Introduction

MicroRNAs (miRNAs) have emerged as important post-transcriptional and translational regulators that have been implicated in playing an important roles in a wide variety of stress conditions including heat, drought, salinity, heavy metal, chilling temperature, nutrient stress and disease1. In plants, miRNA genes are transcribed by RNA polymerase II to produce primary miRNA (pri-miRNA), further processed to yield precursor miRNA (pre-miRNA). Pre-miRNAs have defined stem loop secondary structures that are processed by the dicing complex. Core components of the dicing complex include DICER‐LIKE1 (DCL1/RNase III), HYPONASTIC LEAVES1 (HYL1/DRB1) and SERRATE (SE) that act to yield mature miRNA/miRNA* duplexes2. The miRNA-miRNA* duplex is subsequently exported to the cytoplasm where they are 2′‐O‐methylated at the 3′ end by HEN1. One strand from the duplex is incorporated into ARGONAUTE1 (AGO1) to form an active RNA‐induced silencing complex (RISC) and the complementary miRNA* is degraded. DCL proteins appear to function as molecular rulers that measure and cleave small RNA duplexes at a specific length. The precision in the position of the first cut along the precursor is of utmost importance as the second cut is usually made by measuring a fixed distance from the precursor end3. Based on conservation of specific structural elements, both base to loop and loop to base modes of processing have been recognized to exist in plant precursor miRNAs3,4

More than forty miRNA gene families are associated with abiotic stress in plants. Of these, thirteen families of miRNAs are responsive to salt and drought stress5. A large number of salinity regulated miRNAs have been identified in switchgrass that play either a direct or indirect role in salt stress alleviation6. Upregulation of miR156, miR158, miR159, miR165, miR167, miR168, miR169, miR171, miR319, miR393, miR394, miR396 and downregulation of miR397, miR398 has been reported under salinity stress in Arabidopsis7. While some information is available about miRNAs and their genes from cultivated rice and certain wild rice species, data on genes and information on their regulation in wild rice species under salinity is limited or in certain cases, totally lacking8,9,10,11,12. In cultivated rice, Oryza sativa, certain members of the miR396 gene cluster function as negative regulators under salinity13,14,15,16. The phylogeny and evolution of the miR396 family has been analysed only in certain AA genomes of Oryza species, including Oryza sativa12. Of the eight MIR396 genes found in O. sativa japonica cv. Nipponbare, two MIR396 genes (osa-MIR396a and osa-MIR396c, the latter on the reverse strand) are clustered in a less than 10 kb genomic region on Chromosome 2.

The genus Oryza is represented by 27 species, broadly divisible into 11 genome types, of which 6 are diploid (n = 12: AA, BB, CC, EE, FF and GG) and 5 are polyploid [n = 24: BBCC, CCDD, HHJJ, HHKK and KKLL17. In this study, we report the isolation of pre-miR396c sequences from wild Oryza species that represent seven genome types and examine their structural characteristics vis-à-vis their sequence. Absolute conservation of the miR396c duplex region and also a substantial degree of conservation of certain elements of the secondary structure of pre-miR396c was observed. An SNP variation (A/G; rs10234287911) was identified in the pre-miR396c -sequence in O. sativa genomes18 and validated in a set of forty-three O. sativa landraces. The effect of the SNP variation on miR396c expression levels under salinity and elevated temperature in O. sativa landraces has also been examined. In addition, expression of pri-miR396c, pre-miR396c and miR396c target genes under salinity and heat is also reported.

Results

Mature miR396c sequence is conserved across most Oryza species Pre-miRNA sequences from Oryza species for miR396c were retrieved from the NCBI database using BLAST. Since pre-miRNA sequences retrieved are reported overwhelmingly for diploid Oryza (AA) genomes19, we attempted to PCR amplify genomic stretches corresponding to the pre-miRNAs from both diploid (other than AA genomes) and tetraploid Oryza species belonging to other genotypes and also sought to validate data currently available for diploid AA Oryza genomes. The Oryza wild species in this study represents seven genome types (AA, BB, CC, EE, BBCC, CCDD and KKLL) of 10 representative genome types recognized by the OMAP resource20. The twelve Oryza genomes analysed in this study are listed in Table 1. Alignment of these sequences suggests that there is no variation in the mature miRNA regions of miR396c in the analysed Oryza species (with the exception of O. coarctata; Supplementary Fig. 1), though SNPs are observed in the pre-miRNA sequences (Fig. 1A & B).

Table 1 List of Oryza species used for the current study and accession numbers of precursor miR396c sequences.
Full size table
Figure 1
figure 1

Multiple Sequence alignment (MSA) and phylogenetic analysis of pre-miR396c in Oryza species. For eleven Oryza species, a high degree of conservation of the pre-miR396c sequence was observed. (A): Alignment of pre-miR396c sequences from Oryza species. Conservation of miRNA (Red) and miRNA* (Yellow) sequences are indicated. Seven SNPs and one indel were identified between miRNA and miRNA* regions and bases are highlighted in pink or green. Of these, residues marked #1-#3 were heterozygous in O. punctata (listed as O.punctata1-6 for ease of representation) and #4 heterozygous in O. sativa (japonica as Osj1/Osj2), O. sativa (indica as Osi1/Osi2), O. australiensis (as O.australiensis1/ O.australiensis1) and O. barthii (as O.barthii1/O.barthii2). A single base pair insertion at the same position was observed in pre-miR396c sequences of both O. punctata and O. officinalis (highlighted in blue). (B): Optimal Neighbour-joining tree generated using MSA data (MEGA11). Evolutionary distances were computed using the Maximum Composite Likelihood method, represented as the number of base substitutions per site. The genome type for each Oryza species is indicated within brackets.

