Barley Seeds miRNome Stability during Long-Term Storage and Aging

Seed aging is a complex biological process that has been attracting scientists’ attention for many years. High-throughput small RNA sequencing was applied to examine microRNAs contribution in barley seeds senescence. Unique samples of seeds that, despite having the same genetic makeup, differed in viability after over 45 years of storage in a dry state were investigated. In total, 61 known and 81 novel miRNA were identified in dry seeds. The highest level of expression was found in four conserved miRNA families, i.e., miR159, miR156, miR166, and miR168. However, the most astonishing result was the lack of significant differences in the level of almost all miRNAs in seed samples with significantly different viability. This result reveals that miRNAs in dry seeds are extremely stable. This is also the first identified RNA fraction that is not deteriorating along with the loss of seed viability. Moreover, the novel miRNA hvu-new41, with higher expression in seeds with the lowest viability as detected by RT-qPCR, has the potential to become an indicator of the decreasing viability of seeds during storage in a dry state.


Introduction
The aging process does not discriminate against any living organism. It begins at birth, leads to inevitable death, and is based on retrograde cellular changes occurring during the lifetime [1]. Life expectancy is a feature that is highly differentiated among living organisms. It is worth noting that it is significantly prolonged in plants in relation to animals, e.g., such plants as Pinus aristata Engelm. in White Mountain, California, which is about 5000 years old [2]. For most species, plants at the seed stage are able to survive extremely long, e.g., Nelumbo nucifera Gaertn. seeds dating back to around 1300 years old with a germination capacity of 84% [3]. Due to environmental degradation, intensive plant species conservation is required. Seeds that have been developed evolutionary for generative reproduction and as spore organs constitute the vast majority of germplasm stored in Gene Banks. They are preserved there in case of the species extinction in the natural environment. However, the fundamental problem is molecular changes associated with the natural aging process, reducing seed viability and vigor.
In the last decade, intensive scientific research has been conducted to identify changes in nucleic acids, proteins, and metabolites during seed dormancy. This omnibus approach has not yet led to important, innovative insights, but has enriched the understanding of the dormancy and germination process. In addition, the resulting data will be a valuable background for further studies. Despite the progress made in recent years, some important questions remain unanswered. The molecular characterization of seeds aging, dormancy,

Overview of Small RNA Library Sequencing
In order to explain the role of miRNA in the seeds aging process, nine miRNA libraries were constructed.
A total of 28,019,900 reads from the first run and 25,796,390 reads from the second run were obtained. After filtering the reads in terms of quality and length (<17 bp and >25 bp), there were obtained on average: 815,627 reads for renewed seeds (Rc-regenerated in 2017/2018 season), 480,785 for seeds with low viability (Lv, low viability; 2% of germability after storage 1972-2018), and 870,166 for highly viable seeds (Hv, high viability; 86.7% of germability after storage 1972-2018). Over 40% were mapped to the H. vulgare reference genome, of which about 10% were rRNA and tRNA (Table 1). In miRNAs libraries, the lengths of the miRNAs sequences were 18-22 nucleotides. The most frequent was 21 nucleotides fraction and it represented 52% of known miRNAs and 72% of novel miRNAs. The 18 nucleotides fraction was 22% of known miRNAs and 20 nucleotides was 15% of novel miRNAs. The smallest fraction of miRNAs was 22 nucleotides for both groups. Neither known nor novel miRNAs of length 23 nt and 24 nt were identified in the studied samples ( Figure 1).
The principal aim of the study is the profiling of miRNA in dry barley seed after long term storage and its role in determining and regulating seed viability.

Overview of Small RNA Library Sequencing
In order to explain the role of miRNA in the seeds aging process, nine miRNA libraries were constructed.
A total of 28,019,900 reads from the first run and 25,796,390 reads from the second run were obtained. After filtering the reads in terms of quality and length (<17 bp and >25 bp), there were obtained on average: 815,627 reads for renewed seeds (Rc-regenerated in 2017/2018 season), 480,785 for seeds with low viability (Lv, low viability; 2% of germability after storage 1972-2018), and 870,166 for highly viable seeds (Hv, high viability; 86.7% of germability after storage 1972-2018). Over 40% were mapped to the H. vulgare reference genome, of which about 10% were rRNA and tRNA (Table 1). In miRNAs libraries, the lengths of the miRNAs sequences were 18-22 nucleotides. The most frequent was 21 nucleotides fraction and it represented 52% of known miRNAs and 72% of novel miRNAs. The 18 nucleotides fraction was 22% of known miRNAs and 20 nucleotides was 15% of novel miRNAs. The smallest fraction of miRNAs was 22 nucleotides for both groups. Neither known nor novel miRNAs of length 23 nt and 24 nt were identified in the studied samples ( Figure 1).

