Small RNA and degradome sequencing reveals important microRNA function in Astragalus chrysochlorus response to selenium stimuli

Summary

Selenium (Se), an essential element, plays important roles in human health as well as environmental sustainability. Se hyperaccumulating plants are thought as an alternative selenium resource, recently. Astragalus species are known as hyperaccumulator of Se by converting it to nonaminoacid compounds. However, Se‐metabolism‐related hyperaccumulation is not elucidated in plants yet. MicroRNAs (miRNAs) are key molecules in many biological and metabolic processes via targeting mRNAs, which may also play an important role in Se accumulation in plants. In this study, we identified 418 known miRNAs, belonging to 380 families, and 151 novel miRNAs induced by Se exposure in Astragalus chyrsochlorus callus. Among known miRNAs, the expression of 287 families was common in both libraries, besides 71 families were expressed only in Se‐treated sample, whereas 60 conserved families were expressed in control tissue. miR1507a, miR1869 and miR2867‐3p were mostly up‐regulated, whereas miR1507‐5p and miR8781b were significantly down‐regulated by Se exposure. Computational analysis shows that the targets of miRNAs are involved in different types of biological mechanisms including 47 types of cellular component, 103 types of molecular function and 144 types of biological process. Degradome analysis shows that 1256 mRNAs were targeted by 499 miRNAs. We conclude that some known and novel miRNAs such as miR167a, miR319, miR1507a, miR4346, miR7767‐3p, miR7800, miR9748 and miR‐n93 target transcription factors, disease resistance proteins and some specific genes like cysteine synthase and might be related to plant hormone signal transduction, plant–pathogen interaction and sulphur metabolism pathways.

Introduction

MicroRNAs (miRNAs) are a class of small noncoding RNA molecules with the length of 21–24 nucleotides (Bartel, 2004). They regulate gene expression post‐transcriptionally via targeting mRNAs for degradation and/or translational inhibition in all eukaryotic organisms (Saini et al., 2012). miRNAs play an essential role in biological and metabolic processes in plants such as growth, development, maturation, cell differentiation and response to various abiotic and biotic stresses (Barrera‐Figueroa et al., 2012; Dalmay, 2006; Jones‐Rhoades et al., 2006; Zhang, 2015; Zhang and Wang, 2015). Thus, determination of functional aspects of miRNAs and their targets are important for breeding strategies and plant biotechnology.

Selenium is one of the nonmetallic elements, and it is a component of selenocysteine (Birringer et al., 2002; Whanger, 2002). Se shares the same assimilation pathway with sulphur (S) as they have a similar chemical structure, so Se can be assimilated in plants as well (Sors et al., 2005). Humans and animals need this element as a micronutrient in low concentrations, but it could be very toxic in higher concentrations (Hung et al., 2012). In some cases, plants that can accumulate high level of Se are desired for bioremediation and biofortification studies as well as human health, and recent findings revealed that Se transportation could be closely associated with phytoremediation and biofortification (Hung et al., 2012; Schiavon et al., 2015). The genus Astragalus L. belongs to Leguminosae, the largest flowering family and known as accumulator of high level of Se (Freeman et al., 2006; Shrift and Virupaksha, 1965). Astragalus bisulcatus is the best known plant of Astragalus species that accumulates selenium in high concentrations with the accumulation level of 0.65% in its tissues (Pickering et al., 2003). Astragalus plants contain many active secondary compounds such as saponins, phenolics and polysaccharides. Species of Astragalus are known for their immunostimulant, hepatoprotective, antiperspirant, diuretic, antiviral and tonic effects (Benchadi et al., 2013). Although Astragalus species are hyperaccumulator of Se, their adverse characteristics such as slow growth, low biomass and nonedibility restrict the use of these plants directly for human dietary and biological applications such as bioremediation (Hung et al., 2012). Therefore, biotechnology approaches are needed to improve undesirable properties of high level Se accumulated plants. miRNAs can be the key molecules for promoting Se accumulation in plants.

The effects of metal stress on miRNAs have been studied with boron in barley (Ozhuner et al., 2013), aluminium in Medicago truncatula (Chen et al., 2012) and cadmium in rice (Ding et al., 2011). In these studies, it was found that plants respond to metal stress by altering miRNA expressions. All these stressors are affecting plants growth and development. It is known that they can be toxic when their concentration is above a certain limit. The aim of our study was to identify Se‐related miRNAs and their putative targets in Astragalus chrysochlorus Boiss. & Kotschy (2n = 16). This plant is shown to accumulate selenium as a secondary accumulator (Arı et al., 2010), and although several studies have been reported on identification and characterization of Se‐related genes in Astragalus species (Arı et al., 2010; Çakır and Arı, 2013; Neuhierl and Bock, 1996; Sors et al., 2009), there is no study about Se‐induced miRNA discovery and expression of miRNAs and their targets in Astragalus. For this purpose, we carried out high‐throughput sequencing analysis of small RNAs in A. chrysochlorus. We also performed degradome sequencing for miRNA target identification.

Results

Small RNA library sequencing

A total of 23 646 078 and 20 850 840 small RNA reads (Table S1) were obtained for Se‐treated and untreated samples, respectively. After removed the unnecessary sequences (adaptor, RNAs shorter than 18 nt, polyA), a total of 22 646 781 and 20 044 787 sequences were remained in Se‐treated and control libraries, respectively. These total reads contained miRNA, rRNA, snRNA, tRNA, snoRNA, tRNA and unannotated sequences (Table S2). The small RNA sequences were ranged from 16 to 27 nt in length with the majority were 20–24 nt in length (Figure S1). In both Se‐treated and control libraries, 21–22 nt long small RNAs were the most abundant. While the 21 nt long small RNAs were 26.68% and 28.39%, the 22 nt long small RNAs were 27.28% and 27.14% in libraries constituted from Se‐treated and control callus tissues, respectively. In total small RNAs, 9.38% (unique, 44.48%) were specifically found in Se‐treated sample, whereas 7.98% (unique, 38.83%) were specific in control sample; there are 82.64% (unique, 16.68%) small RNAs commonly existed in both samples (Figure 1).

Figure 1.