Full size image

Seven SNPs and one indel are present in the stretch between the miRNA-miRNA* regions in the eleven pre-miR396c Oryza sequences examined. Among 7 SNPs detected, four SNPs were due to heterozygosity in selected Oryza species (Fig. 1A). A single base pair insertion at the same position was observed in pre-miR396c sequences of both O. punctata and O. officinalis. Halophytic wild rice O. coarctata pre-miR396c sequence is significantly different from O. sativa pre-miR396c and shows six base changes in the miRNA-miRNA* region (three each in miRNA and miRNA* regions; Supplementary Fig. 2). The pre-miR396c sequences from eleven Oryza species are 141–142 bp in length while for O. coarctata the pre-miR396c sequence is substantially reduced (133 bp). This 9 bp deletion falls within the intervening miRNA-miRNA* region. The O. coarctata pre-miR396c sequence reported here is identical to that present in the O. coarctata genome21 (Supplementary Fig. 2), suggesting miR396c function in halophytic wild rice may be significantly different.

Pre-miR396c secondary structures from Oryza species can be divided into three groups

The predicted secondary structure of pre-miR396c sequences from twelve Oryza species were examined vis-à-vis their sequence using Mfold22 (Fig. 2-1 and 2-2). Based on the secondary structures, the twelve pre-miRNA sequences were divided into three groups. Within group 1 and 2, certain pre-miR396c structural features are absolutely conserved. These include the miRNA-miRNA* regions (conserved across eleven Oryza species). In addition, a five bp bulge 3′ relative to the miRNA-miRNA* region and a 9 bp lower conserved region (LR) 5′ relative to the miRNA-miRNA* duplex are also found to be absolutely conserved in the eleven Oryza species. Group I contains pre-miRNA secondary structures from O. punctata and O. minuta that differ in length by one base (O. punctata; 142 bases; Fig. 2-1A–F). The O. punctata pre-miR396c sequence shows heterozygosity at positions 74 and 79 (M = A or C), in addition to insertion of a single base at position 102. As a result of heterozygosity at positions 74 and 79, variations in O. punctata pre-miR396c structures are observed that, in turn, result in variation of ΔGfolding from − 60.98 to − 65.39 kcals/mol (Fig. 2-1). These structures all show two conserved bulges in the Upper region (UR), including a Branched Terminal Loop (BTL) that differ slightly in their structure, contributing to variation in ΔGfolding. The O. minuta pre-miR396c structure also shows a BTL in the UR (Fig. 2-1G). Further, the second bulge in the base paired region below the BTL in O. minuta has 8 bases and is shifted in its position relative to that in O. punctata.

Figure 2
figure 2

Predicted secondary structures for pre-miR396c using Mfold. Pre-miR396c structures from twelve Oryza species can be divided into three groups. Three elements are absolutely conserved across eleven Oryza spp. Pre-miR396c structures: (i) miRNA-miRNA* duplex (miRNA in green)(ii) a five bp bulge above the miRNA-miRNA* region (indicated by a grey box) (iii) 9 bp lower conserved region (LR) below the miRNA-miRNA* duplex (of which 6 bases show pairing). (A–G): Group I contains pre-miR396c structures from O. punctata (142 bases) and O. minuta (141 bases). O. punctata pre-miRNA 396c sequence shows heterozygosity at positions 74 and 79 (M = A or C) in addition to insertion of a single base at position 102 (indicated by red arrows; Fig. 2-1). (H–J): Group II contains. officinalis, O. alta and O. glaberrima pre-miR396c structures differing only in an ‘A’ or ‘G’ at the 103 position of the pre-miR396c sequence (Fig. 2-2). (K): Group 3 contains only O. coarctata pre-miR396c. The miRNA-miRNA* duplex region is indicated in green and perfect pairing is disrupted due to 6 SNPs.

Full size image

Among the wild and cultivated Oryza species examined, six Oryza species showed identical sequences within Group 2 and only variation at position 103 (G or A); only G: O. glaberrima, only A (O. officinalis, O. alta, O. nivara, O. rufipogon); heterozygosity (G/A; O. barthii, O. australiensis) in the specific accessions examined. In the two cultivated Asian rice accessions examined, (O. sativa sub. indica; O. sativa sub. japonica), also falling within Group 2, position 103 showed heterozygosity (G/A) in the pre-miR396c sequences. In Group 2, the O. officinalis pre-miR 396c secondary structure has one additional bulge in the UR; O. alta and O. glaberrima, pre-miR396c structures show two bulges, two loops and three paired stems in the UR; (Fig. 2-2H-J). O. alta and O. glaberrima structures differ at position 103 (R = G or A) in the UR of the pre-miR396c sequence. The presence of an “A” at the 103rd position results in a perfect U38-A103 base pair while the presence of a ‘G’ at the 103rd position disrupts this base pairing, resulting in a minor increase in the ΔGfolding of the pre-miR396c by -0.8kcals/mol.

The O. coarctata pre-miR396c structure is unique and occurs in group 3 (Fig. 2-2K). It shows one bulge, three loops and three stems in the UR region. The O. coarctata pre-miR396c secondary structure in addition has other unique features: (i) the 6 base variation in the miRNA-miRNA* region results in a loop at the 3′ end of the mature miRNA sequence, (ii) near perfectly base paired LR (eight of nine bases paired), (iii) a G-U wobble base pair in the LR (iv) a 7 bp terminal loop and (v) a low folding energy (-55.3 kcals/mol) compared to other Oryza species.

