 773 797 12.89% libraries, respectively (Table 2; Supplementary Table S2). these tags were significantly different between the two librar- A total of 720 362 (CK) and 595 759 (Cd200) unique sequences ies. The relative abundances of both 21 nt and 22 nt sRNAs in were successfully mapped to the radish reference sequences, the Cd200 library were markedly lower than those in the CK respectively. Thereafter, the non-coding RNAs, including library, suggesting that both of the miRNA classes might be rRNAs, tRNAs, snRNAs, and snoRNAs, were annotated and repressed under Cd stress. Nevertheless, the abundance of 23 removed. Querying the remaining sequences against miRbase nt sRNAs was relatively larger in the Cd200 compared with 19.0 identified 31 280 (CK) and 28 788 (Cd200) unique reads the CK library, indicating that the 23 nt sRNAs may play matchings known miRNAs. There remained 4 446 875 (CK) more important roles in plant response to Cd stress. and 3 862 789 (Cd200) unannotated unique sRNA sequences to be screened for identification of novel miRNAs (Table 3). Identification of known miRNA families in radish The majority of total sRNA reads ranged from 20 nt to 24 nt in length, and 21 nt and 24 nt sequences were dominant Based on mapping the unique sRNA sequences to miRBase in both libraries (Fig. 1). The 24 nt sRNAs were the most 19.0 with a maximum of two mismatches, 42 308 and 38 abundant, making up 31.53% (CK) and 29.59% (Cd200) 313 known miRNA sequences were identified in the CK and of the total sequence reads. This result was consistent with Cd200 libraries, respectively (Table 2). A total of 93 unique that previously reported for other plant species, such as sequences belonging to 16 conserved miRNA families were Arabidopsis (Hsieh et al., 2009), Oryza (Jeong et al., 2011), identified in the two libraries (Table 4; Supplementary Table Medicago (Lelandais-Brière et al., 2009; Wang et al., 2011), S3 at JXB online). Among these conserved miRNA families, and Populus (Chen et al., 2012a). Moreover, the fractions of miR156, miR166, miR158, and miR164 contained 13, 11,

Table 2. Statistical analysis of sequencing reads from the CK and Cd200 sRNA libraries in radish

Total sRNAs Unique sRNAs

CK Raw reads 15 983 864 Clean reads 15 779 290 4 615 663 Mapped to genomic 7 396 733 720 362 Match known miRNAs 1 771 311 31 280 Unannotated sRNAs 10 536 202 4 446 875 Cd200 Raw reads 13 664 651 Clean reads 13 495 250 4 071 113 Mapped to genomic 6 592 663 595 759 Match known miRNAs 1 152 100 28 788 Unannotated sRNAs 8 038 038 3 862 789

Libraries for CK and Cd200 were constructed from R. sativus roots exposed to solution without Cd and containing 200mg l–1 Cd (pH 5.5), respectively.

Table 3. Distribution of small RNAs among different categories in radish

Category CK Cd200 Unique sRNAs Total sRNAs Unique sRNAs Total sRNAs

Total small RNAs 4 615 663 (100%) 15 779 290 (100%) 4 071 113 (100%) 13 495 250 (100%) miRNA 31 280 (0.68%) 1 771 311 (11.23%) 28 788 (0.71%) 1 152 100 (8.54%) rRNA 118 421 (2.57%) 3 186 388 (20.19%) 155 413 (3.78%) 3 328 122 (24.66%) snRNA 6539 (0.14%) 39 651 (0.25%) 5233 (0.13%) 21 793 (0.16%) snoRNA 2354 (0.05%) 5761 (0.04%) 1796 (0.04%) 4020 (0.03%) tRNA 10 194 (0.22%) 239 977 (1.52%) 17 094 (0.42%) 951 177 (7.05%) Unannotated 4 446 875 (96.34%) 10 536 202 (66.77%) 3 862 789 (94.88%) 8 038 038 (59.56%) Identification of Cd-responsive microRNAs and their targets in radish | 4275

Fig. 1. Size distribution of small RNAs in Cd-free (CK) and Cd-treated (Cd200) libraries from radish roots. (This figure is available in colour at JXB online.) nine, and eight members, respectively; whereas three miRNA both libraries. The majority of these non-conserved miRNA families including miR319, miR397, and miR398 had only families comprised only one member, whereas three miRNA one member. Moreover, 15 unique sequences belonging to families (miR824, miR825, and miR827) and miR403 con- 10 non-conserved miRNA families were also detected from tained two and three members, respectively (Fig. 2A).

Table 4. Known miRNA families and their transcript abundance identified from CK and Cd200 libraries in radish

Family No. of members miRNA reads Total miRNA reads Ratio (Cd200/CK) CK Cd200

Conserved miRNA miR156 13 291 882 175 380 467 262 0.60 miR158 9 114 972 2904 178 876 0.03 miR159 4 2081 1233 3314 0.59 miR160 2 45 122 167 2.71 miR162 2 1861 742 2603 0.40 miR164 8 54 932 36 291 91 223 0.66 miR165 4 6986 1707 8693 0.24 miR166 11 189 669 57 169 246 838 0.30 miR167 7 2628 12 794 15 422 4.87 miR168 3 72 829 61 084 133 913 0.84 miR169 4 783 604 1387 0.77 miR172 3 2103 1933 4036 0.92 miR319 1 7241 1572 8813 0.22 miR390 3 11 993 3748 15 741 0.31 miR391 2 2157 1228 3385 0.57 miR393 2 30 224 254 7.47 miR395 4 210 477 687 2.27 miR396 2 950 21 753 5773 22.90 miR397 1 3252 3013 6265 0.93 miR398 1 164 753 917 4.59 miR399 2 23 55 78 2.39 miR408 5 417 843 201 889 619 732 0.48 Non-conserved miRNA miR403 3 1168 659 1 827 0.56 miR824 2 1178 568 1746 0.48 miR825 2 115 42 157 0.37 miR827 2 36 472 508 13.11 miR854 1 485 145 630 0.30 miR857 1 1598 3162 5760 1.98 miR1023 1 5 8 13 1.6 miR1442 1 183 26 209 0.14 miR2111 1 6 32 38 5.33 miR5021 1 15 76 91 5.07 4276 | Xu et al.

Fig. 2. Sizes and abundance of identified known miRNA families from radish. (A) Distribution of known miRNA family size in radish. (B) Counts of each known miRNA family in radish. (This figure is available in colour at JXB online.)