Identification of Known and Novel miRNAs
Briefly, 34 known miRNAs occurred in all samples, two miRNAs unique to renewed seeds (Rc) sample, eight miRNAs unique to low viable seeds (Lv), and ten miRNAs unique to highly viable seeds (Hv) samples were identified (Supplementary Table S1). Four common miRNAs were identified for Rc samples and Lv samples, one miRNA for Hv and Lv samples, and two miRNAs for Rc and Hv samples ( Figure 2).
The most numerous group was the 168 miRNA family containing ten different isomiRs and the miR5048 and miR166 families containing eight and seven isomiRs, respectively. As isomiRs, variants of mature miRNAs differing from their reference sequences in the miRBase [11] were considered. The variation results from 5 -or 3 -trimming variants arising during miRNA biogenesis or during pri-miRNA post-trancriptional editing [18,19]. The smallest family was miR397 containing only one member ( Figure 3).
Canonical mature miRNAs represented slightly over 50% of all identified known miRNAs. The remaining fraction consisted mainly of non-coincidental 3 trimmed isomiRs (average 47%), 5 trimmed isomiRs (about 0.2%) and two-side trimmed (about 0.5%). 0.1% of known miRNAs with very low frequencies were considered to be random isomiRs. For three families, i.e., miR1120, miR168, and miR6200, the canonical forms with a length of 24 nt (miR1120 and miR168) and 22 nt (miR6200) were not identified in the examined samples. The occurring canonical miRNAs were mostly 21 nt in length. Only in the case of the miR6201 family were there molecules with a length of 22 nt.