A conserved SNP variation (A/G; rs10234287911) is present in pre-miR396c sequences within the genus Oryza. Pre-miR396c diversity across dicots, monocots and 3024 cultivated rice genomes was examined and is represented using a Circos plot (Fig. 3)23. The outermost ring is indicative of overall nucleotide sequence conservation across dicotyledonous and monocotyledonous species, with a darker shade indicating a higher conservation score. In case of pre-miR396c we have presented data for 74 precursors from 34 dicot species and 17 precursors from 7 monocot species. Ring II indicates the frequency and distribution of paired (blue bars) and unpaired bases (brown bars) for every position in the pre-miRNA precursor, with inner darker blue connecting ribbons representative of base pairing across Oryza species. The O. coarctata pre-miR396c sequence was not included in the Circos analysis because of a substantial variation in length vis-à-vis other Oryza species (133 bases vs. 141 bases). The division of the pre-miR396c sequences from Oryza species into three groups, based on secondary structure, is represented by orange (Group 1), dark green (Group 2) and light blue (Group 3) ribbons respectively. Rings III, IV and V represent pre-miR396c sequence conservation across Oryza species, dicotyledonous and monocotyledonous species respectively. Ring VI indicates nucleotide conservation of pre-miR396c sequences across 3024 cultivated rice O. sativa genomes. High conservation of the miRNA-miRNA* regions was observed in both dicots and monocots; further, high conservation was also observed in the lower stem region, preceding the miRNA-miRNA* duplex in dicots as well as monocots.

Figure 3
figure 3

Circos analysis of sequence and structural conservation of pre-miR396c Oryza, dicot and monocot species. Ring I represents overall conservation of bases; ring II represents a histogram depicting frequency of base pairing (blue bars) and unpaired bases (brown bars) in Oryza species. Ring III represents conservation score for each base among Oryza species; darker colour indicates a higher conservation score (9) while a lighter colour indicates lower conservation score (1). Ring IV and Ring V represents the conservation scores observed for dicot and monocot species respectively. Ring VI represents the SNP observed in 3024O. sativa genomes.

Full size image
Figure 4
figure 4

(A) Alignment of miR396 mature sequences from O. sativa. Mature miR396a-c sequences group more closely (boxed in red) relative to miR396d-h. (B): Pictorial representation of the miR396c locus in O. sativa (Os02g0804000; negative stand; chromosomal coordinates indicated) with corresponding non-coding RNA (ncRNA; XR_003240687.1) and cDNA/EST sequences reported in the NCBI database. Exonic segments are indicated in blue, and the intronic region in green. The region marked in yellow in exon 1 corresponds to the pre-miR396c region. Numbers adjacent to dotted green, blue lines (5′ end) and red lines (5′ end) next to cDNA/EST sequences indicates corresponding base in the miR396c genomic sequence (Os02g0804000). Of these, cDNA (AK062523.1) shows partial intron retention.

Full size image

As observed above for the wild and cultivated Oryza species in group 2, a single SNP variation (A/G; rs10234287911) corresponding to the same 103rd position in the pre-miR396c sequences was also observed in the examined 3024 O. sativa sequences (3K O. sativa rice genomes). This SNP variation occurs in the stem structure of the pre-miRNA396c, in the UR just above the miRNA-miRNA* paired region (Supplementary Fig. 3). As already mentioned, an ‘A’ at the 103rd position in the O. sativa pre-miRNA sequence results in a perfect U38-A103 base pair while the presence of a ‘G’ at the 103rd position disrupts this base pairing, resulting in a minor increase in the ΔGfolding of the pre-miR396c by  − 0.8 kcals/mol. In addition, the presence of A103 results in a contiguous stretch of seven base paired residues just above a five bp bulge in the upper region in contrast to the presence of G103, resulting in 5 base paired residues flanking U38-G103 and G39-U102. Of the 3024 O. sativa genomes examined, 2184 genomes had ‘G’ at the 103rd position while 740 O. sativa genomes had ‘A’; in addition, 83 O. sativa sequences were heterozygous for this allele in pre-miR396c (Supplementary Table 1), similar to the data obtained for heterozygosity in O. sativa (indica and japonica; R = A or G) as well as O. barthii and O. australiensis from our sequencing study. In addition, only a ‘G’ or ‘A’ was found at position 103 in all Oryza species except O. coarctata, and is suggestive of evolutionary selection for either an ‘A’ or a ‘G’ base at the 103rd position.

Pri-miR396c in O. sativa shows complex post-transcriptional processing

Since MIR396 is a multigene family in O. sativa, mature miR396a-h sequences were compared. Of these, miR396a-c showed near sequence identity, with mature miR396c differing from miR396a/b by one base (3' terminal U; Fig. 4A). This suggests that miR396c can be distinguished from miR396a/b using specific probes or primers. A noncoding RNA (XR_003240687.1) corresponding to MIR396c (Os02g0804000) was retrieved. In addition, cDNA/ESTs corresponding to the O. sativa pre-miR396C cDNA: AK062523.1; ESTs: CT829989.1, CI029080.1, CK053360.1, CK045291.1, CT856578.1, CT850193.1 and CI296040.1) were also retrieved. The genomic sequence of Osa-MIR396c (Os02g0804000) was aligned with the non-coding RNA (XR_003240687.1), showing the presence of two exons and an intervening intron (Fig. 4B). Of the cDNA/ESTs retrieved, only CT829989.1 showed correct splicing but retained only 132 bp as part of the second exon while other cDNAs/ESTs showed partial intron retention (Supplementary Fig. 4). The presence of intron retaining pri-miR396c isoforms in cultivated rice, O. sativa, suggests that MIR396c regulation and osa-miR396c biogenesis might be complex. Extending analysis to the entire Os02g0804000 locus (inclusive of the intronic segment) led to the identification of 27 SNPs and 7 indels among 3024 rice accessions (pri-miR396c region, including intronic segment). The indels include a GA repeat region in the first exon (following the pre-miR396c) and a TA repeat region in the intron.