The read number differed drastically among the 26 known is a prerequisite for the annotation of a new miRNA. In miRNA families. A few conserved miRNA families, such as total, 28 novel miRNAs representing 22 unique miRNA miR156, miR166, miR168, and miR408, showed extraordi- sequences were identified with complementary miRNA* narily high expression levels in both libraries. miR408 was the as potentially novel miRNAs. The majority of these novel most abundant, with 417 843 (CK) and 201 889 (Cd200) reads miRNAs were produced from one locus, whereas rsa- accounting for 35.3% and 34.1% of all conserved miRNA reads, miRn5, rsa-miRn9, and rsa-miRn14 were generated from respectively (Fig. 2B); miR166 was the second most abundant two, three, and four loci, respectively (Table 5). The length in both libraries. Several miRNA families including miR164, of the 22 novel unique miRNAs ranged from 21 nt to 23 nt, miR172, miR319, miR390, and miR397 had moderate abun- with the majority being 21 nt long (17; 77.3%). The length dance. Nevertheless, a few non-conserved miRNA families, of the novel miRNA precursors varied from 72 nt to 208 such as miR827, miR1442, miR2111, and miR5021, showed nt, with an average of 134 nt. The average minimum free relatively lower expression levels, which were represented by energy (MFE) value was –46.5 kcal mol–1, with a range <100 reads in the CK or Cd200 library (Supplementary Table of –86.7 kcal mol–1 to –18.2 kcal mol–1. The secondary S3 at JXB online). Moreover, different members in the same structures of the 22 novel miRNA precursors are shown miRNA family also displayed significantly different expres- in Supplementary Fig. S2 at JXB online. The expression sion levels. For instance, the abundance of miR156 members levels of novel miRNAs and their miRNA* varied sig- varied from 10 to 197 470 reads. These results indicated that nificantly between the CK and Cd200 library (Table 5; the expression level of conserved and non-conserved miRNAs Supplementary Table S4 at JXB online), demonstrating the varies dramatically in radish, which was in agreement with pre- complexity of miRNA generation and expression under Cd vious studies reporting that the non-conserved miRNAs were stress in radish. represented by relatively lower levels than conserved miRNAs (Jeong et al., 2011; Yu et al., 2012; J.H. Yang et al., 2013). Identification of Cd-responsive miRNAs in radish

Identification of novel candidate miRNAs in radish To identify differently regulated miRNAs under Cd stress in radish, a differential expression analysis of miRNAs Based on the criteria for the annotation of novel miRNAs between the CK and Cd200 library was performed. A total (Meyers et al., 2008), a characteristic stem–loop precursor of 22 known miRNAs belonging to 15 miRNA families Identification of Cd-responsive microRNAs and their targets in radish | 4277

Table 5. Novel miRNAs and their transcript abundance identified from CK and Cd200 sRNA libraries in radish

miRNA Mature sequence (5′–3′) Size LP MFE miRNA reads Log2 Total Total Loci (Cd200/CK) miRNA miRNA* CK Cd200 reads reads rsa-miRn1 UCGCUUGGUGCAGGUCGGGAA 21 141 –72.20 8263 9468 0.42 17731 199 1 rsa-miRn2 UGAAGCUGCCAGCAUGAUCUA 21 118 –48.00 1524 4358 1.74 5882 16 1 rsa-miRn3 AAGCUAGAGACUUAAAACAAG 21 139 –23.95 65 8 –2.80 73 19 1 rsa-miRn4 GCGUAUGAGGAGCCAAGCAUA 21 106 –49.30 1045 574 –0.64 1619 168 1 rsa-miRn5 GUGGUGACGGUGGUGGUGCGA 21 99 –36.40 5 0 –8.31 5 1 2 rsa-miRn6 CAGGGAACAAGCAGAGCAUGG 21 110 –46.10 3972 2741 –0.31 6713 1557 1 rsa-miRn7 AUAUACUGAAGUUUAUACUCU 21 208 –37.00 58 362 2.87 420 36 1 rsa-miRn8 GUAUGAGGAGCCAAGCAUAU 21 107 –46.60 1137 563 –0.79 1700 198 1 rsa-miRn9 GUACGACGAAGAUGAGCCGACA 23 110 –19.60 36 25 –0.30 61 5 3 rsa-miRn10 UGGAGGCAGCGGUUCAUCGAUC 22 140 –45.70 378 452 0.48 830 2370 1 rsa-miRn11 GCUCAAGAAAGCUGUGGGAAA 21 147 –39.04 0 242 14.18 242 72 1 rsa-miRn12 AAACUGCCUAAACAAACAUAUC 22 171 –40.44 56 38 –11.87 59 14 1 rsa-miRn13 GCUGGAGGCAGCGGUUCAUCGAUC 23 142 –46.80 827 2326 1.72 3153 300 1 rsa-miRn14 AGAUGACAGUGAGGCUUCUUA 21 108 –18.20 0 28 11.02 28 1 4 rsa-miRn15 CCCGCCUUGCAUCAACUGAAU 21 137 –66.30 222 34 –2.48 256 19 1 rsa-miRn16 UCGCUUGGUGCAGGUCGGGAC 21 142 –73.60 12154 14602 1.31 37886 3276 1 rsa-miRn17 UUGGACUGAAGGGAGCUCCUU 21 201 –86.70 8 0 –8.99 8 1 1 rsa-miRn18 CGCCUUGCAUCAACUGAAUCA 21 108 –56.70 1721 1423 1.71 6544 170 1 rsa-miRn19 AUGGAUGUAUGAUAUGAUGGA 21 136 –41.40 0 21 10.60 21 3 1 rsa-miRn20 GGAAUGUUGUUUGGCUCGAAG 21 72 –20.00 10 27 1.66 37 1 1 rsa-miRn21 UCGGACCAGGCUUCAUUCCCC 21 133 –67.20 36853 53357 0.76 90210 436 1 rsa-miRn22 AAGCUGCCAGCGUGAUCUUAAC 22 101 –41.80 2638 3280 0.54 5918 2018 1

LP (nt), The length of precursor; MFE (kcal mol–1), minimal folding free energy. and 11 novel miRNAs belonging to eight miRNA families only detected in the CK library, whereas three miRNAs were identified to be differentially expressed in response (rsa-miRn11, rsa-miRn14, and rsa-miRn19) were detected to Cd stress (Supplementary Table S5 at JXB online). The only in the Cd200 library (Table 6), suggesting that these majority of these Cd-responsive miRNAs were down- novel miRNAs might be induced or repressed under Cd regulated in the Cd200 library compared with the CK stress in radish. The results revealed that these differentially library, whereas five known miRNAs (miR167a, miR167d, regulated miRNAs may play crucial roles in response to Cd miR396a, miR396b, and miR398) and three novel miRNAs stress in radish. (rsa-miRn3, rsa-miRn11, and rsa-miRn19) showed up- regulated patterns (Fig. 3A, B), suggesting that the down- Degradome sequencing and data summary regulation of miRNAs might play more important roles in plant responses to Cd stress (Khraiwesh et al., 2012). The Although a few miRNAs targets from radish have been pre- miRNAs with the greatest change in expression levels were viously predicted (Muvva et al., 2012; Xu et al., 2013), no rsa-miRn11 and rsa-miRn19, with ratios of 14.18-fold miRNA targets for radish have yet been experimentally char- and 10.60-fold, respectively. Among these differentially acterized. In this study, the miRNA-cleaved mRNAs in radish expressed novel miRNAs, rsa-miRn5 and rsa-miRn17 were were first systematically confirmed using the high-throughput