Common and unique sequences between Se‐treated and control libraries (a) unique sRNAs (b) total sRNAs

Identification and expression patterns of known miRNAs

To identify known miRNAs in A. chrysochlorus, clean reads generated from two libraries aligned against miRNA database (Release 20) (Kozomara and Griffiths‐Jones, 2011). A total of 418 miRNAs, belonging to 380 families, were detected in both Se‐treated and control samples. Of these miRNAs, 71 were expressed only in Se‐treated samples; 60 were expressed only in control sample. A total of 287 miRNA families were expressed in both treated and untreated samples (Figure 2). For example, miR1869 and miR6195 were only detected in Se‐treated samples, whereas miRNAs, such as miR156, miR157 and miR159, were detected in both libraries. Among the 380 miRNA families, 160 miRNAs were differentially expressed in both libraries after normalization of miRNA reads to ‘reads per million’ (RPM) (Table 1, Figures 3 and 4). miR2867‐3p was the miRNA with most fold change with 17.8‐fold up‐regulated by Se exposure, followed by miR1869 and miR1507a. Their fold changes were 17.25 and 16.65, respectively. miR319b, miR535a, miR846‐5p, miR3633a‐3p, miR3711, miR3946, miR4414b, miR5232, miR5241a, miR5369, miR9662a‐3p and miR9741 were also found to be significantly up‐regulated in Se‐treated tissues. On the other hand, miR165a‐5p miR397a, miR399i, miR419, miR848‐3p, miR1507‐5p, miR2920, miR5077, miR5225‐5p, miR5239, miR5721, miR6266a, miR7503, miR7539 and miR8675c were significantly down‐regulated by Se treatment (Figure 4). The most significantly down‐regulated one was miR1507‐5p.

Figure 2.

Distribution of miRNAs between Control and Selenium Treatment. (a) Conserved miRNAs; (b) Novel miRNAs.

Table 1.

Differentially expressed known miRNAs after Se exposure in Astragalus chrysochlorus

miRNA	Normalized expression levela		Fold change (log2 Se treatment/control)
	Se treatment	Control
miR1044‐3p	12.49	0	10.28
miR1081	1.45	0	7.18
miR1085‐3p	30.11	0	11.55
miR1112‐3p	3.79	0	8.56
miR1114	0	2.64	−8.04
miR1120b‐3p	0	1.94	−7.60
miR1147.2	5.03	0	8.97
miR1440b	0.01	33.97	−11.73
miR1507‐5p	0.01	1884.97	−17.52
miR1507a	1030.47	0	16.65
miR1531‐3p	0	15.56	−10.60
miR158a‐3p	1.28	0	7.00
miR161‐5p.1	13.73	0	10.42
miR165a‐5p	0	24.59	−11.26
miR166c	0	3.94	−8.62
miR166g‐5p	4.28	0	8.74
miR167a	3721.19	8271.42	−1.15
miR167f‐3p	32.14	105.71	−1.71
miR169n‐3p	0	18.00	−10.81
miR171b‐3p	33.07	11.62	1.50
miR171m	5.60	0	9.13
miR171n	0	4.14	−8.69
miR1861c	0	26.24	−11.35
miR1869	1566.09	0	17.25
miR1873	0	4.53	−8.82
miR2084	0	12.62	−10.30
miR2097‐5p	3.62	0	8.50
miR2108b	1.28	0	7.00
miR2628	1.85	0	7.53
miR2670e	91.13	0	13.15
miR2867‐3p	2285.31	0	17.80
miR2920	0	8.43	−9.71
miR2937	0	3.59	−8.48
miR319a‐3p	4.54	0	8.82
miR319b	32.27	3.14	3.36
miR3437‐3p	0	4.68	−8.87
miR3446‐5p	14.88	0	10.53
miR3476	1.89	0	7.56
miR3633a‐3p	8.30	1.74	2.24
miR3633a‐5p	374.84	995.72	−1.40
miR3637‐5p	2.11	0	7.72
miR3711	1.05	0	6.72
miR393a	12.05	4.88	1.30
miR3946	23.44	9.87	1.24
miR395	0	1.64	−7.36
miR395n	41.06	0	12.00
miR397a	246.92	811.38	−1.71
miR399b‐3p	0	3.84	−8.58
miR399i	3.62	23.79	−2.71
miR415	6.97	1008.59	−7.17
miR419	3.92	9.42	−1.26
miR4240	51.13	0	12.32
miR4244	75.86	31.57	1.26
miR4346	67.29	30.03	1.16
miR4348a	8.87	0	9.79
miR4386	26.97	0	11.39
miR4388	0	6.23	−9.28
miR4414a‐5p	0	33.32	−11.70
miR4414b	278.80	0	14.76
miR4415a‐3p	0.52	1.14	−1.11
miR4415b‐5p	64.90	0	12.66
miR447a.2‐3p	15.63	6.88	1.18
miR477a	8.52	0	9.73
miR477d	0	5.48	−9.10
miR5029	0	19.70	−10.94
miR5049‐3p	34.53	0	11.75
miR5070‐3p	4.32	0	8.75
miR5077	31.26	63.30	−1.01
miR5083	0	1.19	−6.90
miR5176‐3p	10.64	0	10.05
miR5208a	0	8.93	−9.80
miR5224b	22.56	0	11.13
miR5225‐5p	0	44.65	−12.12
miR5232	19.38	6.48	1.57
miR5239	59.16	136.94	−1.21
miR5241a	70.34	13.32	2.40
miR5258	0	11.67	−10.18
miR5264	0	2.09	−7.71
miR5265	1.72	0	7.42
miR5270a	1.54	0.69	1.14
miR5273	6.18	0	9.27
miR5287a	0	13.76	−10.42
miR5287b	5.47	0	9.09
miR530‐3p	0	12.02	−10.23
miR5337a	2.69	0	8.07
miR535a	188.85	0	14.20
miR535d	0	1.64	−7.36
miR5368	42.65	8.78	2.28
miR5369	68.66	0	12.74
miR5485	10.59	0	10.04
miR5503	22.16	9.82	1.17
miR5519	584.01	272.53	1.09
miR5528	4.81	0	8.91
miR5536	81.33	40.35	1.01
miR5575	0	4.83	−8.91
miR5634	0	1.496	−7.22
miR5664	0	11.17	−10.12
miR5671a	425.35	205.48	1.04
miR5721	5.12	147.22	−4.84
miR5763a	0	23.39	−11.19
miR5776	45.17	0	12.14
miR5792	20.48	0	11.00
miR5837.2	93.30	0	13.18
miR6021	33.60	11.72	1.51
miR6029	0	6.53	−9.35
miR6116‐5p	0	33.47	−11.70
miR6151f	7.94	0	9.63
miR6172	4.54	0	8.82
miR6195	193.75	0	14.24
miR6196	18.85	45.24	−1.26
miR6209	137.06	53.87	1.34
miR6266a	0	9.778	−9.93
miR6267c‐3p	1.67	0	7.39
miR6290	25.47	0	11.31
miR6449	1.41	0	7.14
miR6462c‐5p	8.25	0	9.68
miR6470	0	1.945	−7.60
miR7503	0	108.90	−13.41
miR7533a	101.42	46.29	1.13
miR7539	3.57	79.27	−4.47
miR7700‐5p	41.28	0	12.01
miR7752‐3p	0	5.487	−9.10
miR7767‐3p	11.43	0	10.15
miR7779‐5p	2.51	0	7.97
miR7797a	0	29.23	−11.51
miR7800	1.45	0	7.18
miR7819	7.68	3.392	1.17
miR7825	0	10.37	−10.01
miR812g	1.54	0	7.27
miR8141	296.68	109.95	1.43
miR8149	60.18	0	12.55
miR8155	1.05	2.24	−1.08
miR8182	1.72	0	7.42
miR845a	7.33	0	9.51
miR846‐5p	4.28	0	8.74
miR848‐3p	0	1.496	−7.22
miR857	1.32	0	7.04
miR8633	84.47	0	13.04
miR8638	0	10.17	−9.99
miR8658	0	5.23	−9.03
miR866‐5p	78.37	157.59	−1.00
miR8675c	0	1.097	−6.77
miR8691	1.89	0	7.56
miR869.2	2.82	0	8.14
miR8704	0	7.98	−9.64
miR8717	0	7.08	−9.46
miR8743a	0	55.87	−12.44
miR8757a	3.97	1.7960	1.14
miR8781b	0	120.28	−13.55
miR908.2	6.27	1.74	1.84
miR9471a‐5p	3.92	0	8.61
miR9478‐3p	2.78	7.38	−1.40
miR9497	0	8.73	−9.76
miR9568‐3p	2.69	1.19	1.16
miR9657b‐5p	8.96	0	9.80
miR9662a‐3p	1.81	0	7.50
miR9741	37.93	0	11.88
miR9742	0	84.26	−13.04
miR9748	112.81	55.67	1.01
miR9766	0	6.78	−9.40