Identification of Known and Novel miRNAs
Briefly, 34 known miRNAs occurred in all samples, two miRNAs unique to renewed seeds (Rc) sample, eight miRNAs unique to low viable seeds (Lv), and ten miRNAs unique to highly viable seeds (Hv) samples were identified (Supplementary Table S1). Four common miRNAs were identified for Rc samples and Lv samples, one miRNA for Hv and Lv samples, and two miRNAs for Rc and Hv samples ( Figure 2).
The most numerous group was the 168 miRNA family containing ten different iso-miRs and the miR5048 and miR166 families containing eight and seven isomiRs, respectively. As isomiRs, variants of mature miRNAs differing from their reference sequences in the miRBase [11] were considered. The variation results from 5′-or 3′-trimming variants arising during miRNA biogenesis or during pri-miRNA post-trancriptional editing [18,19]. The smallest family was miR397 containing only one member ( Figure 3).
Canonical mature miRNAs represented slightly over 50% of all identified known miRNAs. The remaining fraction consisted mainly of non-coincidental 3'trimmed isomiRs (average 47%), 5'trimmed isomiRs (about 0.2%) and two-side trimmed (about 0.5%). 0.1% of known miRNAs with very low frequencies were considered to be random isomiRs. For three families, i.e., miR1120, miR168, and miR6200, the canonical forms with a length of 24 nt (miR1120 and miR168) and 22 nt (miR6200) were not identified in the examined samples. The occurring canonical miRNAs were mostly 21 nt in length. Only in the case of the miR6201 family were there molecules with a length of 22 nt.  Table S1).   Lv-low viable seeds after long-term storage in a dry state; Hv-highly viable seeds seeds after long-term storage in a dry state.
In addition, 81 novel miRNAs were identified, of which 48 miRNA were found in all the samples. Two miRNAs unique to Hv sample and nine miRNAs unique to Rc sample were identified. Moreover, ten novel miRNAs common to Rc and Hv samples, seven to Rc and Lv samples, and five to Lv and Hv samples were found ( Figure 5, Supplementary  Table S2). Overall, the level of expression of the novel miRNAs was much lower than the known ones. The highest expression had hvu-new80. As many as 70 novel miRNAs had an expression level below 100 reads per million (RPM) (Supplementary Table S2).   Lv-low viable seeds after long-term storage in a dry state; Hv-highly viable seeds seeds after long-term storage in a dry state.
In addition, 81 novel miRNAs were identified, of which 48 miRNA were found in all the samples. Two miRNAs unique to Hv sample and nine miRNAs unique to Rc sample were identified. Moreover, ten novel miRNAs common to Rc and Hv samples, seven to Rc and Lv samples, and five to Lv and Hv samples were found ( Figure 5, Supplementary  Table S2). Overall, the level of expression of the novel miRNAs was much lower than the known ones. The highest expression had hvu-new80. As many as 70 novel miRNAs had an expression level below 100 reads per million (RPM) (Supplementary Table S2). Lv-low viable seeds after long-term storage in a dry state; Hv-highly viable seeds seeds after long-term storage in a dry state.
In addition, 81 novel miRNAs were identified, of which 48 miRNA were found in all the samples. Two miRNAs unique to Hv sample and nine miRNAs unique to Rc sample were identified. Moreover, ten novel miRNAs common to Rc and Hv samples, seven to Rc and Lv samples, and five to Lv and Hv samples were found ( Figure 5, Supplementary  Table S2). Overall, the level of expression of the novel miRNAs was much lower than the known ones. The highest expression had hvu-new80. As many as 70 novel miRNAs had an expression level below 100 reads per million (RPM) (Supplementary Table S2).
Among the novel miRNAs, eighteen miRNAs were found with a mature sequence matching in the miRBase [11]. Some novel miRNAs are isomiRs in already known families: thirteen novel miRNAs were assigned to known families, one novel miRNA per miR159, miR166, miR168, miR156, miR396, miR171, and miR397, and five to miR167 family ( Table 2). The secondary structures of miRNA hairpin precursors were predicted according to the criteria reported by Axtell and Meyers [20] (Supplementary Table S3; Supplementary Figure S1). Among the novel miRNAs, eighteen miRNAs were found with a mature sequence matching in the miRBase [11]. Some novel miRNAs are isomiRs in already known families: thirteen novel miRNAs were assigned to known families, one novel miRNA per miR159, miR166, miR168, miR156, miR396, miR171, and miR397, and five to miR167 family ( Table 2). The secondary structures of miRNA hairpin precursors were predicted according to the criteria reported by Axtell and Meyers [20] (Supplementary Table S3; Supplementary Figure S1).

Figure 5.
Venn diagram for number of novel miRNAs detected in barley dry seeds. Rc-renewed seeds sample; Lv-low viable seeds after long-term storage in a dry state; Hv-highly viable seeds after long-term storage in a dry state.

Differential Expression of miRNAs
The differential expression (DE) analysis was performed to compare the level of miRNA in barley seeds with different viability and storage time. It based on the normalized read counts for each identified miRNA. Samples after long-term storage having different viability (Hv and Lv) did not show significant differences in the miRNA expression level. Only two miRNAs significantly differentiated the regenerated seed (Rc) sample from the long-term stored samples (Hv and Lv). Both of them were novel miRNAs. Hvu-new60 was present only in Rc sample and hvu-new2 was present only in samples after long-term storage (Supplementary Table S4).
The expression levels of some novel and known miRNAs were verified using RT-qPCR methods ( Figure 6). A similar expression profile was observed for the majority of miRNAs in the tested sample. The hvu-new48 failed to amplify. However, this novel miRNA was found in a very small number in NGS sequencing results. As in DE analysis, the results of RT-qPCR did not indicate the presence of significant differences in the level Figure 5. Venn diagram for number of novel miRNAs detected in barley dry seeds. Rc-renewed seeds sample; Lv-low viable seeds after long-term storage in a dry state; Hv-highly viable seeds after long-term storage in a dry state.