Table 6. The novel cadmium-responsive miRNAs identified from CK and Cd200 libraries in radish

miRNA CK Cd200 Fold change log2 Regulation P-value Significance (Cd200/CK) Count Normalized Count Normalized rsa-miRn3 65 4.1191 8 0.5926 –2.80 Down-regulated 0 ** rsa-miRn4 1045 66.2231 574 42.5185 –0.64 Down-regulated 0 rsa-miRn5 5 0.3169 0 0.0010 –8.31 Down-regulated 8.07E-10 ** rsa-miRn11 0 0.0010 242 17.9259 14.13 Up-regulated 3.61E-09 ** rsa-miRn14 0 0.0010 28 2.0741 11.02 Up-regulated 0 ** rsa-miRn15 222 14.0684 34 2.5185 –2.48 Down-regulated 6.73E-10 ** rsa-miRn17 8 0.5070 0 0.0010 –8.99 Down-regulated 1.57E-16 ** rsa-miRn19 0 0.0010 21 1.5556 10.60 Up-regulated 8.15E-07 ** 4278 | Xu et al.

Fig. 3. Validation and comparative relative expression of differentially expressed known (A) and novel (B) miRNAs between the CK and Cd200 libraries in radish by qRT-PCR. (This figure is available in colour at JXB online.) degradome sequencing technology. In total, ~25.87 million Identification and annotation of targets for raw reads were obtained, which resulted in generating 25 741 radish miRNAs 860 clean reads representing 8 037 807 unique reads from the mixed degradome library. A total of 6 287 867 unique reads Through the degradome sequencing approach, a total of 55 representing 78.2% of the total unique sequences were suc- and 16 sliced targets for 17 conserved and seven non-­conserved cessfully mapped to the R. sativus reference sequences. The miRNAs were identified, respectively. Among these 71 tar- CleaveLane pipeline was adopted to identify the sliced tar- gets, 45 targets fell into category I, whereas 16 and 10 targets gets for the known miRNAs and novel miRNA candidates belonged to category II and III, respectively. A number of the (Addo-Quaye et al., 2009). The abundance of the sequenced identified targets for the known radish miRNAs were tran- tags was plotted on each transcript. The sliced target tran- scription factors, such as the auxin response factor (ARF) scripts were grouped into three categories based on the rela- family, AP2-type transcription factor, NAC (No Apical tive abundance of the degradome signatures at the target Meristem) domain transcription factor, zinc finger proteins, mRNA sites (Fig. 4). Based on the previous criteria, category squamosa promoter-binding protein (SBP), and TCP family I is defined as abundance at a position where the most abun- transcription factor (Supplementary Table S6 at JXB online), dant reads match the transcript, and with only one maximum which could play essential regulatory roles in various aspects on the transcript with more than one raw read at the posi- of plant growth and development. Moreover, many miRNA tion. Category II is described as abundance at the position targets were involved in a wide range of biological processes, less than the maximum but higher than the median for the including those encoding F-box family protein, protein–­ transcript and with more than one raw read at the position. protein interaction family proteins, transmembrane protein, Category III is comprised of all the other transcripts sliced and transducin family protein. In addition, a few transcripts by miRNAs. targeted by known miRNAs were involved in plant response Identification of Cd-responsive microRNAs and their targets in radish | 4279

Fig. 4. Target plots (t-plots) of miRNA targets in different categories confirmed by degradome sequencing. The normalized signature abundance throughout the length of the indicated transcripts is shown. Representative t-plots for class I (A), class II (B), and class III (C) categories are shown. Blue arrows indicate signatures consistent with miRNA-directed cleavage. The solid lines and dot in miRNA:mRNA alignments indicate matched RNA base pairs and GU mismatch, respectively. On the left of the t-plots, the cleavage sites and normalized signature abundance are shown in the left and right column, respectively. to biotic and abiotic stresses, such as those encoding leucine- were also identified by degradome sequencing analysis. In total, rich repeat (LRR) domain-containing proteins (CL2282. 18 target genes for 10 novel miRNAs were successfully identi- Contig9 and EY922772; miR166 and miR390), heat shock fied (Table 8). Among these candidate targets, 11 targets fell into protein (Contig132627, miR396), ABC transporter protein category I, whereas five and two targets belonged to category (comp31028, miR159), and iron transporter protein (CL2282. II and III, respectively. Compared with the known miRNAs, Contig10, miR166) (Fig. 5). These results indicated that the low splicing frequency targets were found for the novel radish conserved and non-conserved miRNAs might play vital roles miRNAs, which was consistent with previous studies for some in diverse biological processes under Cd stress in radish. Brassica species (M.Y. Xu et al., 2012; J.H. Yang et al., 2013). Although the novel miRNAs were sequenced at a relatively Several targets were transcription factors, whereas some other lower level as compared with the known miRNAs, their targets targets appeared to be involved in signal transduction pathways 4280 | Xu et al. including ARFs, zinc finger proteins, and protein–protein inter- qRT-PCR validation of Cd-responsive miRNAs action domain family protein (Table 8). Moreover, a few targets could be involved in stress response, such as disease resist- To confirm the Solexa sequencing results and study the ance protein (Unigene17426, rsa-miRn11) and cold-inducible dynamic expression patterns of the Cd-responsive miRNAs plasma membrane protein (Unigene13282, rsa-miRn17). In at different Cd treatment time points (0, 1, 6, 12, 24, and addition, a small proportion of these novel miRNAs targeted 48 h) in radish, the expression patterns of six known and four some genes with no functional annotations. These findings sug- novel (two miRNAs with their miRNA*s) Cd-responsive gested that the novel miRNAs may play special roles in some miRNAs were validated by qRT-PCR (Fig. 6). As expected, biological and developmental processes under Cd stress in rad- the qRT-PCR data showed a high degree of agreement ish. However, no targets were discovered for 12 novel miRNAs with the expression profiles obtained by sRNA sequencing in the degradome sequencing data, partly due to the limited between the CK and Cd200 libraries under Cd treatment number of accessible radish reference sequences. at 12 h. For known miRNAs, transcripts of miR157a and