^aAll miRNA expressions were normalized to read per million (RPM). If miRNA expression measured as zero, normalized expression valued as 0.01 according to Murakami et al. (2006). Fold change was calculated using the formula, fold change = log₂(treatment/control) (Marsit et al., 2006). Significance was calculated as fold change log₂ > 1 or log₂ < −1 and P‐value < 0.01. The miRNAs given in the table above are significantly expressed.

Figure 3.

Small RNA expression profiles of control and Se‐treated callus of Astragalus chrysochlorus.

Figure 4.

Expression profiles of randomly selected miRNAs with different abundance in Se‐treated Astragalus chrysochlorus calli.

Identification of potentially novel miRNAs and their expression pattern

Among a total of 151 novel miRNAs, 55 and 57 miRNAs were identified in selenium treated and control samples, respectively, whereas 39 miRNAs found in both libraries (Figure 2). Of them, 30 miRNAs were differentially expressed after Se treatment (Table 2). Among these 30 miRNAs, 14 were only expressed in Se‐treated sample, whereas 12 were only expressed in control. Only 4 of them were expressed in both treated and untreated samples. Among these novel miRNAs, 16 were up‐regulated and 14 were down‐regulated (Table 2). The most up‐regulated miRNAs are miR‐n1 (11.05), miR‐n53 (11.06), miR‐n61 (10.86), miR‐n74 (10.44), miR‐n87 (10.46), miR‐n90 (17.65) and miR‐n93 (12.48). Of them, miR‐n90 was up‐regulated most with 17.65‐fold change. The most significantly down‐regulated one was miR‐n146 with 9.49‐fold change and miR‐n122 was also down‐regulated by 9.24‐fold.

Table 2.

Differentially expressed novel miRNAs after Se exposure in Astragalus chrysochlorus

miRNA	Normalized expression levela		Fold change (log2 Se treatment/control)
	Se treatment	Control
miR‐n1	21.23	0	11.05
miR‐n101	0	3.14	−8.29
miR‐n109	0	1.09	−6.77
miR‐n113	0	1.69	−7.40
miR‐n117	0	3.29	−8.36
miR‐n122	0	6.08	−9.24
miR‐n135	0	1.19	−6.90
miR‐n142	0	1.696	−7.40
miR‐n144	0	2.99	−8.22
miR‐n146	0	7.23	−9.49
miR‐n147	0	1.59	−7.31
miR‐n148	0	5.88	−9.20
miR‐n15	2.11	0	7.72
miR‐n21	5.87	0	9.19
miR‐n26	2.47	6.43	−1.37
miR‐n28	1.45	0	7.18
miR‐n35	2.82	9.32	−1.72
miR‐n42	1.19	0	6.89
miR‐n5	2.34	0	7.87
miR‐n53	21.37	0	11.06
miR‐n61	18.63	0	10.86
miR‐n70	14.26	5.28	1.43
miR‐n71	562.11	214.96	1.38
miR‐n74	13.90	0	10.44
miR‐n79	1.32	0	7.04
miR‐n8	1.63	0	7.35
miR‐n87	14.08	0	10.46
miR‐n90	2062.14	0	17.65
miR‐n93	57.31	0	12.48
miR‐n95	0	1.49	−7.22

^aAll miRNA expressions were normalized to read per million (RPM). If miRNA expression measured as zero, normalized expression valued as 0.01 according to Murakami et al. (2006). Fold change was calculated using the formula, fold change = log₂(treatment/control) (Marsit et al., 2006). Significance was calculated as fold change log₂ > 1 or log₂ < −1 and P‐value < 0.01. The miRNAs given in the table above are significantly expressed.

Degradome sequencing analysis

To validate the cleavage sites of miRNAs, we performed high‐throughput degradome sequencing. A total of 29 371 471 reads were obtained by degradome sequencing. After removing the adaptors and other RNAs, a total of 4 955 825 unique reads were obtained. Figure 5 shows the target plots of identified targets of randomly selected genes. In total, 1339 predicted sites were identified. The predicted sites were determined to be cleaved by 499 miRNAs. The total predicted sites were in 1256 genes with 2027 cleavage events. The target genes were annotated and classified as transcription factors and their subunits (WRKY, trihelix transcription factor, bHLH143, RF2b, MYC2, GTE4, TCP8, Myb family transcription factor APL, ethylene‐responsive transcription factors, heat stress transcription factors), enzyme coding genes such as kinases and transferases (probable leucine‐rich repeat receptor‐like protein kinase, probable LRR receptor‐like serine/threonine protein kinase, calcium‐dependent protein kinase 3, uracil phosphoribosyltransferase, mRNA cap guanine‐N7 methyltransferase 1, chloroplastic homogentisate phytyltransferase 2, mitochondrial aminomethyltransferase), resistance proteins (TMV resistance protein, pleiotropic drug resistance protein), leucine‐rich repeat, leucine zipper and zinc finger proteins and other structural and functional proteins.

Figure 5.

Target plots (t‐plots) of miRNAs and their targets. The red arrows indicate the most abundant peaks and cleavage sites. (a) miR162‐3p targeting endonuclease Dicer homologue‐1‐like protein, (b) miR1513a targeting blue light‐activated histidine kinase, (c) miR2118b targeting hypoxanthine‐guanine phosphoribosyltransferase‐like protein, (d) miR172c targeting putative ethylene‐responsive transcription factor RAP‐2‐7‐like protein, (e) miR159b‐3p targeting hypothetical protein 11M19.5, (f) miR166 h‐3p targeting homeobox leucin zipper protein ATHB‐15‐like protein.