Differential Expression of miRNAs
The differential expression (DE) analysis was performed to compare the level of miRNA in barley seeds with different viability and storage time. It based on the normalized read counts for each identified miRNA. Samples after long-term storage having different viability (Hv and Lv) did not show significant differences in the miRNA expression level. Only two miRNAs significantly differentiated the regenerated seed (Rc) sample from the long-term stored samples (Hv and Lv). Both of them were novel miRNAs. Hvu-new60 was present only in Rc sample and hvu-new2 was present only in samples after long-term storage (Supplementary Table S4).
The expression levels of some novel and known miRNAs were verified using RT-qPCR methods ( Figure 6). A similar expression profile was observed for the majority of miRNAs in the tested sample. The hvu-new48 failed to amplify. However, this novel miRNA was found in a very small number in NGS sequencing results. As in DE analysis, the results of RT-qPCR did not indicate the presence of significant differences in the level of tested miRNAs in seed samples with different viability and storage duration. However, ANOVA analysis of RT-qPCR results showed significant differences in hvu-new41 expression levels. The Tukey's honest significant difference test showed that the level of the miRNA is significantly higher in Lv sample than in Rc, while Hv sample had an intermediate level. of tested miRNAs in seed samples with different viability and storage duration. However, ANOVA analysis of RT-qPCR results showed significant differences in hvu-new41 expression levels. The Tukey's honest significant difference test showed that the level of the miRNA is significantly higher in Lv sample than in Rc, while Hv sample had an intermediate level. Figure 6. Expression level of selected miRNA by RT-qPCR in barley dry seeds. Rc-renewed seeds sample; Lv-low viable seeds after long-term storage in a dry state; Hv-highly viable seeds after long-term storage in a dry state. Letters indicate homogeneous groups based on the Tukey's HSD test.

Prediction Targets of miRNA
The analysis of the targets functions was performed for all miRNAs. Target predictions indicated more than one target for each miRNA. In total, 11,771 potential targets were identified for the known and novel miRNAs, of which only 856 originated from the barley genome (Supplementary Table S5). The most suitable targets identified in barley genome were TC240039 (UniRef A4HRC2) encoding NAC transcription factor (expectation 1) for hvu-new63 and TC253359 (UniRef Q40073) encoding Ribulose bisphosphate carboxylase/oxygenase activase A for miR1120 (expectation 3). The novel miRNA hvu-new60 had an affinity for thirteen potential targets with five encoding proteins, while hvu-new2 had eleven potential targets and encoding proteins. Four proteins were encoded by seven identified targets for miRNA hvu-new41 (Table 3). Approximately 15% of the potential targets in the barley genome for both new and known miRNAs had the translational inhibition suggested as a possible process of miRNA-RISC mediated gene inhibition. The translational inhibition potential was reported by psRNATarget [21] whenever a mismatch was found in the central complementary region of the miRNA sequence.
Furthermore, all the potential targets of miRNAs were subjected to Gene Ontology (GO) analysis to determine their GO terms, which could provide insight into the miRNAs function. The GO analysis showed that the highest number of putative targets (over 75%) was involved in cellular and metabolic processes. Potential targets for novel miRNAs were involved in 114 biological processes and those for known miRNAs only in 38. Interestingly, 57% of potential targets for novel miRNAs were involved in primary metabolic processes. No targets for known miRNAs were assigned to this category (Figure 7a, Supplementary Table S6). The most common group among identified potential targets were those related to the molecular functions involved in binding and catalytic activity (over Figure 6. Expression level of selected miRNA by RT-qPCR in barley dry seeds. Rc-renewed seeds sample; Lv-low viable seeds after long-term storage in a dry state; Hv-highly viable seeds after long-term storage in a dry state. Letters indicate homogeneous groups based on the Tukey's HSD test.