Expression of miR396c, pre-miR396c and pri-miR396c in rice landraces under salinity and heat stresses

The SNP variation (A/G; rs10234287911) identified in the Oryza spp. above was validated in additional 43 landraces from coastal regions of India24. PCR primers were designed to amplify a 141 bp fragment of pre-miR396c (Supplementary Fig. 5). Sequence analysis revealed that among 43 landraces, 22 showed the presence of an ‘A’ base at the 103rd position whereas 21 landraces had a ‘G’ at the same position. Thus, both bases appeared to be present at almost equal frequency in the examined landraces. Since this SNP variation occurred outside of the miRNA-miRNA* region, we examined if the variation may have a role in miR396c biogenesis from pre-miRNA. Over-expression of miR396c is associated with reduced salinity tolerance in rice13. Further, higher temperatures may alter pre-miRNA structure stability and hence miR396c abundance under heat stress (measured as miR396c expression) was examined. Therefore, we randomly selected seven landraces, three having either an ‘A’ (Pokkali, IR28 and FL478) and four with a ‘G’ (Nona Bokra, Anakodan, Aduisen and Orkyma) at the 103rd position in the pre-miR396c sequence and we examined miR396c expression in the landraces/varieties under salinity and heat stress conditions with appropriate untreated controls. Under control, salinity and heat treatments, miR396c expression in landraces/varieties grouped either ‘A’ or ‘G’ differed significantly as measured by statistical pair wise (Students t-test) analysis (Fig. 5A-C). MiR396c expression under salinity was increased significantly in landraces/varieties grouped ‘G’ while there was no significant difference in landraces/varieties grouped ‘A’ (Fig. 5D). Heat stress did not bring about a significant change in miR396c expression in landraces grouped either ‘A’ or ‘G’ (Fig. 5D). Further, landraces/varieties with an ‘A’ at the 103rd position in the pre-miR396c sequences showed more tightly clustered expression under control conditions, while landraces/varieties with a ‘G’ at the 103rd position in the pre-miR396c sequences showed more distributed expression (Fig. 5A–C and E–G). Pre-miR396c expression under control (untreated), salinity and heat were also examined (Fig. 6A–G). Again, under salinity and heat treatments, pre-miR396c expression in landraces/varieties grouped either ‘A’ or ‘G’ differed significantly (highly significant under heat stress; Fig. 6A–C and E–G). Pre-miR396c expression was increased highly significantly under both salinity and heat stress in landraces/varieties grouped ‘A’ and to a lesser extent in landraces grouped ‘G’ (Fig. 6D). Significant positive correlation was seen between pre-miR396c and miR396c expression under control conditions in landraces/varieties grouped ‘G’ and under salinity treatment for landraces grouped ‘A’ (Supplementary Fig. 6A and B respectively). Finally, pri-miR396c expression under control (untreated), salinity and heat was also examined in the same set of samples (Fig. 7A–G). Pri-miR396c expression under control and heat treatments, differed significantly in landraces/varieties grouped either ‘A’ or ‘G’ but not under salinity (Fig. 7A–C and E–G). Further, pri-miR396c expression was significantly upregulated under heat stress in landraces/varieties grouped ‘A’ and to a lesser extent under both salinity and heat in landraces grouped ‘G’ (Fig. 7D). A significant correlation was found between pri-miR396c and pre-miR396c accumulation under control and heat stress conditions in landraces/varieties grouped ‘G’ and heat stress and salinity in landraces/varieties grouped ‘A’ Supplementary Fig. 6C–F). Expression of miR396c target genes (OsTBP and OsGRF3) in the same biological samples was examined under control, salinity and heat treatments. Overall, qRT-PCR data for the samples shows that the expression level of target genes was inversely correlated with miR396c expression and in most cases, statistically significant (Fig. 8A–C).

Figure 5
figure 5

qRT-PCR analysis of mature miR396c expression under control (untreated), salinity and heat treatments in rice landraces/varieties grouped either ‘A’ or ‘G’. Violin plots representing ΔCT values for miR396c expression under control conditions in (A), salinity in (B) or heat stress in (C) for landraces/varieties grouped either as ‘A’ or ‘G. In (E–G), ΔCT values for miR396c expression for individual landraces under the same conditions is shown. Pairwise statistical analysis of miR396c expression in landraces/varieties grouped either ‘A’ or ‘G’ (A–C) or within landraces grouped ‘A’ or ‘G’ (control vs salinity or heat) using pairwise Student’s t-test (GraphPad V 6.0). Data is for three biological replicates per landrace/varieties per treatment (n = 3) with two/three technical internal replicates per treatment. ****P < 0.0001, **P < 0.01; ns: non significant.

Full size image
Figure 6
figure 6

qRT-PCR analysis of pre-miR396c expression under control (untreated), salinity and heat treatments in rice landraces/varieties grouped either ‘A’ or ‘G’. Violin plots representing ΔCT values for pre-miR396c expression under control conditions in (A), salinity in (B) or heat stress in (C) for landraces/varieties grouped either as ‘A’ or ‘G. In E–G, ΔCT values for pre-miR396c expression for individual landraces under the same conditions is shown. Pairwise statistical analysis of pre-miR396c expression in landraces/varieties grouped either ‘A’ or ‘G’ (A–C) or within landraces/varieties grouped ‘A’ or ‘G’ (control vs salinity or heat) using pairwise Student’s t-test (GraphPad V 6.0). Data is for three biological replicates per landrace/varieties per treatment (n = 3) with two/three technical internal replicates per treatment. ****: P < 0.0001, **P < 0.01; P < 0.05; ns: non significant.