Fig. 5. Target plots (t-plots) for a set of Cd-responsive miRNA targets confirmed by degradome sequencing. (A) rsa-miR156, (B) rsa- miR159, (C) rsa-miR166, (D) rsa-miR393. Blue arrows indicate signatures consistent with miRNA-directed cleavage. The solid lines and dot in miRNA:mRNA alignments indicate matched RNA base pairs and GU mismatch, respectively. Identification of Cd-responsive microRNAs and their targets in radish | 4281

Table 7. Identified targets for known Cd-responsive miRNAs in radish miRNA Target ID Cleavage site Category TP100M Score Target annotation miR156 Contig20162 111 I 22.562 1 Squamosa promoter-binding protein Contig44431 487 I 16.338 1 Squamosa promoter-binding protein Contig27694 126 I 49.403 3 Glutathione S-transferase 5 (GST5) miR157 Rsa#S42048961 357 II 119.034 2 Squamosa promoter-binding protein miR158 EV566868 183 I 33.065 4 Bax inhibitor-like protein miR159 EY896930 325 I 192.166 3 MYB family transcription factor Rsa#S42022839 268 III 31.898 3 MYB family transcription factor Contig31028 106 I 48.236 2.5 ABC transporter family protein miR166 CL14723.Contig4 798 I 153.655 2.5 HD-ZIP transcription factor CL2282.Contig1 888 I 23.729 3 Ring zinc finger protein CL3287.Contig3 798 I 23.729 2.5 Iron transporter protein CL528.Contig1 699 I 23.729 2.5 Homeodomain-leucine zipper protein CL7584.Contig4 888 II 23.729 2.5 Leucine-rich repeat (LRR) domain-containing proteins CL14723.Contig1 663 I 243.903 2.5 Peptide chain release factor subunit CL2282.Contig8 681 I 23.729 3 Unknown protein miR169 EX750227 370 II 12.448 3 AP2 domain-containing transcription factor CL4257 256 I 27.23 2.5 CCAAT-binding transcription factor Contig7182 82 I 56.794 3 CCAAT-binding transcription factor miR393 Contig26439 256 I 222.508 2 Phytochelatin synthase 1 miR396 CL8017.Contig3 445 I 16.727 3.5 Auxin response factor 8 (ARF8) Contig132627 136 III 45.124 3.5 Heat shock protein 90 Unigene34651 136 III 45.124 3.5 Uncharacterized protein FD946893 185 I 27.23 3.5 Hypothetical protein Unigene23674 490 I 17.894 4 Drought-stressed protein FD986768 531 III 17.894 4 Drought-stressed protein FY448028 648 II 16.727 3.5 Uncharacterized protein CL8017.Contig1 725 I 16.727 2.5 Transmembrane protein-related miR408 Unigene1805 426 II 33.454 3 Ascorbate oxidase Contig13527 78 I 33.454 2.5 Auxin response factor 6(ARF6) miR827 EY923470 114 II 24.896 1.5 Ring finger family protein miR2111 EV552506 301 II 19.061 4 60S ribosomal protein L26-1

Detailed information of targets for all the known miRNAs cab be found in Supplementary Table S6 at JXB online. Categories are defined according to Addo-Quaye et al. (2008). Cleavage site, nucleotide number from the 5′ end of cDNA; TP100M, transcripts per 100 million. miR159a were down-regulated, gradually declined at 1, However, rsa-miRn11 was expressed at far greater levels 6, and 12 h, and then steadily increased at 24 h and 48 h. than rsa-miRn11*, whereas rsa-miRn17* was expressed at miR166a showed a similar down-regulated expression pat- markedly higher levels than rsa-miRn17 (Fig. 6). tern, except that it slightly increased at 1 h of Cd stress. miR167a and miR2111a gradually increased and peaked at 12 h, then sharply decreased and remained at a similar level Discussion at 24 h and 48 h. miR319 was down-regulated and remained at an extremely low expression level at all time points (1, 6, Cd is a widespread toxicant and poses a potential significant 12, 24, and 48 h) (Fig. 6). threat to human health. Recently, high-throughput sequenc- For these novel Cd-responsive miRNAs, rsa-miRn3 was ing has been widely applied to identify comprehensively down-regulated at 1, 6, and 12 h, and then reached a rela- plant miRNAs responsive to abiotic stress, which could tively high expression level at 48 h. rsa-miRn5 was slightly provide powerful information for better understanding of up-regulated at 1 h, gradually declined at 6 h and 12 h, and miRNA-mediated regulatory networks in plant response to then reached a relatively low expression level at 24 h and various stresses (Ruiz-Ferrer and Voinnet, 2009; Khraiwesh 48 h. Moreover, the increased expression profiles of rsa- et al., 2012). Although many functional studies have shown miRn11 and rsa-miRn17 were accompanied by a decrease that several known miRNAs were involved in Cd stress in in rsa-miRn11* and rsa-miRn17*, respectively, indicating some plant species (Ding et al., 2011; Khraiwesh et al., 2012; that the relative expression patterns of novel miRNAs and Zhou et al., 2012a), no study on comprehensive identification their miRNA*s were opposite under Cd stress in radish. of Cd-regulated miRNAs and their targets has yet been 4282 | Xu et al.

Table 8. Candidate targets for novel Cd-responsive miRNAs in radish miRNA Target ID Cleavage site Category TP100M Score Target annotation rsa-miRn1 Contig19272 733 I 18.234 4 Auxin response factor 16 (ARF16) CL2405.Contig1 113 I 18.234 4 Auxin response factor16 (ARF16) rsa-miRn3 Contig81072 75 II 35.628 2.5 Protein–protein interaction domain family protein Contig11248 461 II 35.628 2.5 Protein–protein interaction domain family protein rsa-miRn4 Unigene18150 158 I 21.792 3 Zinc finger-containing protein rsa-miRn8 Contig23480 856 I 31.56 3.5 MYB family transcription factor Contig21929 12 II 31.56 3.5 MYB family transcription factor Unigene24462 334 II 31.56 3.5 Leucine-rich repeat (LRR) domain-containing proteins rsa-miRn10 CL10523.Contig1 126 I 10.862 3 AP2 domain-containing transcription factor rsa-miRn11 Unigene17426 62 I 45.68 3.5 Disease resistance protein rsa-miRn13 CL1028.Contig2 215 III 19.824 3 CCAAT-binding transcription factor Contig2969 83 I 19.824 3 Uncharacterized protein rsa-miRn17 Unigene13282 256 II 106.52 1 Cold-inducible plasma membrane protein rsa-miRn19 Contig10826 32 I 86.583 4 Zinc finger-containing protein Unigene2104 306 I 86.583 4 Zinc finger-containing protein CL3587 78 I 86.583 4 Uncharacterized protein rsa-miRn21 Contig14638 137 III 42.358 3.5 Unknown protein CL6275.Contig2 94 I 42.358 3.5 Calcium-dependent protein kinase