Prediction Targets of miRNA
The analysis of the targets functions was performed for all miRNAs. Target predictions indicated more than one target for each miRNA. In total, 11,771 potential targets were identified for the known and novel miRNAs, of which only 856 originated from the barley genome (Supplementary Table S5). The most suitable targets identified in barley genome were TC240039 (UniRef A4HRC2) encoding NAC transcription factor (expectation 1) for hvu-new63 and TC253359 (UniRef Q40073) encoding Ribulose bisphosphate carboxylase/oxygenase activase A for miR1120 (expectation 3). The novel miRNA hvu-new60 had an affinity for thirteen potential targets with five encoding proteins, while hvu-new2 had eleven potential targets and encoding proteins. Four proteins were encoded by seven identified targets for miRNA hvu-new41 (Table 3). Approximately 15% of the potential targets in the barley genome for both new and known miRNAs had the translational inhibition suggested as a possible process of miRNA-RISC mediated gene inhibition. The translational inhibition potential was reported by psRNATarget [21] whenever a mismatch was found in the central complementary region of the miRNA sequence.
Furthermore, all the potential targets of miRNAs were subjected to Gene Ontology (GO) analysis to determine their GO terms, which could provide insight into the miRNAs function. The GO analysis showed that the highest number of putative targets (over 75%) was involved in cellular and metabolic processes. Potential targets for novel miRNAs were involved in 114 biological processes and those for known miRNAs only in 38. Interestingly, 57% of potential targets for novel miRNAs were involved in primary metabolic processes. No targets for known miRNAs were assigned to this category (Figure 7a, Supplementary Table S6). The most common group among identified potential targets were those related to the molecular functions involved in binding and catalytic activity (over 55%). Similarly to biological processes, the putative targets for the novel miRNAs represented more groups of molecular functions, although the difference was not so significant (46 for the novel miRNAs and 32 for known miRNAs). Uniquely 21% of the potential targets for the novel miRNAs had a molecular function related to hydrolase activity. In contrast, 26% of the selected targets for known miRNAs performed a small molecule binding function. None of the targets identified for novel miRNAs had such a function (Figure 7b, Supplementary  Table S6). In turn, the most common cellular anatomical entities associated with potential targets for both known and novel miRNAs were intracellular, cytoplasm, and organelle (over 70%, 55%, and 50%, respectively). Among the unique potential targets for the novel miRNAs were those associated with chloroplast internal structures, i.e., thylakoid and its membranes and photosystem (Figure 7c, Supplementary Table S6). 3 Hordoindoline-b1 hinb-1 Q5IUH9 Figure 7. Analysis of target function using g:Profiler toolset [22] for known and novel miRNA detected in barley seeds, (a) biological process; (b) molecular function; (c) cellular anatomical entity.

Discussion
Small non-coding RNAs, including miRNAs, play a key role in plant ontogenesis [23]. More and more research on the role of miRNAs in seed development and germination is being conducted [5,[24][25][26][27]. However, miRNome in dry seeds with quiescent metabolism has been relatively rarely analyzed [25,28,29]. So far, the effect of long-term seed storage and the inextricably linked ageing on miRNA in barley seeds has not been studied.
Here, the unique plant material was used. Two samples of barley seeds exposed to long-term storage in a dry state were used. Their uniqueness results from the origin of a single seed lot. This guarantees no differences resulting from environmental conditions during ripening, post-harvest treatment and preparation for storage. Therefore, it can be assumed that, at the beginning of storage, they had not only homogeneous genetic background but also homogeneous transcriptome and miRNome. During long-term storage, the moisture content of the seeds in one flask increased probably due to its slight leakage. This resulted in a decrease in seed viability. The significant difference in the viability of the two seed samples made them a perfect material for studying the effects of aging at a molecular level. The studies presented here were aimed to determine the relationship between the aging process of seeds and miRNome.
As a result of high-throughput sequencing in the presented study, 61 known miRNAs were identified that were matched to 13 families available for barley at miRBase ver. 22.1 [11]. Also, 81 novel miRNAs, of which 13 were matched as new isomiRs for barley miRNAs, were found.