Full size image
Figure 7
figure 7

qRT-PCR analysis of pri-miR396c expression under control (untreated), salinity and heat treatments in rice landraces/varietiesgrouped either ‘A’ or ‘G’. Violin plots representing ΔCT values for pri-miR396c expression under control conditions in (A), salinity in (B) or heat stress in (C) for landraces/varieties grouped either as ‘A’ or ‘G. In (E–G), ΔCT values for pri-miR396c expression for individual landraces/varieties under the same conditions is shown. Pairwise statistical analysis of pri-miR396c expression in landraces grouped either ‘A’ or ‘G’ (A–C) or within landraces/varieties grouped ‘A’ or ‘G’ (control vs salinity or heat) using pairwise Student’s t-test (GraphPad V 6.0). Data is for three biological replicates per landrace/varieties per treatment (n = 3) with two/three technical internal replicates per treatment. ****: P < 0.0001, **; P < 0.05; ns: non significant.

Full size image
Figure 8
figure 8

Expression analysis of (A) miR396c and target genes (B) OsGRF3 and (C) OsTBP under control, salinity and heat stress treatments in selected landraces/varietiesgrouped either ‘A’ or ‘G’. Pairwise statistical analysis of miR396c or OsGRF3 or OsTBP expression in landraces grouped either ‘A’ or ‘G’ (A–C) for each landrace/varieties (control vs. salinity or heat) using pairwise Student’s t-test (GraphPad V 6.0). Data is for three biological replicates per landrace/varieties per treatment (n = 3) with two/three technical internal replicates per treatment. a: P < 0.0001; b: P < 0.001; c: P < 0.01, d: P < 0.05; e: non significant.

Full size image

Small RNA northern blot analysis from landraces grouped ‘A’ or ‘G’, under control (Supplementary Fig. 7A), salinity and heat treatments (Supplementary Fig. 7B, C; Supplementary Fig. 12) did not show any differences in miR396c expression levels. Mature miR396a differs from miR396c at the terminal 3' base. However, miR396a expression levels are considerably lower as compared to miR396c, as longer autoradiographic exposure was required (5 days) to obtain a signal compared to miR396c (3 days; Supplementary Fig. 7A). Given that probes can occasionally cross-hybridize, RNA-seq data from25 was used to confirm expression level differences between miR396a and miR396c (Supplementary Fig. 8).

Discussion

Plant miRNA precursors show high structural diversity and are processed by different modes to produce mature miRNAs 3,4,20. Oryza species span ~ 15 million years of evolution and the absolute conservation of mature miRNA sequences of miR396c (O. coarctata is an exception) in eleven Oryza genome types is suggestive of a strong evolutionary constraint to maintain its interaction with target genes in the genus Oryza.

Pre-miR396c sequences show absolute conservation of miRNA and miRNA* regions in most Oryza species examined. Based on the pre-miR396c structures, three distinct structural groups could be recognized, with O. coarctata pre-miR396c forming a separate group. As mentioned previously, three elements were absolutely conserved across the eleven Oryza pre-miR396c structures (miRNA/miRNA* regions, 5 base bulge 3′ to miRNA/miRNA* region and 9 bp stem in LR) and may be indicative of base to loop pre-miRNA processing, similar to miR172a4. Group 1 members show the presence of a branched terminal loop (BTL). BTLs seen in Arabidopsis pri-miR166 alter processing direction from a base-to-loop mode to a bidirectional mode that has implications on miRNA abundance26. BTLs reduce the efficiency of Arabidopsis miR157c processing and replacement by a 4 nucleotides loop increases processing efficiency27. Group 2 members show a common conserved 10 bp terminal loop (TL) in the UR. In animal systems, the primary miRNA transcript is processed to pre-miRNA by Drosha which is then processed by Dicer to generate the mature miRNA-miRNA* duplex. Around 14% of all human pri-miRNAs have conserved terminal loops that are indicative of a requirement to provide a docking site for auxiliary factors that govern pri-miRNA processing28. Lin28 (abnormal cell lineage factor 28) interacts with the miRNA precursors of the let-7 family, by recognizing different elements of the TLs of miRNA precursors while multifunctional KSRP [KH (K-homology) splicing regulator protein] interacts with pre-let-7a, recognizing a G-rich site in the TL29. The presence of the conserved 10 bp loop sequence in eight Oryza species is strongly suggestive of its recognition as an important element in the pre-miR396c processing machinery. In O. coarctata this 10 bp stretch is absent and may have different miRNA processing from other Oryza species.