conducted in radish. In this study, using high-throughput non-conserved miRNAs. These results were consistent with Solexa technology, ~16 and 14 million sRNA raw reads were previous reports in A. thaliana (Rajagopalan et al., 2006), obtained from the CK and Cd200 sRNA library of rad- M. truncaula (Chen et al., 2012b), B. napus (Zhou et al., ish root, respectively. Due to the unavailability of the full 2012a), and cucumber (Martínez et al., 2011). Similar to genome sequences of R. sativus and limited numbers of GSS previous reports, a large proportion of targets for conserved and EST sequences in the public NCBI databases, an mRNA miRNAs were transcription factors that regulate develop- transcriptome from radish roots was sequenced for use as a mental and growth processes, and fewer were associated reference sequence, which could provide more valuable infor- with signal transduction and response to environmental mation for prediction of candidate pri-miRNA sequences. stress (Voinnet, 2009; Cuperus et al., 2011). To the authors’ knowledge, this is the first report on the sys- An increasing number of studies have revealed a range tematic transcriptome-based identification and characteriza- of conserved and non-conserved miRNAs that were dif- tion of miRNAs and their targets in response to Cd stress in ferentially regulated under Cd stress in rice (Huang et al., radish. 2009; Ding et al., 2011), M. truncatula (Zhou et al., 2008), and B. napus (Huang et al., 2010; Zhou et al., 2012a), Characteristics of R. sativus miRNAs responsive to which greatly advanced our understanding of the regula- Cd stress tory roles of plant miRNAs in adaptive response to heavy metal stresses (Khraiwesh et al., 2012). In rice, using the Previous studies have shown that many known miRNAs miRNA microarray approach, a total of nine miRNA fami- were highly evolutionarily conserved in the plant kingdom, lies (miR156, miR162, miR168, miR166, miR171, miR396, whereas some other miRNAs were only conserved in one miR390, miR1432, and miR444) were down-regulated or a few plant species (Martínez et al., 2011; Chen et al., under Cd stress, whereas only miR528 was significantly 2012a). In the current study, the majority of conserved miR- up-regulated (Ding et al., 2011). In M. truncatula, Zhou NAs were represented with high or moderate abundance, et al. (2008) reported that miR171, miR319, miR393, and whereas several non-conserved miRNAs were represented miR529 were up-regulated, whereas miR166 and miR398 by <100 reads. This observation revealed that the conserved were down-regulated under Cd stress using a qRT-PCR- miRNAs showed relatively higher expression abundance based assay. Moreover, seven up-regulated miRNA fami- than the non-conserved miRNAs (M.Y. Xu et al., 2012; lies (miR158, miR161, miR172, miR398, miR400, miR857, J.H. Yang et al., 2013). As expected, the majority of these and miR1885) and 10 down-regulated miRNA families highly conserved miRNAs in a variety of plant species were (miR159, miR162, miR164, miR171, miR319, miR394, also the most abundant classes in the radish root miRNA miR395, miR396, miR858, and miR2111) were identified library (Table 4). Moreover, 13 out of 16 evolutionarily in response to Cd stress in B. napus (Zhou et al., 2012a). conserved miRNAs were represented by more than one In the present study, a total of 22 known miRNAs and member, whereas the majority of non-conserved miRNAs eight unique novel miRNAs were identified to be respon- were represented only by a single member, suggesting that sive to Cd stress (Table 6; Supplementary Table S5 at JXB the conserved miRNAs had more family members than the online). As expected, most of these previously identified Identification of Cd-responsive microRNAs and their targets in radish | 4283

Fig. 6. qRT-PCR validation of known Cd-responsive miRNAs in radish. (A–F) and (G–L) Known and novel miRNAs, respectively. Small RNAs (<200 nt) were extracted from radish root treated or not with Cd (0, 1, 6, 12, 24, and 48 h). The amount of expression was normalized to the level of 5.8S rRNA. The normalized miRNA levels at 0 h were arbitrarily set to 1. Different letters indicate significant differences at P<0.05 according to Duncan’s multiple range tests.