Target identification and GO and KEGG pathway analyses

The targets of identified miRNAs were subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) to perceive their roles biologically. It was determined that the target genes are involved in 47 types of cellular component, 103 types of molecular function and 144 types of biological process. The detailed summary of GO classification was given in Figure 6. In Se‐treated tissues, there were 3017 targets genes classified into different groups, but the biological regulations (206), cellular processes (637), metabolic processes (588), regulation of biological processes (201), single‐organism processes (351) and response to stimulus (245) were the most abundant ones in biological process categories. About cellular component category, there were 2183 target genes involved in different groups. The most abundants were cell (538), cell part (538), macromolecular complexes (153), membrane (186), organelle (422) and organelle parts (162). The last category was molecular function and 1397 targets were determined. The most abundant groups were binding (689) and catalytic activity (572), respectively. The most abundant categories obtained by known and novel miRNA analysis were summarized in Tables 3 and 4, respectively. According to KEGG analysis, 968 target genes were annotated to 239 pathways. Consistent with GO analysis, we determined the putative targets of down‐regulated and up‐regulated miRNAs (Tables 5 and 6). The KEGG pathway related to plant–pathogen interaction pathway and the related miRNAs were shown in Figure 7. According to these analysis, calcium‐dependent protein kinase 1 (CPDK) which is related to this pathway was affected by miR5049‐3p expression. Cyclic nucleotide gated channel 10 (CGNGs) is an ion channel and affected by miR5485 expression. Calcium‐binding calmodulin‐like protein 7 was affected by miR4244 expression. Disease resistance protein (RPM1) was affected by miR1507a and miR1507c‐5p. Disease resistance protein (RPS2) was affected by miR1510a, miR5652 and miR3633a‐3p. Disease resistance protein (RSP5) was affected by miR5255, miR1510a, miR2118, miR2118a‐3p and miR3633a‐3p. Chloroplast heat‐shock protein 90 (HSP90), which has a role in protein processing in endoplasmic reticulum, was affected by miR9722 and miR9748. Transcription factor MYC2 is involved in environmental information processing and plant hormone signal transduction and affected by miR9748. All these proteins are involved in environmental adaptation. Also leucin rich repeats receptor‐like serine/threonine protein kinase (FLS2) and somatic embryogenesis receptor kinase 4 were also nonspecific serine/threonine protein kinase and affected by miR414 and miR5205b, respectively. WRKY transcription factor 25 is affected by miR5766 and miR831‐5p. Serine/threonine protein kinase (PBS1) is affected by miR6196, miR6180, miR5658 and miR419. Jasmonate ZIM domain‐containing protein (JAZ5) is involved in plant hormone signal transduction and affected by miR7127a and miR5248. Chitin elicitor receptor kinase 1 (CERK1) is a kind of serine/threonine protein kinases, affected by miR1850.1.

Figure 6.

Summary of GO classifications of miRNA targets in Astragalus chrysochlorus.

Table 3.

The most abundant GO categories in Se‐treated tissues obtained by known miRNA analysis

Go term (Se treatment)	Category	Enrichment factor	P‐value
GO:0046499 S‐adenosylmethioninamine metabolic process	Process	30.65813	0.00119
GO:0008215 spermine metabolic process	Process	30.65813	0.00119
GO:0006597 spermine biosynthetic process	Process	30.65813	0.00119
GO:0006557 S‐adenosylmethioninamine biosynthetic process	Process	30.65813	0.00119
GO:0008295 spermidine biosynthetic process	Process	20.43875	0.01696
GO:0008216 spermidine metabolic process	Process	20.43875	0.01696
GO:2001251 negative regulation of chromosome organization	Process	20.43875	0.01696
GO:0042149 cellular response to glucose starvation	Process	14.52227	2.91e‐06
GO:0000819 sister chromatid segregation	Process	10.90067	4.00e‐10
GO:0009567 double fertilization forming a zygote and endosperm	Process	9.81060	0.00086
GO:0006476 protein deacetylation	Process	9.58067	4.59e‐05
GO:0035601 protein deacylation	Process	9.58067	4.59e‐05
GO:0004014 adenosylmethionine decarboxylase activity	Function	29.16322	0.00044
GO:0010385 double‐stranded methylated DNA binding	Function	29.16322	6.04e‐10
GO:0042301 phosphate ion binding	Function	29.16322	0.00044
GO:0070403 NAD+ binding	Function	21.87242	1.28e‐12
GO:0043130 ubiquitin binding	Function	19.44215	0.00635
GO:0032266 phosphatidylinositol‐3‐phosphate binding	Function	19.44215	0.00635
GO:0032182 small conjugating protein binding	Function	19.44215	0.00635
GO:0030295 protein kinase activator activity	Function	13.12345	2.45e‐06
GO:0019209 kinase activator activity	Function	11.41170	1.08e‐05
GO:0004564 beta‐fructofuranosidase activity	Function	10.74434	0.00062
GO:0004575 sucrose alpha‐glucosidase activity	Function	10.74434	0.00062
GO:0051765 inositol tetrakisphosphate kinase activity	Function	9.20944	0.00964
GO:0047325 inositol tetrakisphosphate 1‐kinase activity	Function	9.20944	0.00964
GO:0052726 inositol‐1,3,4‐trisphosphate 5‐kinase activity	Function	9.20944	0.00964
GO:0052725 inositol‐1,3,4‐trisphosphate 6‐kinase activity	Function	9.20944	0.00964
GO:0005677 chromatin silencing complex	Cellular Component	23.44956	5.63e‐11
GO:0000808 origin recognition complex	Cellular Component	17.14749	9.78e‐35
GO:0000795 synaptonemal complex	Cellular Component	9.94057	1.47e‐09
GO:0000794 condensed nuclear chromosome	Cellular Component	8.79359	1.07e‐08

Table 4.