The comparatively higher ΔGfolding of O. coarctata pre-miR396c vis-à-vis other Oryza species may have implications for processing. In Arabidopsis, allelic variation in the miR824-encoding locus favours a thermal resistant substructure in the precursor that impacts processing of mature miR82430. In other Oryza species examined, the 9 bp LR shows 6 paired bases, while in O. coarctata 8 of 9 bases shows pairing that also includes a unique U-G wobble. This U-G wobble arises due to an SNP in the LR just below the miRNA* region. Directed substitutions and variation analysis of Arabidopsis MIR390a and MIR390b loci shows that base pair properties and nucleotide identity 4–6 bases below the miR390/miR390* duplex region contributes to the efficiency and accuracy of precursor processing31. O. coarctata pre-miR396c shows unique ten bp loop within the miRNA-miRNA* region (towards the 3′ end of the mature miRNA sequence). This loop arises due to two SNPs in the mature miRNA and 2 SNPs in the miRNA* 396c sequence at 3′ end. A. thaliana miRNA:miRNA* duplexes feature a high number of natural polymorphisms that can affect base pairing and thus reduce accumulation of mature miRNAs32. The 3 SNPs in the putative miRNA region of O. coarctata miR396c are also likely to alter target specificity.

miR396c in O. sativa can be distinguished from other family members due to sequence specific differences or expression level variation. This has been observed in this study as well as others33,34. SNPs in pri-miRNA, pre-miRNA and mature miRNA sequences can potentially affect the maturation of miRNAs, their expression level and base pairing at the target site respectively, each mechanism in turn controlling/contributing to gene regulation. Predominantly, polymorphisms in pre-miRNA can influence miRNA maturation and thereby regulate miRNA expression35. We identified a G/A SNP variation (rs10234287911) at the 103rd position of O. sativa pre-miR396c examined sequences that potentially disturbs an A-U base pair if G replaces the A, resulting in the introduction of a G-U wobble pair destabilizing a perfectly base paired stem region that follows the miRNA-miRNA* region and a minor decrease in the ΔGfolding of the pre-miRNA 396c by 0.8 kcals/mol (Supplementary Fig. 3). Mutations that destabilize the region below the terminal loop in Arabidopsis precursor miR172a resulting in an open structure abolishes miR172 expression while mutations that stabilize the structure of the loop do not affect miR172a biogenesis36. An SNP in pre-miR1666 decreases mature miRNA expression and is associated with chicken growth traits37. Structurally different alleles of Arabidopsis precursor miR824 likely show differential processing of mature miR824 in A. thaliana30. Interestingly, of the twelve sub-populations recognised in the sequenced rice genomes18, four clusters including East Asian temperate (GJ-tmp), Southeast Asian Subtropical (GJ-sbtrp), South East Asian tropical (GJ-trp) and admix (GJ-admix) categories showed a clear 100% presence of the major ‘G’ allele. The ‘G’ allele also predominates in six other recognized categories in the same data set. On the other hand, the minor allele ‘A’, predominates in two groups containing Aus, Boro and Rayada ecotypes from Bangladesh and India (circum Aus group-cA) and the Basmati and Sadri aromatic varieties (circum Basmati group-cB) (Supplementary Fig. 9; Supplementary Table 1). The minor allele ‘A’ occurs in samples taken from South/South East Asian countries (Supplementary Fig. 9; Supplementary Table 2) while the major allele appears to predominate in South-East Asian countries (samples from India, China occurs in equal numbers in both categories). More precise passport data for the samples might help to correlate more precisely the geographic coordinates with geographical distribution of the SNP.

The G/A SNP variation occur at almost equal frequencies in the 43 landraces/varieties examined. Further, a significant correlation and distinction appears to exist between the two groups (G/A) vis-à-vis expression levels of the mature miR396c in leaf tissues of O. sativa landraces/varieties under control salinity or heat stress. This, in turn, impacts expression of target genes OsTBP and OsGRF3. Further, miR396c expression under salinity is downregulated in most landraces/varieties and as reported previously by13. However, only under salinity there is a significant miR396c expression change in the landraces grouped G. Given the association of miR396c with salinity stress in rice, it may be that specific factors control miR396c levels by pre-miR396c processing under salinity but not heat stress. Many regulatory proteins influencing the folding, stability, and/or processing of pre-miRNAs have been identified and described2,38 suggesting that pre-miRNAs are modified, folded, and processed co-transcriptionally39. More recently, it has been suggested secondary structural features of precursor miRNAs may have a role in not only controlling precursor miRNA processing, but also RISC loading efficiency onto AGO140,41. SNPs in Arabidopsis miRNAs (ecotypes) that are predicted to have subtle effects on pre-miRNA secondary structure mostly occur in the double-stranded stem regions42. Further, if the SNPs have structural effects, these were found to be subtle and did not affect general integrity of the stem-loop structure. This may be similar to the SNP identified in this study. However, Northern analysis of landraces/varieties grouped ‘A’ or ‘G’ under the same set of conditions did not reveal significant expression level differences. Overall, miRNA quantification methods differ in specificity and accuracy43.

Extending expression analysis to pre-miR396c, significant differences were observed between landraces/varieties grouped ‘A’ or ‘G’ under salinity and heat. For pri-miR396c significant differences were observed between landraces/varieties grouped ‘A’ or ‘G’under control and heat stress conditions. The promoter region of the MIR396c locus (Os02g0804000) is enriched in ABRE elements and has TC-rich regions13. Further both pri-miR396c and pre-miR396c expression are upregulated in most landraces/varieties examined under heat. Also, pre-miR396c expression is upregulated under salinity in most landraces but miR396c expression is downregulated in most landraces/varieties under salinity (Pokkali and Orkyma are exceptions). There may thus be mechanisms controlling miR396c biogenesis in rice under salinity in leaf tissues. A partial intron retaining O. sativa miR396c cDNA (pri-miR396c; AK062523.1; Fig. 4), when constitutively overexpressed in Agrostris stolonfera (creeping bentgrass), is processed correctly to give mature miR396c and is associated with stunted growth but increased salinity tolerance44. Further, transgenic bentgrass lines bypass a vernalization requirement for flowering45. Thus, intron retaining pri-miR396c transcripts appear to be functional. The above data suggests complex transcriptional and post-transcriptional (including intron retention) mechanisms of miR396c regulation in rice.