Cd-responsive miRNAs also showed differential regulation reported Cd-regulated miRNA families, such as miR161, under Cd stress. For instance, miR156, miR159, miR166, miR162, and miR172, were not significantly differentially and miR319 were down-regulated, whereas miR398 and regulated under Cd stress in radish, while miR396a and miR857 were up-regulated, which was in agreement with miR396b were up-regulated under Cd stress. Thus, further previous studies in B. napus or rice (Ding et al., 2011; Zhou functional studies are needed to validate the precise regula- et al., 2012a). However, it was found that some previously tory roles of these miRNAs in plant response to Cd stress. 4284 | Xu et al. miRNA-mediated regulatory networks responsive to transporter protein could play critical roles in the regulatory Cd stress networks of Cd uptake, accumulation, translocation, and detoxification in radish. Therefore, it is reasonable that the Recently, high-throughput degradome sequencing has been miRNA-mediated gene regulation could serve as an impor- shown to be a valuable and efficient approach to validate and tant mode of regulatory networks responsive to Cd stress characterize target genes of miRNAs in a variety of plant in plants. Nevertheless, further functional analysis of these species (German et al., 2008; Shamimuzzaman and Vodkin, Cd-responsive miRNAs and their corresponding targets is 2012). Although a wide range of target genes for conserved still necessary for better understanding of the miRNA-medi- and non-conserved miRNAs have been previously predicted ated regulatory mechanisms underlying plant response to Cd in several vegetable crops, only a few target genes have been stress. confirmed experimentally (M.Y. Xu et al., 2012; Yu et al., Although several candidate target transcripts for a propor- 2012). This study represents the first transcriptome-based tion of known miRNAs were successfully identified under Cd analysis of miRNA targets responsive to Cd stress in rad- stress by degradome analysis, there were no detectable sliced tar- ish by degradome sequencing analysis. Consistent with some get transcripts identified for a few conserved miRNAs (miR162, previous studies, several identified radish miRNA targets miR167, miR319, miR391, and miR397) and non-conserved belong to a variety of gene families of transcription factors, miRNAs (miR857, miR1442, and miR5021). Similar results including ARFs, MYBs, SBPs, AP2-like factors, and NAC- were found in M. truncatula (Zhou et al., 2012b), soybean domain proteins (Tables 7, 8), which were found to be highly (Shamimuzzaman and Vodkin, 2012), cotton (X.Y. Yang et al., conserved in other plant species (Martin et al., 2010; M.Y. 2013), and B. napus (M.Y. Xu et al., 2012; Zhou et al., 2012a). Xu et al., 2012). In Arabidopsis, miR156 and miR172 were This may be attributed to the differences in temporal or spatial shown to be involved in the regulation of flowering time expression of miRNAs, and their targets may cause insufficient and floral development by negatively regulating SBP-LIKE degradation of the target genes, resulting in the levels of these (SPLs) proteins and AP2-like factors, respectively (Wu et al., sliced targets being too low to be detected in the degradome 2009). In this study, rsa-miR156 and rsa-miR157 targeted (Kawashima et al., 2009; M.Y. Xu et al., 2012). Furthermore, SBP transcription factors, whereas three conserved miRNAs some plant miRNAs regulate their targets by mRNA cleavage, (rsa-miR165, rsa-miR169, and rsa-miR172) targeted AP2- whereas some other miRNAs may silence their target activ- like factors, suggesting that these five conserved miRNAs ity by translational repression (Fabian et al., 2010; X.Y. Yang might play significant roles in regulating flowering time and et al., 2013). Additionally, because the R. sativus full genome floral development in radish. In addition, some other targets sequence is not available yet, the insufficient number of acces- appeared to be involved in signal transduction, metabolism, sible radish reference sequences might limit the identification of disease resistance, and response to environmental stresses. targets for both known and novel miRNAs. Notably, several key responsive proteins or enzymes for In conclusion, this is the first report on the genome-wide heavy metal uptake and translocation were identified as tar- identification of novel and Cd-responsive miRNAs and their get transcripts for a few conserved miRNAs. rsa-miR156 targets using small RNA sequencing and degradome analy- targeted a transcript encoding a glutathione S-transferase 5 sis in radish. A total of 15 known miRNAs and eight novel (GST5), whereas rsa-miR393 targeted phytochelatin synthase miRNA families were identified to be responsive to Cd stress. 1 (PCS1). In plant cells, one major mechanism related to Cd Using degradome sequencing, 18 and 71 targets cleaved by detoxification is complexation with a range of metal-chelat- novel and known miRNAs were confirmed in radish for the ing peptides such as glutathione (GSH) and phytochelatins first time. Some target transcripts were functionally pre- (PCs), both of which reduced its mobility and sequestered the dicted to code biotic and abiotic stress-responsive proteins phytochelatin–metal complexes into the vacuole (Shimo et al., or enzymes. Expression patterns of these differentially regu- 2011; Satoh-Nagasawa et al., 2012). Moreover, detoxification lated miRNAs and their targets were shown to be regulated of reactive oxygen species (ROS) could be accomplished by by Cd stress. These findings could provide new information synthesis of various antioxidants such as ascorbate, GSH, and for further identification and characterization of miRNAs glutathione S-transferase (DalCorso et al., 2008; Takahashi in radish, and advance our understanding of the functional et al., 2011). These findings suggested that rsa-miR156 and characterization of miRNAs and their targets in regulating rsa-miR393 could be involved in Cd detoxification and medi- plant response to Cd stress. ation via directing regulation of the GST5 and PCS1 genes in radish, respectively. Additionally, iron transporter-like pro- Supplementary data tein and ABC transporter protein, which were known to be important transporter proteins for heavy metal uptake and Supplementary data are available at JXB online. translocation (Zhou et al., 2012a; Shimo et al., 2011), were Figure S1. Venn diagrams for analysis of total (A) and shown to be targeted by rsa-miR159 and rsa-miR166, respec- unique (B) small RNAs between CK and Cd200 libraries tively. The results indicated that rsa-miR159 and rsa-miR166 from radish roots. might be important participants in Cd uptake and transloca- Figure S2. The secondary structures of novel Raphanus tion in plants through regulating their corresponding targets. sativus miRNA precursors. Taken together, the identified miRNA-mediated gene expres- Table S1. qRT-PCR-validated miRNAs and their sion of PCS1, GST5, iron transporter-like protein, and ABC sequences. Identification of Cd-responsive microRNAs and their targets in radish | 4285