The most abundant GO categories in Se‐treated tissues obtained by novel miRNA analysis

Go term (Se Treatment)	Category	Enrichment Factor	P‐value
GO:0043174 nucleoside salvage	Process	27.33333	4.46e‐17
GO:0006166 purine ribonucleoside salvage	Process	27.33333	4.46e‐17
GO:0043101 purine‐containing compound salvage	Process	23.76812	1.30e‐15
GO:0042547 cell wall modification involved in multidimensional cell growth	Process	20.82540	0.01871
GO:0009831 plant‐type cell wall modification involved in multidimensional cell growth	Process	20.82540	0.01871
GO:0018298 protein‐chromophore linkage	Process	15.84541	1.28e‐16
GO:0006465 signal peptide processing	Process	11.09179	0.00192
GO:0052657 guanine phosphoribosyltransferase activity	Function	34.53737	2.53e‐21
GO:0004422 hypoxanthine phosphoribosyltransferase activity	Function	34.53737	2.53e‐21
GO:0008442 3‐hydroxyisobutyrate dehydrogenase activity	Function	13.81495	0.04382
GO:0016168 chlorophyll binding	Function	13.12420	2.14e‐14
GO:0051539 4 iron, 4 sulphur cluster binding	Function	9.18082	2.00e‐12
GO:0005852 eukaryotic translation initiation factor 3 complex	Cellular Component	15.08777	0.00016
GO:0009522 photosystem I	Cellular Component	9.22030	1.29e‐13

Table 5.

Targets of miRNAs up‐regulated by Se exposure

miRNA	Putative target(s)
miR1085‐3p	Cyclin A‐like protein [Medicago truncatula] Rhomboid protease gluP [M. truncatula]
miR1147.2	Putative retrotransposon protein, identical [Solanum demissum]
miR1507a	NB‐LRR type disease resistance protein Rps1‐k‐2 [M. truncatula] NBS‐containing resistance‐like protein [M. truncatula] NBS resistance protein [M. truncatula] Disease resistance protein RGA2 [M. truncatula]
miR2108b	Methyltransferase‐like protein [M. truncatula] Ribonuclease H [M. truncatula]
miR2867‐3p	hypothetical protein MTR_5g051130 [M. truncatula]
miR319a‐3p	Transcription factor PCF5 [M. truncatula]
miR3443‐5p	hypothetical protein MTR_056s0017 [M. truncatula]
miR3633a‐3p	PREDICTED: TMV resistance protein N‐like [Glycine max] Disease resistance protein RPS2 [M. truncatula] NBS‐containing resistance‐like protein [M. truncatula] TIR‐NBS‐LRR type disease resistance protein [M. truncatula] Kinase‐like protein [M. truncatula] Disease resistance protein [M. truncatula] TMV resistance protein N [M. truncatula] Disease resistance protein [M. truncatula] TIR‐NBS‐LRR type disease resistance protein [M. truncatula] Disease resistance protein [M. truncatula]
miR395n	putative polyprotein [Cicer arietinum] Polynucleotidyl transferase, Ribonuclease H fold [M. truncatula]
miR4244	retrotransposon gag protein [Arachis hypogaea] PREDICTED: pentatricopeptide repeat‐containing protein At2g01860‐like [Glycine max] hypothetical protein VITISV_037041 [Vitis vinifera]
miR4346	Auxin influx protein [M. truncatula]
miR4348a	Pentatricopeptide repeat‐containing protein [M. truncatula]
miR447a.2‐3p	PREDICTED: WD and tetratricopeptide repeats protein 1‐like [G. max]
miR5049‐3p	calcium‐dependent protein kinase [Swainsona canescens]
miR5070‐3p	Mitochondrial protein, putative [M. truncatula] ATPase subunit 1 (mitochondrion) [Lotus japonicus]
miR5077	Pentatricopeptide repeat‐containing protein [M. truncatula]
miR5167a‐5p	S‐adenosylmethionine decarboxylase [Medicago falcata]
miR5224b	Pol polyprotein [M. truncatula]
miR5287b	PREDICTED: LOW‐QUALITY PROTEIN: cytokinin hydroxylase‐like [G. max] isoflavonoid glucosyltransferase [Glycyrrhiza echinata] PREDICTED: macrophage migration inhibitory factor homologue isoform 2 [Fragaria vesca subsp. vesca]
miR5337a	PREDICTED: DNA topoisomerase 2‐like [Solanum lycopersicum]
miR535a	PREDICTED: transcription factor GTE10‐like [G. max] PREDICTED: Niemann‐Pick C1 protein‐like [G. max] ATP‐dependent DNA helicase Q1 [M. truncatula]
miR5368	Cell wall‐associated hydrolase, partial [M. truncatula]
miR5369	hypothetical protein MTR_8g103420 [M. truncatula] PREDICTED: phosphatidylserine synthase 2‐like [G. max] PREDICTED: tubby‐like F‐box protein 7‐like [G. max]
miR5485	hypothetical protein 11M19.5 [Arabidopsis halleri] PREDICTED: nuclear‐pore anchor‐like [G. max] PREDICTED: GPN‐loop GTPase 2‐like [G. max] hypothetical protein 11M19.5 [Arabidopsis halleri]
miR5528	PREDICTED: ALBINO3‐like protein 2, chloroplastic‐like [G. max]
miR5671a	hypothetical protein MTR_5g051130 [M. truncatula]
miR5837.2	Retrotransposon gag protein, putative [M. truncatula]
miR6021	retrotransposon gag protein [Arachis hypogaea]
miR6151f	DnaJ protein‐like protein [M. truncatula]
miR6195	Alanyl‐tRNA synthetase [M. truncatula]
miR6462c‐5p	Solute carrier family 25 member [M. truncatula] PREDICTED: omega‐hydroxypalmitate O‐feruloyl transferase‐like [G. max]
miR7533a	Vacuolar protein sorting protein [M. truncatula] Cell wall‐associated hydrolase, partial [M. truncatula] SAM domain family protein [M. truncatula] PREDICTED: putative pentatricopeptide repeat‐containing protein At5g59900‐like [G. max]
miR7767‐3p	Disease resistance protein [M. truncatula] TIR‐NBS‐LRR RCT1‐like resistance protein [Medicago sativa] TIR‐NBS‐LRR RCT1‐like resistance protein [M. truncatula] PREDICTED: TMV resistance protein N‐like [G. max] TIR‐NBS‐LRR RCT1‐like resistance protein [M. truncatula] TIR‐NBS‐LRR RCT1‐like resistance protein [M. sativa]
miR7779‐5p	Disease resistance protein RGA2 [M. truncatula]
miR7800	Cysteine synthase [M. truncatula]
miR8141	Ycf68 [M. truncatula]
miR8182	PREDICTED: flavonol synthase/flavanone 3‐hydroxylase‐like [G. max]
miR857	PREDICTED: tRNA (adenine‐N(1)‐)‐methyltransferase noncatalytic subunit trm6‐like [G. max] Chitin‐inducible gibberellin‐responsive protein [M. truncatula] PREDICTED: probable histone‐arginine methyltransferase 1.4‐like [G. max] PREDICTED: putative kinase‐like protein TMKL1‐like [G. max] Rhomboid protease gluP [M. truncatula] Acetylglutamate kinase‐like protein [M. truncatula]
miR8633	Chloroplast inner envelope protein (IEP110) [M. truncatula]
miR9568‐3p	PREDICTED: photosystem II CP47 chlorophyll apoprotein‐like [Solanum lycopersicum]
miR9748	E3 ubiquitin‐protein ligase HUWE1 [M. truncatula] Serine/threonine protein phosphatase 4 regulatory subunit [M. truncatula] PREDICTED: putative receptor‐like protein kinase At1g80870‐like [G. max] Lin‐9‐like protein [M. truncatula] Mitochondrial chaperone BCS1 [M. truncatula] PREDICTED: transcription initiation factor IIA large subunit‐like [G. max] Zinc finger protein [M. truncatula] Zinc finger CCCH domain‐containing protein [M. truncatula] PREDICTED: 97 kDa heat‐shock protein‐like [G. max] PREDICTED: putative receptor‐like protein kinase At1g80870‐like [G. max] Nucleosome assembly protein 1‐like protein [M. truncatula] PREDICTED: BTB/POZ domain‐containing protein At1g03010‐like [G. max] PREDICTED: FACT complex subunit SPT16‐like [G. max] PREDICTED: protein CWC15 homologue [G. max] PREDICTED: heparan‐alpha‐glucosaminide N‐acetyltransferase‐like [G. max] BCCIP‐like protein [M. truncatula] PREDICTED: coiled‐coil domain‐containing protein 75‐like [G. max] PREDICTED: ATP‐dependent helicase BRM‐like [G. max] Knotted‐1 homeobox protein [M. truncatula] PREDICTED: pre‐mRNA‐processing factor 6‐like [G. max] PREDICTED: mRNA‐capping enzyme‐like [G. max] PREDICTED: anaphase‐promoting complex subunit 4 [G. max] Ascorbate peroxidase [M. truncatula] Ycf68 [M. truncatula] Cytochrome c oxidase subunit 5B [M. truncatula] TdcA1‐ORF2 protein [M. truncatula] PREDICTED: protein DA1‐related 2‐like [G. max] 4‐hydroxy‐3‐methylbut‐2‐en‐1‐yl diphosphate synthase [M. truncatula] BZIP transcription factor ATB2 [M. truncatula] putative basic helix‐loop‐helix protein BHLH2 [L. japonicus] Zinc finger‐like protein [M. truncatula] PREDICTED: zinc finger CCCH domain‐containing protein 24‐like [G. max] PREDICTED: transformation/transcription domain‐associated protein‐like [G. max] GT‐2 factor [M. truncatula] alpha 1,4‐fucosyltransferase [M. truncatula] PREDICTED: probable exocyst complex component 6‐like [G. max] PREDICTED: probable WRKY transcription factor 40‐like isoform 2 [G. max] Manganese‐dependent ADP‐ribose/CDP‐alcohol diphosphatase [M. truncatula] cell division control protein 2 homologue 2 [Saccharum hybrid cultivar R570]