The conservation of SNP rs10234287911 in the pre-miR396c sequence in wild as well as cultivated Oryza species suggests a possible role in mature miRNA biogenesis. Replacing the G/A hypothetically with either C or T results in miR396c secondary structures with significantly increased free energy values and lowered stability (Supplementary Fig. 4). Further, analysis of the 3000 rice genome data shows (i) the predominance of one allele (G) in the analysed population, (ii) exclusive or near exclusive presence in certain subpopulations of the major allele (Supplementary Fig. 9) and the pre-dominance of the minor allele in Aus and aromatic sub-populations (Supplementary Fig. 10). The pre-miRNA structure with the major ‘G’ allele shows less structured base pairing in the region closer to the terminal loop while minor ‘A’ allele has a more stable structure. MiRNA maturation efficiency may be governed by dynamic structural transitional states of pri- and precursor miRNAs (including 5′ UG/GU 3′ wobble pairs), adding another possible layer in regulation of miRNA biogenesis in response to environmental or developmental cues43,46. We hypothesize that the more stable structure conferred by the ‘A’ allele may be processed more readily vis-à-vis the structure conferred by the ‘G’ allele (with 5′ UG/GU 3′ wobble pairs). This may be reflective of the distribution pattern of mature miR396c expression levels in the two allele types. Agroinfiltration of cloned of O. sativa pre-miR396c structures with G103 or A103 variation(s) would help in determining conclusively the role of this SNP, if any, in controlling mature miR396c abundance.

Conclusion

We present for the first time, pre-miRNA sequence data and secondary structural characteristics for miR396c across Oryza species. Conserved structural determinants within precursor miR396c sequences suggest a conserved base-to-loop mode of miRNA biogenesis. Pre-miR396c from halophytic O. coarctata shows substantial variation in length, sequence of precursor as well as miRNA/miRNA* regions that can impact biogenesis and target specificity. Further, a SNP identified in the precursor region of miR396c sequences (3024 O. sativa genomes) may have implications for miRNA abundance and target gene regulation under salinity but not under heat stress.

Materials and methods

All methods were performed in accordance with the relevant guidelines and regulations.

Genomic DNA isolation and primer design

O. sativa miR396c pre-miRNA sequence (MI0001048) of was obtained from miRbase47 and used as query sequence to retrieve corresponding plant pre-miRNAs from Oryza species (non-redundant; whole genome shotgun; genome survey sequence) present in the NCBI database using BLAST. In order to isolate the corresponding pre-miRNA sequences from other Oryza species (not reported so far), primers were designed to amplify the corresponding pre-miRNA from genomic DNA samples. DNA was isolated using CTAB from young leaf tissues of twelve Oryza spp.48. Oryza species used for PCR amplification of pre-miRNA sequences are listed in Table 1 along with source data. All necessary permissions were taken to obtain the samples.

PCR amplification and sequencing of pre-miRNAs from Oryza species

PCR amplification was carried out in a reaction volume of 10 µl containing 50 ng genomic DNA template, 0.5 μM forward and 0.5 μM reverse primers (Table 2), Ampliqon Taq DNA Polymerase Master Mix RED (2x). Reaction conditions were as follows: initial denaturation (94 °C; 5 min), 35 cycles of denaturation (94 °C; 30 s), annealing (53 °C; 30 s), extension (72 °C; 30 s), with a final extension step at 72 °C (7 min) in a GeneAmp PCR System 9700 Thermal cycler (Applied Biosystems, California, USA). The amplified products were separated by agarose gel electrophoresis (1.5%), PCR products eluted using a gel purification kit (Favor Gen) and sequenced from both ends (Eurofins, India). The chromatogram data (forward and reverse sequences) obtained was analysed for base calling accuracy and a consensus sequence was determined for each PCR product using DNA Baser V 4.36. The sequences were aligned using Clustal Omega and analysed for variations49.

Table 2 Free energy of folding (ΔGfolding) for the secondary structure of pre-miR396c.
Full size table

Secondary structure prediction of pre-mir396c from Oryza species

The pre-miRNA sequences obtained through sequencing were analysed using Mfold22 using default parameters. The folding temperature was fixed at 37 °C. The predicted structures were analysed and grouped manually based on the number of bulges, stem and loops.

Analysis of variation in pre-miR396c sequence in Oryza species, dicots and monocots

Pre-miRNA sequences corresponding to miR396c from monocotyledonous and dicotyledonous species were retrieved from miRbase. Pre-miRNA sequences corresponding to the 3000 rice genomes were also retrieved50, 18 (https://snp-seek.irri.org/). ConSurf V 3.0 was used for predicting nucleotide conservation scores51, with the number with the highest value representing the most conserved base. Conservation in pre-miRNA sequences is represented using Circos23. The SNP data was downloaded from Rice SNP-Seek Database (https://snp-seek.irri.org/) as an Excel file and sorted based on subpopulations and the SNP. This data was manually fed into the ArcGIS software to generate map based data. Since the pre-miR396c is located on the reverse strand on rice chromosome 2 (coordinates), the data obtained from Rice SNP-Seek Database (a single SNP C/T at position 10,234,287,911) refers to the ‘non-coding’ strand (reference genome information IRGSP v 2.0) with reference to the pre-miR396c sequence. The corresponding sequence in the ‘coding strand of pre-miR396c would be G/A and is referred to as such throughout the study (map data however corresponds to the reference strand IRGSP v 2.0).

PCR amplification and sequencing of pre-mi96c from Oryza sativa landraces

Pre-miR396c sequences were PCR amplified from genomic DNA of O. sativa rice landraces (Table 3; includes source data) and sequenced from both ends (Eurofins, India). All necessary permissions were taken to obtain the samples. Chromatogram data (forward and reverse sequences) obtained was analysed for base calling accuracy and a consensus sequence was determined for each PCR product using DNA Baser V 4.36.