Table S2. Summary of cleaning data produced from CK truncatula by genome-wide high-throughput sequencing. Planta 235, and Cd200 sRNA libraries of radish. 375–386. Table S3. Detailed information of the known miRNAs Cuperus JT, Fahlgren N, Carrington JC. 2011. Evolution and identified from radish. functional diversification of MIRNA genes. The Plant Cell 23, Table S4. Detailed information of novel candidate miR- 431–442. NAs identified from radish. DalCorso G, Farinati S, Maistri S, Furini A. 2008. How plants Table S5. Summary of Cd-responsive miRNAs in radish. cope with cadmium: staking all on metabolism and gene expression. Table S6. Detailed information of targets for known radish Journal of Integrative Plant Biology 50, 1268–1280. miRNAs confirmed by degradome sequencing. Ding YF, Chen Z, Zhu C. 2011. Microarray-based analysis of cadmium-responsive microRNAs in rice (Oryza sativa). Journal of Acknowledgements Experimental Botany 62, 3563–3573. Ding YF, Zhu C. 2009. The role of microRNAs in copper and This work was in part supported by grants from the cadmium homeostasis. Biochemical and Biophysical Research National Natural Science Foundation of China (31372064, Communications 386, 6–10. 31171956), the Fundamental Research Funds for the Central Universities (KYZ201209), the National Key Technologies Fabian MR, Sonenberg N, Filipowicz W. 2010. Regulation of R & D Program of China (2012BAD02B01), the Program mRNA translation and stability by microRNAs. Annual Review of for NCET, MOE (NCET-10-047), and the PAPD. We thank Biochemistry 79, 351–379. Dr Y. Chen from The University of Western Australia for his German MA, Luo SJ, Schroth G, Meyers BC, Green PJ. 2009. critical review and helpful comments during the preparation Construction of Parallel Analysis of RNA Ends (PARE) libraries for the of this manuscript. study of cleaved miRNA targets and the RNA degradome. Nature Protocols 4, 356–362. German MA, Pillay M, Jeong DH, et al. 2008. Global identification References of microRNA–target RNA pairs by parallel analysis of RNA ends. Addo-Quaye C, Eshoo TW, Bartel DP, Axtell MJ. 2008. Nature Biotechnology 26, 941–946. Endogenous siRNA and miRNA targets identified by sequencing of the Grant CA, Clarke JM, Duguid S, Chaney RL. 2008. Selection Arabidopsis degradome. Current Biology 18, 758–762. and breeding of plant cultivars to minimize cadmium accumulation. Addo-Quaye C, Miller W, Axtell MJ. 2009. CleaveLand, a pipeline Science of the Total Environment 390, 301–310. for using degradome data to find cleaved small RNA targets. Hafner M, Landgraf P, Ludwig J, Rice A, Ojo T, Lin C, Holoch Bioinformatics 25, 130–131. D, Lim C, Tuschl T. 2008. Identification of microRNAs and other Allen E, Xie ZX, Gustafson AM, Carrington JC. 2005. MicroRNA- small regulatory RNAs using cDNA library sequencing. Methods 44, directed phasing during trans-acting siRNA biogenesis in plants. Cell 3–12. 121, 207–221. Hsieh LC, Lin SI, Shih AC, Chen JW, Lin WY, Tseng CY, Li WH, Barakat A, Sriram A, Park J, Zhebentyayeva T, Main D, Abbott Chiou TJ. 2009. Uncovering small RNA-mediated responses to A. 2012. Genome wide identification of chilling responsive microRNAs phosphate deficiency in Arabidopsis by deep sequencing. Plant in Prunus persica. BMC Genomics 13, 481. Physiology 151, 2120–2132. Bartel DP. 2004. MicroRNAs: genomics, biogenesis, mechanism, and Huang SQ, Peng J, Qiu CX, Yang ZM. 2009. Heavy metal- function. Cell 116, 281–297. regulated new microRNAs from rice. Journal of Inorganic Biochemistry Benson DA, Karsch-Mizrachi I, Lipman DJ, Ostell J, Sayers EW. 103, 282–287. 2011. GenBank. Nucleic Acids Research 39, D32–D37. Huang SQ, Xiang AL, Che LL, Chen S, Li Hui, Song JB, Yang Besson-Bard A, Gravot A, Richaud P, Auroy P, Duc C, Gaymard ZM. 2010. A set of miRNAs from Brassica napus in response to F, Taconnat L, Renou JP, Pugin A, Wendehenne D. 2009. Nitric sulfate-deficiency and cadmium stress. Plant Biotechnology Journal 8, oxide contributes to cadmium toxicity in Arabidopsis by promoting 887–899. cadmium accumulation in roots and by up-regulating genes related to Ishikawaa S, Ishimaru Y, Igura M, Kuramata M, Abe T, Senoura iron uptake. Plant Physiology 149, 1302–1315. T, Hase Y, Arao T, Nishizawa NK, Nakanishi H. 2012. Ion-beam Brodersen P, Sakvarelidze-Achard L, Bruun-Rasmussen irradiation, gene identification, and marker-assisted breeding in the M, Dunoyer P, Yamamoto YY, Sieburth L, Voinnet O. 2008. development of low-cadmium rice. Proceedings of the National Widespread translational inhibition by plant miRNAs and siRNAs. Academy of Sciences, USA , 109, 19166–19171. Science 320, 1185–1190. Jeong DH, Park S, Zhai J, Qurazada SGR, De Paoli E, Meyers Chen L, Ren YY, Zhang YY, Xu JC, Sun FS, Zhang ZY, Wang YW. BC, Green PJ. 2011. Massive analysis of rice small RNAs, 2012a. Genome-wide identification and expression analysis of heat- mechanistic implications of regulated microRNAs and variants for responsive and novel microRNAs in Populus tomentosa. Gene 504, differential target RNA cleavage. The Plant Cell 23, 4185–4207. 160–165. Jones-Rhoades MW, Bartel DP, Bartel B. 2006. MicroRNAs and Chen L, Wang TZ, Zhao MG, Tian QY, Zhang WH. 2012b. their regulatory roles in plants. Annual Review of Plant Biology 57, Identification of aluminum-responsive microRNAs in Medicago 19–53. 4286 | Xu et al.