Table 6.

Target of miRNAs down‐regulated by Se exposure

miRNA	Putative target(s)
miR1531‐3p	DAG protein, chloroplast precursor, putative [Ricinus communis]
miR165a‐5p	Pentatricopeptide repeat‐containing protein [Medicago truncatula]
miR167a	PREDICTED: auxin response factor 8‐like [G. max]
miR1873	PREDICTED: protein GPR107‐like [G. max]
miR3437‐3p	gag polyprotein [Cicer arietinum]
miR395	Disease resistance protein [M. truncatula] NBS‐LRR type disease resistance protein [C. arietinum] putative NBS‐LRR type disease resistance protein [Pisum sativum] RGA‐D protein [C. arietinum]
miR399i	PREDICTED: probable ubiquitin‐conjugating enzyme E2 24‐like [G. max] phosphate transporter 5 [G. max]
miR415	PREDICTED: LOW‐QUALITY PROTEIN: pre‐mRNA‐splicing factor cwc22‐like [Glycine max] PREDICTED: pre‐mRNA‐splicing factor CWC22 homologue [G. max]
miR419	PREDICTED: cysteine‐rich receptor‐like protein kinase 10‐like [G. max] PREDICTED: condensin‐2 complex subunit H2‐like [G. max] dehydration responsive element binding protein [Halimodendron halodendron]
miR477d	Nuclear transcription factor Y subunit A‐7, partial [M. truncatula] Replication protein A 70 kDa DNA‐binding subunit [M. truncatula] hypothetical protein MTR_8g086620 [M. truncatula]
miR5029	PREDICTED: histidine decarboxylase‐like [G. max]
miR5287a	PREDICTED: pre‐mRNA‐processing factor 40 homologue B‐like [G. max] Mitochondrial protein, putative [Medicago truncatula]
miR530‐3p	PREDICTED: fimbrin‐like protein 2‐like [G. max]
miR5575	hypothetical protein VITISV_032489 [Vitis vinifera]
miR5634	PREDICTED: kinesin‐4‐like [G. max] PREDICTED: HIPL1 protein‐like [G. max] putative non‐LTR retroelement reverse transcriptase [Arabidopsis thaliana] PREDICTED: peptide methionine sulfoxide reductase B2, chloroplastic‐like isoform 1 [G. max]
miR5636	Polynucleotidyl transferase, Ribonuclease H fold [M. truncatula]
miR6116‐5p	PREDICTED: arginyl‐tRNA synthetase, cytoplasmic‐like [G. max]
miR7503	hypothetical protein MTR_5g051130 [M. truncatula] Mitochondrial protein, putative [M. truncatula] ATPase subunit 1 (mitochondrion) [Lotus japonicus]
miR7752‐3p	Ubiquitin [M. truncatula]
miR7797a	Putative retrotransposon protein, identical [Solanum demissum] putative polyprotein [C. arietinum] Thioredoxin fold [M. truncatula] LIM and UIM domain‐containing [M. truncatula] Cc‐nbs‐lrr resistance protein [M. truncatula]
miR8658	PREDICTED: putative ribonuclease H protein At1g65750‐like [G. max]
miR8717	non‐ltr retroelement reverse transcriptase [Rosa rugosa] PREDICTED: putative ribonuclease H protein At1g65750‐like [Fragaria vesca subsp. vesca]
miR9742	delta‐pyrroline‐5‐carboxylate synthetase [G. max] hypothetical protein MTR_3g030260 [M. truncatula] PREDICTED: probable leucine‐rich repeat receptor‐like serine/threonine protein kinase At5g15730‐like [G. max]
miR9766	Pyruvate kinase [M. truncatula] PREDICTED: transaldolase‐like [G. max]

Figure 7.

The KEGG Plant–pathogen interaction pathway and novel and known miRNAs which obtained in this study possibly targeting the genes involved in this pathway.