Table 3 List of rice landraces used for SNP validation and accession numbers of precursor miR396c sequences. Landraces used in this study are according to25.
Full size table

Mining pri-miR396c sequences from databases

O. sativa miR396c (Osa-miR396c) sequence was used to query the NCBI (nr and EST) databases using BLAST. Sequences were aligned using Clustal Omega and cDNA and genomic sequence exon–intron boundaries defined using Splign [https://www.ncbi.nlm.nih.gov/sutils/splign/splign.cgi]. Primers corresponding to O. sativa pri-miR396c, pre-miR396c (as defined by miRbase) were designed based on this sequence information and used for qRT-PCR.

Salinity and heat stress treatments

Four landraces with ‘A’ at the 103rd position in the expected PCR fragment (Supplementary Fig. 3 for reference; O. sativa ssp. indica cv. Pokkali, IR28, FL478) and four other landraces with ‘G’ nucleotide at the 103rd position (O. sativa ssp. indica cv. Nona Bokra, Anakodan, Orkyma and Aduisen) were selected randomly from the sequenced landraces. The plants were grown in controlled conditions (25 ± 2 °C; 12/12-h light/dark photoperiod; and 60% moisture) and transferred to the temperature and humidity controlled green house after 20 days. Salinity stress was imposed in hydroponics as described in52. Salinity stress was imposed using 100 mM NaCl (2 days) Similarly, the selected landraces with ‘A’ or ‘G’ at the 103rd position, grown in controlled conditions as described previously were also subjected to heat stress, incubated at 42 °C for 2 days. Parallel control samples were maintained at 25 ± 2 °C. Leaf tissues were harvested and frozen in liquid nitrogen.

Total RNA Isolation and RT-qPCR

MiRNA was isolated using ReliaPrep miRNA Cell and Tissue Miniprep Kit (Promega, India) using manufacturer’s instructions and quantified using Thermo Scientific, Multiskan GO v 3.2. First strand cDNA was synthesized from total miRNA (2.5 μg) using a cDNA synthesis kit (Superscript III; Invitrogen, USA) following manufacturer’s instructions and stem loop primers53. STEM-LOOP RT-qPCR was performed as described in54 and expression of Osa-Pri-miR396c, Osa-Pre-miR396c and mature Osa-miR396c were quantified in the same sample with specific primers (Table 4) as described in54,55 using SYBR green chemistry (Quant Studio 6 Flex, Applied Biosystems, USA) O. sativa U6 SnoRNA (Osa-U6) served as housekeeping control for O. sativa pri-miR396c, pre-miR396c and mature miR396c56. The relative expression ratio for pri, pre and mature miRNA396c was determined using the ∆CT and ∆∆CT methods for three biological replicates, each containing three seedlings (each replication was checked in triplicate) and significance estimated using Student’s t-test (GraphPad Prism 6). Candidate gene expression (OsGRF3, OsTBP) was tested for the same set of salinity stress samples for which STEM-LOOP RT-qPCR was carried out. Oryza eukaryotic initiation factor 4-α (eIF4-α) and actin (ACT) served as housekeeping controls54. cDNA was prepared using the cDNA synthesis kit (Superscript III; Invitrogen, USA) following manufacturer’s instructions and qRT-PCR carried out using SYBR green chemistry. Primers used for this study are listed in Table 4.

Table 4 Primers used for amplifying miR396c from Oryza species and qRT-PCR.
Full size table

Total RNA Isolation, small RNA Northern Hybridisation and Small RNA-seq data analysis

The leaf samples treated as mentioned above were ground in liquid nitrogen and total RNA was extracted using TRIzol (Invitrogen) as per the manufacturer’s instructions. RNA and quantified using RNA BR dye (Thermo Fisher Scientific) in a Qubit Fluorimeter. Northern hybridisation (small RNA) was performed according to57,58. Briefly, 10 µg of total RNA was electrophoresed in denaturing acrylamide gel [15% with 8 M Urea; 19:1 acrylamide: bis-acrylamide ratio; at 80 V (3 h)]. RNA was semi-dry blotted onto a Hybond N+ membrane (GE Healthcare) at 10 V overnight (4 °C) and crosslinked to the membrane using UV(UVP Crosslinker)]. Hybridisation in UltraHyb-Oligobuffer (Ambion) with 32P end-labelled oligo probes (miR396a: 5′ CAGTTCAAGAAAGCTGTGGAA 3′ or miR396c-5p: 5′ AAGTTCAAGAAAGCTGTGGAA 3′) was performed at 35 °C for 12 h. Probes were end labelled using T4 Polynucleotide Kinase (NEB) as per manufacturer’s instructions with labelled γ-32P-ATP (BARC,India) and purified using G-25 columns (Illustra Microspin; GE Healthcare) Following hybridization, the blot was washed twice with 2 × SSC, 0.5% SDS (30 min each at 35 °C).The blots were exposed to phosphor screen (GE Healthcare) and imaged using a Phosphorimager (Typhoon molecular imager; GE Healthcare). The blots were stripped at 80 °C and re-probed with end labelled U6 probe (equimolar mix of 5′ GGCCATGCTAATCTTCTCTGTATCGTT 3′ and 5′ GGCCATGCTAATCTTCTCTGTATCGTT 3′). For estimation of miR396a, miR396c expression levels, Reads per Million (RPM) values (two replicates each; three tissues; three O. sativa accessions and two wild Oryza species) from25 were used.