Kawashima CG, Yoshimoto N, Maruyama-Nakashita A, polymerase chain reaction profiling and small RNA sequencing. Plant Tsuchiya YN, Saito K, Takhashi H, Dalmay T. 2009. Sulphur Physiology 150, 1541–1555. starvation induces the expression of microRNA-395 and one of its Rajagopalan R, Vaucheret H, Trejo J, Bartel DP. 2006. A diverse target genes but in different cell types. The Plant Journal 57, and evolutionarily fluid set of microRNAs in Arabidopsis thaliana. 313–321. Genes and Devevelopment 20, 3407–3425. Khraiwesh B, Zhu JK, Zhu JH. 2012. Role of miRNAs and siRNAs Ruiz-Ferrer V, Voinnet O. 2009. Roles of plant small RNAs in biotic in biotic and abiotic stress responses of plants. Biochimica et stress responses. Annual Review of Plant Biology 60, 485–510. Biophysica Acta 1819, 137–148. Satoh-Nagasawa N, Mori M, Nakazawa N, Kawamoto T, Nagato Lelandais-Brière C, Naya L, Sallet E, Calenge F, Frugier F, Y, Sakurai K, Takahashi H, Watanabe A, Akagi H. 2012. Mutations Hartmann C, Gouzy J, Crespia M. 2009. Genome-wide Medicago in rice (Oryza sativa) heavy metal ATPase 2 (OsHMA2) restrict the truncatula small RNA analysis revealed novel microRNAs and isoforms translocation of zinc and cadmium. Plant and Cell Physiology 53, differentially regulated in roots and nodules. The Plant Cell 21, 213–224. 2780–2796. Shamimuzzaman M, Vodkin L. 2012. Identification of soybean seed Li BS, Duan H, Li JG, Deng XW, Yin WL, Xia XL. 2013. Global developmental stage-specific and tissue-specific miRNA targets by identification of miRNAs and targets in Populus euphratica under salt degradome sequencing. BMC Genomics 13, 310. stress. Plant Molecular Biology 81, 525–539. Shimo H, Ishimaru Y, An G, Yamakawa T, Nakanishi H, Li BS, Qin YR, Duan H, Yin WL, Xia XL. 2011. Genome-wide Nishizawa NK. 2011. Low cadmium (LCD), a novel gene related to characterization of new and drought stress responsive microRNAs in cadmium tolerance and accumulation in rice. Journal of Experimental Populus euphratica. Journal of Experimental Botany 62, 3765–3779. Botany 62, 5727–5734. Li R, Yu C, Li Y, Lam TW, Yiu SM, Kristiansen K, Wang J. Sunkar R, Kapoor A, Zhu JK. 2006. Posttranscriptional induction 2009. SOAP2: an improved ultrafast tool for short read alignment. of two Cu/Zn superoxide dismutase genes in Arabidopsis is mediated Bioinformatics 25, 1966–1967. by downregulation of miR398 and important for oxidative stress Li YF, Zheng Y, Addo-Quaye C, Zhang Li, Saini A, tolerance. The Plant Cell 18, 2051–2065. Jagadeeswaran G, Axtell MJ, Zhang W, Sunkar R. 2010. Takahashi R, Ishimaru Y, Senoura T, Shimo H, Ishikawa S, Transcriptome-wide identification of microRNA targets in rice. The Arao T, Nakanishi H, Nishizawa NK. 2011. The OsNRAMP1 Plant Journal 62, 742–759. iron transporter is involved in Cd accumulation in rice. Journal of Liang G, Yang F, Yu D. 2010. MicroRNA395 mediates regulation of Experimental Botany 62, 4843–4850. sulfate accumulation and allocation in Arabidopsis thaliana. The Plant Uraguchi S, Mori S, Kuramata M, Kawasaki A, Arao T, Ishikawa Journal 62, 1046–1057. S. 2009. Root-to-shoot Cd translocation via the xylem is the major Macovei A, Tuteja N. 2012. microRNAs targeting DEAD-box process determining shoot and grain cadmium accumulation in rice. helicases are involved in salinity stress response in rice (Oryza sativa Journal of Experimental Botany 60, 2677–2688. L.) BMC Plant Biology 12, 183. Voinnet O. 2009. Origin, biogenesis, and activity of plant microRNAs. Cell 136, 669–687. Man MZ, Wang X, Wang Y. 2000. POWER_SAGE: comparing statistical tests for SAGE experiments. Bioinformatics 16, 953–959. Wang LZ, He QW. 2005. Chinese radish . Beijing: Scientific and Technical Documents Publishing House. Martin RC, Asalina M, Liu PP, et al. 2010. The microRNA156 and microRNA172 gene regulation cascades at post-germinative stages in Wang TZ, Chen L, Zhao MG, Tian QY, Zhang WH. 2011. Arabidopsis. Seed Science Research 20, 79–87. Identification of drought-responsive microRNAs in Medicago truncatula by genome-wide high-throughput sequencing. BMC Martínez G, Forment J, Llave C, Pallás V, Gómez G. 2011. High- Genomics 12, 367. throughput sequencing, characterization and detection of new and conserved cucumber miRNAs. PLoS One 6, e19523. Wiebe K, Harris NS, Faris JD, Clarke JM, Knox RE, Taylor GJ, Pozniak CJ. 2010. Targeted mapping of Cdu1, a major locus McCormick KP, Willmann MR, Meyers BC. 2011. Experimental regulating grain cadmium concentration in durum wheat (Triticum design, preprocessing, normalization and differential expression turgidum L. var durum). Theoretical and Applied Genetics 121, analysis of small RNA sequencing experiments. Silence 2, 2. 1047–1058. Meyers BC, Axtell MJ, Bartel B, et al. 2008. Criteria for annotation Wu G, Park MY, Conway SR, Wang JW, Weigel D, Poethig of plant microRNAs. The Plant Cell 20, 3186–3190. RS. 2009. The sequential action of miR156 and miR172 regulates Morozova O, Marra MA. 2008. Applications of next-generation developmental timing in Arabidopsis. Cell 138, 750–759. sequencing technologies in functional genomics. Genomics 92, 255–264. Xu L, Wang LJ, Gong YQ, Dai WH, Wang Y, Zhu XW, Wen TC, Muvva C, Tewari L, Aruna K, Ranjit P, Zahoorullah SMD, Liu LW. 2012. Genetic linkage map construction and QTL mapping Matheen KA, Veeramachaneni MDH. 2012. In silico identification of cadmium accumulation in radish (Raphanus sativus L.). Theoretical of miRNAs and their targets from the expressed sequence tags of and Applied Genetics 125, 659–670. Raphanus sativus. Bioinformation 8, 98–103. Xu L, Wang Y, Xu YY, Wang LJ, Zhai LL, Zhu XW, Gong YQ, Pant BD, Musialak-Lange M, Nuc P, May P, Buhtz A, Kehr J, Ye S, Liu LW. 2013. Identification and characterization of novel and Walther D, Scheible WR. 2009. Identification of nutrient-responsive conserved microRNAs in radish (Raphanus sativus L.) using high- Arabidopsis and rapeseed microRNAs by comprehensive real-time throughput sequencing. Plant Science 201–202, 108–114. Identification of Cd-responsive microRNAs and their targets in radish | 4287

Xu MY, Dong Y, Zhang QX, Zhang L, Luo YZ, Sun J, Fan YL, Zhang J, Xu Y, Huan Q, Chong K. 2009. Deep sequencing of Wang L. 2012. Identification of miRNAs and their targets from Brachypodium small RNAs at the global genome level identifies Brassica napus by high-throughput sequencing and degradome microRNAs involved in cold stress response. BMC Genomics 10, 449. analysis. BMC Genomics 13, 421. Zhou L, Liu Y, Liu Z, Kong D, Duan M, Luo L. 2010. Genome-wide Yang JH, Liu XY, Xu BC, Zhao N, Yang XD, Zhang MF. 2013. identification and analysis of drought-responsive microRNAs in Oryza Identification of miRNAs and their targets using high-throughput sativa. Journal of Experimental Botany 61, 4157–4168. sequencing and degradome analysis in cytoplasmic male-sterile and Zhou ZS, Huang SQ, Yang ZM. 2008. Bioinformatic identification its maintainer fertile lines of Brassica juncea. BMC Genomics 14, 9. and expression analysis of new microRNAs from Medicago truncatula. Yang XY, Wang LC, Yuan DJ, Lindsey K, Zhang XL. 2013. Small Biochemical and Biophysical Research Communications 374, RNA and degradome sequencing reveal complex miRNA regulation 538–542. during cotton somatic embryogenesis. Journal of Experimental Botany Zhou ZS, Song JB, Yang ZM. 2012a. Genome-wide identification of 64, 1521–1536. Brassica napus microRNAs and their targets in response to cadmium. Yu X, Wang H, Lu, YZ, Ruiter M, Cariaso M, Prins M, Tunen A, Journal of Experimental Botany 63, 4597–4613. He YK. 2012. Identification of conserved and novel microRNAs that Zhou ZS, Zeng HQ, Liu ZP, Yang ZM. 2012b. Genome-wide are responsive to heat stress in Brassica rapa. Journal of Experimental identification of Medicago truncatula microRNAs and their targets Botany 63, 1025–1038. reveals their differential regulation by heavy metal. Plant, Cell and Zeng QY, Yang CY, Ma QB, Li XP, Dong WW, Nian H. 2012. Environment 35, 86–99. Identification of wild soybean miRNAs and their target genes Zuker M. 2003. Mfold web server for nucleic acid folding and responsive to aluminum stress. BMC Plant Biology 12, 182. hybridization prediction. Nucleic Acids Research 31, 3406–3415.