Discussion

In plants, small RNAs regulate the gene expression post‐transcriptionally. Deep sequencing strategy is a powerful technology to discover miRNAs in plant species. It has been employed to identify miRNAs in many plant species. Astragalus species are known to accumulate Se in their tissues by converting it to nonamino acid compounds. However, there are still mysterious parts of selenium tolerance mechanism in plants. The aim of this study was to identify Se‐responsive miRNAs and their putative targets by deep sequencing. To achieve this, small RNA libraries were constructed from both control and Se‐treated callus tissues of Astragalus chrysochlorus. Se treatment significantly affected the expression of miRNAs. In total, 418 known and 151 novel miRNAs were identified. When the expression of Se‐treated and control tissues were compared, average normalized reads for known miRNAs were 83.8 and 251.54, respectively, but unknown miRNAs were 93.54 and 65.59, respectively (Tables 1 and 2). The most significant expression difference was occured for miR2867‐3p by 17.8‐fold up‐regulation. miR1869 and miR1507a were up‐regulated by 17.25 and 16.65, respectively. Among the down‐regulated miRNAs, miR1507‐5p was the most down‐regulated one with 17.52‐fold change. Although it was the most up‐regulated miRNA, the potential targets of miR2867‐3p were not annotated. In wheat, miR2867‐3p related to fungal stress was found to target disease resistance protein rga3‐like and categorized in the group of response to stimulus (Inal et al., 2014). miR1507a was found to be related in nitrogen fixation in soya bean (Wang et al., 2009). Despite the fact that it was reported to be down‐regulated by the phosphorus deficiency in Glycine max roots (Zeng et al., 2010), our results showed that miR1507a was up‐regulated significantly in Se‐treated tissues in A. chrysochlorus. The targets of this miRNA were determined as disease resistance protein RGA2 and NBS resistance protein. In another study, Chen et al. (2012) found that miR1507 expression was also decreased after Al treatment. High level of selenium accumulation in hyperaccumulator plant species enables to protection against herbivore and fungal pathogens (Freeman et al., 2010). Consistent with the KEGG pathway analysis, we found that plant–pathogen interaction metabolism is one of the most affected pathway by Se exposure in A. chyrsochlorus. known as secondary accumulator of selenium, in our high‐throughput results. miR535a was also found to be up‐regulated significantly in Se‐related tissues, and its potential targets, transcription factor GTE10‐like, Niemann‐Pick C1 protein‐like, and ATP‐dependent DNA helicase Q1, play role in plant hormone signal transduction pathway.

Van Hoewyk et al. (2008) investigated 40 μm Se stress in Arabidopsis shoots and roots by microarray analysis. The Se‐responsive genes are involved in calcium signalling, ethylene and jasmonic acid synthesis and ethylene‐responsive transcription factor family; stress‐induced and disease‐induced proteins were up‐regulated significantly by selenate treatment as a defence response (Van Hoewyk et al., 2008). In our study, miRNAs, such as miR1507a, miR3633a‐3p, miR7767‐3p and miR395, were found to target stress‐induced and disease‐induced proteins according to GO analysis. While miR1507a, miR3633a‐3p and miR7767‐3p were expressed in Se‐treated tissues, only miR395 was expressed in control tissues. These findings suggest that Se tolerance mechanism may differ by other stress mechanisms. Another miRNA affected by Se treatment in our study is miR393. It is reported that expression of miR393 is regulated by Cd, Hg and Al (Xie et al., 2007). miR393 down‐regulates the F‐box auxin receptors TIR1/AFBs and bHLH transcription factors (Jones‐Rhoades and Bartel, 2004; Navarro et al., 2006). Zhou et al. (2008) reported that miR393, miR171, miR319 and miR529 were up‐regulated in the leaves, but miR166 and miR398 were down‐regulated when Medicago truncatula was treated with Cd, Hg and Al. Decrease in miR166 expression after Cd and Al exposure was also shown by Chen et al. (2012) and Ding et al. (2011). In callus tissues of A. chrysochlorus, miR319 (miR319a‐3p and miR319b) and miR393 (miR393a) family miRNAs were up‐regulated by Se treatment. The fold change of these miRNAs was 8.82, 3.36 and 1.3, respectively. miR319 targets TCP (TeosinteBranched/Cycloidea/PCF) transcription factor (Zhou et al., 2008) and affects the plant growth and development. miR171 family miRNAs (miR171b‐3p, miR171 m, miR171n) were also affected by Se treatment. miR171b‐3p and miR171 m were up‐regulated, and miR171n was significantly down‐regulated by Se exposure. Zhou et al. (2012) reported that miR169 and miR395 were up‐regulated in response to Hg‐toxicity and miR171 was down‐regulated in Medicago truncatula (Zhou et al., 2012). In our study, miR169n‐3p and miR395 were down‐regulated by Se treatment by 10.81‐ and 7.36‐fold, respectively.

Gene ontology analysis was applied for understanding the function of miRNAs' target genes. Among the known miRNAs, miR7800 was found to target the gene encoding cysteine synthase gene of sulphur/selenocompound metabolism that functions in selenocysteine and cysteine synthesis. In this study, miR7800 was up‐regulated by Se exposure, which suggests that cysteine synthase may be down‐regulated. We also found that miR6196 targets γ‐glutamylcysteine synthetase that combines glutamate and methyl‐selenocysteine to generate gamma‐glutamyl‐methylselenocysteine. This compound is thought to be the storage form of Se in hyperaccumulators (Freeman et al., 2007; Kubachka et al., 2007). In our study, miR6196 was down‐regulated by Se treatment. It is thought that Se could be accumulated in different forms. The opposite expression of these miRNAs may be due to plant species, different Se concentration or maybe different exposure time. Certain novel miRNAs target the genes functioning in sulphur metabolism and selenocompound metabolism. miR‐n9 was found to target NADPH‐dependent thioredoxin reductase 3‐like protein. miR‐n69 targets cystathionine beta lyase. This enzyme also plays a role both in sulphur metabolism and selenocompound metabolism. miR‐n116, miR‐n131 and miR‐n147 target serine acetyltransferase in sulphur metabolism. It is thought that these miRNAs should be investigated to understand whether they play a role in Se tolerance and accumulation. KEGG pathway analysis showed that plant–pathogen interaction pathway could be affected in high level by Se treatment in A. chrysochlorus calli. It is found that many genes that encode important proteins and enzymes in this pathway may be repressed by miRNAs with Se treatment. In this pathway, CDPK (calcium‐dependent protein kinase) is thought to be down‐regulated by miR5049‐3p, while MYC2 (transcription factor MYC2) and HSP90 (heat‐shock protein 90 kDa) expressions were decreased by miR9748. Disease resistance proteins RPS2 and RPS5 were down‐regulated by miR3633a‐3p, whereas RPM1 protein was down‐regulated by miR1507a. All these proteins can change the hypersensitive response of the plant, and these results are consistent with the results in Van Hoewyk et al. (2008).

In conclusion, we identified a large number of known and novel miRNAs responsive to Se exposure. We detected 151 novel and 418 known miRNAs, some of them are expected to be regulated by Se treatment. In silico analysis showed that some miRNAs are found to be involved in targeting genes leading to Se metabolism. Se‐responsive miRNAs are mostly involved in plant–pathogen interactions, plant hormone signal transduction, calcium signalling and sulphur metabolism; all these might be related to selenium tolerance and accumulation directly/indirectly (Figure 8). miR1507a, miR3633a‐3p, miR4244, miR5049‐3p, miR5485, miR7767‐3p and miR9748 were expressed significantly by Se exposure. The results we obtained from our study show that miRNAs involved in plant–pathogen interaction pathway may decrease the plant's hypersensitive response and increase the tolerance.

Figure 8.

Possible mechanisms affected by Se‐induced miRNAs.

Materials and methods

Plant material and culturing conditions

The seeds of Astragalus chrysochlorus, collected in 2004 from Sertavul, Karaman, Turkey, were surface‐sterilized with 70% alcohol for 1 min followed by 15 min in 5% bleach and then were rinsed with sterile distilled water three times. The sterilized seeds were germinated on growth‐regulator‐free MS (Murashige and Skoog, 1962) medium (pH 5.8) supplemented with 3% (w/v) sucrose and 0.8% agar (w/v). Callus tissues were proliferated on MS medium supplemented with 0.5 mg/L 2,4‐dichlorophenoxyacetic acid (2,4‐D) and were subcultured every 3 weeks. All tissue culture experiments were carried out in a growth chamber illuminated with fluorescent light (ca. 1400 mol⁻² ms⁻¹) over a 16/8 day and night at 25 ± 2 °C. The selenium treatment of calli was carried out in MS medium supplemented with 0 and 5 mg/L sodium selenate for 21 days. Each treatment was replicated three times and each time treatment was replicated in five individual culture dishes, and each dish contained nine callus tissues. Twenty‐1‐day‐old callus tissues belong to control, and Se treatment were collected and frozen in liquid nitrogen and then stored at −80 °C.

RNA isolation, library construction and sequencing

Total RNAs were isolated for each Se‐treated and untreated samples using TRIZOL (Invitrogen) according to the manufacturer's instructions. Quality and quantity of RNAs were measured with a Nanodrop 2000 spectrophotometer (Nanodrop Technologies, Thermo Scientific, Wilmington, DE). The RNA samples were then pooled for treated and control samples, respectively, and then stored at −80 °C for further analysis. The library construction and Illumina (Solexa) based‐small RNA and degradome sequencing were carried out by the Beijing Genomics Institute (BGI, Shenzhen, China). Briefly, small RNAs were fractionated according to their sizes (18–30 nt long) and ligated to 5′ and 3′ oligonucleotide RNA adaptors. They were reverse‐transcribed and amplified by PCR (Hafner et al., 2008). For degradome sequencing, the cleaved products were ligated to a single‐stranded 5′ RNA adaptor. The ligation products were separated by oligo(dT) cellulose; after cDNAs were synthesized, MmeI digestion was performed (NEB). Ligation of a double‐stranded adaptor to the 3′ end was performed followed by 21 cycles of PCR amplification (German et al., 2008). For both small RNA and degradome libraries, the purified PCR products were sequenced by SBS sequencing with a Solexa/Illumina genome analyser.

Small RNA and Degradome sequencing analyses

After sequencing, the vector sequences were removed first and the sequences between 18 and 30 nt were used for analyses as our previous reports (Xie et al., 2015a; Xie et al., 2015b). Briefly, miRNA length distribution was identified for clean, common and spesific sequences. Small RNA annotation was performed using tag2annotation software (BGI) to analyse the length distribution. The RNA‐Seq data of A. chrysochlorus (unpublished data) obtained from our study was used as reference genome for mapping with the clean sequences by SOAP (Short Oligonucleotide Alignment Program) to determine the distribution on the genome (Li et al., 2008), as no available genome sequence information for this plant. The sequences that were matched up precisely were used for further analysis. The obtained small RNA sequences were compared to the GenBank (Benson et al., 2006) and Rfam database (11.0) (Gardner et al., 2009) by alignment and BLAST search to investigate rRNA, tRNA, snoRNA, snRNA and scRNA. The small RNAs were annotated with the priority rule [rRNAetc (in which GenBank > Rfam) > known miRNA > repeat > exon > intron] (Calabrese et al., 2007). To screen the known miRNAs, alignments were performed against the miRBase (Release 20) database (Griffiths‐Jones et al., 2008) and differentially expressed known miRNAs among the samples that were identified previously. The small RNAs, which were unannotated, were screened for prediction of novel miRNAs using Mireap (BGI) with default parameters. The same method was used for degradome sequencing analysis, clean reads were generated and the alignments were performed against the A. chrysochlorus RNA‐seq data, Rfam (11.0) database and GenBank database. The filtered reads were aligned to exons and introns of A. chrysochlorus mRNAs to investigate the fragments of degraded mRNAs. Then, clean tags were mapped to the reference genome (A. chrysochlorus RNA‐seq data) using SOAP2.20 (Li et al., 2009) with allowing only two mismatches. 5′‐position of sequences were chosen to predict miRNA cleavage sites using CleaveLand pipeline v3.0.1 (Addo‐Quaye et al., 2009). Target (T‐) plots were generated evaluating the position 10 or 11 of the miRNA.

Target prediction and functional classification based on GO and KEGG analyses

To predict potential mRNAs of A. chryschlorus, psRobot Small RNA Target Prediction Tool was used with default parameters (Wu et al., 2012). Glycine max (Soybean, JGI Glyma1.0 annotation of the chromosome‐based Glyma1 assembly) and Medicago truncatula (Barrel medic, Release Mt3.0 from the Medicago Genome Sequence Consortium) were selected as transcript libraries for target analysis. Gene Ontology (GO) provides a dictionary to render the characteristics of genes and their products. The results are classified by three ontologies in GO and these are molecular function, cellular component and biological process. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis can also be used for the putative target genes. To determine the function of A. chryschlorus miRNAs, we used Blast2GO (http://www.blast2go.com) to determine the predicted target genes. As a first step, the mRNAs, which were targeted by miRNA, were aligned with BLASTX against nr database. As a second step, the hits were subjected to the GO and KEGG databases. Pathway enrichment and categorization were performed using GO (http://www.geneontology.org/) and KEGG databases (http://www.genome.jp/kegg/kegg1.html).

Supporting information

Figure S1 Size distribution of small RNA sequences in Se‐treated and untreated‐ Astragalus chrysochlorus.

Table S1 Summary of deep sequencing dataset.

Table S2 Small RNA deep sequencing data from Se‐treated and untreated Astragalus chrysochlorus.