miRNA and Seed Aging
Seed aging is associated with changes in macromolecules such as proteins, lipids and nucleic acid [30][31][32]. These changes are most likely caused by reactive oxygen species (ROS) [33][34][35][36]. In dry seeds a metabolism is quiescent because of the low water content, the viscosity increases, and the glassy state forms and the chemical reactions are reduced [37,38]. Aging process progressively occurs due to ROS origin damage. ROS are formed as a result of non-enzymatic reactions such as lipid peroxidation or Maillard reactions [39][40][41]. Compared to DNA, which is double stranded, bound to chromatin and located in the nucleus, RNA undergoes damage easily. mRNA is susceptible to breakage and nucleotides can be removed or modified [42]. Dry, mature seeds contain various long-lived mRNA, which are present in seeds from late embryogenesis to early seed germination and are crucial for protein synthesis during the early phase of germination [43,44]. The relationship between RNA fragmentation and reduced viability has been demonstrated in soybean seeds and the damages occur randomly throughout the transcriptome [45,46]. Random RNA threads breakage occurred also during the aging of Arabidopsis seeds [47]. Dry seeds also contain long-lived miRNAs, which act as a regulator of gene expression during germination [28,48,49]. Therefore, it is crucial to answer the question to what extent longterm seed storage and the inextricably linked ageing is reflected in the miRNome. The material used in the study provided a unique opportunity to find answers to the questions about the role and relationship of miRNA with the process of seeds aging preserved in a dry state. As numerous studies have shown a pivotal role of miRNAs in senescence in both animals [50] and plants [51], the question arises about the role of miRNAs in seed aging. Recent studies on artificial seed aging have identified a relationship between miRNAs and the loss of seed viability [52,53]. Artificial aging, also known as accelerated aging, is an extensively used method in seed aging studies. The deterioration of the seeds quality is accelerated by affecting them with high temperature and humidity for a relatively short time [54,55]. Despite many years of investigations, researchers' opinions on the uniformity of changes caused by accelerated and natural aging are divided. However, when analyzing the role of miRNAs in accelerated and dry aging, we are talking about two definitely different processes. The quiescent metabolism in dry seeds prevents the degradation of miRNA by exoribonucleases [56].
Taking into account previous studies indicating the presence of changes at the level of macromolecules caused by the process of seed aging, the presented here constant level of known and novel miRNAs in barley seeds regardless of their viability or storage time is quite surprising. Due to the lack of metabolic activity in dry seeds, it indicates very high stability of this non-coding RNA fraction. The uneven degradation of different RNA classes in barley seeds with low viability was indicated by earlier studies of Puchta et al. [57]. Although that study did not include the small non-coding RNA class, its results were consistent with uneven RNA degradation in male and female mouse hearts [58]. The high stability of miRNAs in vitro was also reported in other studies. Researchers indicated that miRNA molecules with increased GC-ratio or lacking AU/UA in the seed sequences and tail region had higher stability [59,60]. Extraordinarily interesting is the higher in vitro stability of miRNA compared to mRNA during NaOH, RNase A, Exonuclease T and Exonuclease T7 or Benzonase treatment [58]. Moreover, miRNAs seemed to be also insensitive to heat degradation [61]. These results suggest the stability of miRNAs to be generally robust. Therefore, it can be assumed that the lack of significant differences in miRNAs level of observed in the study results from their high stability. Moreover, it may also indicate that miRNA molecules are not destroyed by ROS, as happens with other macromolecules.
In cells, miRNA are closely associated with RNA-induced silencing complex (RISC) and only a relatively small fraction is considered to be independent of RISC [62]. The association of miRNAs with the components of the RISC complex increases their stability [63]. The presence of RISC complex was confirmed by proteomic studies of barley seeds [64]. Besides, EMBL-EBI Expression Atlas indicates the highest expression of RISC is related to the development and germination of barley seeds [65]. Moreover, in plants, the stability of miRNA is additionally modified by 2 -O-methylation at 3 end catalyzed by methyltransferase HEN1 [66]. Thus, in dry seeds, the miRNA-protein complex can provide an additional form of protection for this RNA fraction and prevents degradation during long-term storage.
The above-mentioned properties make miRNAs are the sole RNA faction identified so far that is not suffering from the deterioration associated with seeds aging. This discovery may have unprecedented significance for biobanks and indicate the direction of further research into the storage of human tissues and fluids. Perhaps, part of the material should be stored at low temperatures, but after prior freeze-drying.

miRNome of Barley Seeds
Analysis of embryos derived from dry seeds with quiescent metabolic activity revealed the presence of numerous isomiRs. Initially, the isomiRs observed in high-throughput sequencing results were thought to be either sequencing artefacts or to arise as a result of imprecise excision of miRNAs from the pre-miRNA hairpin by Dicer during biogenesis [67][68][69]. It is currently believed that isomiRs are biologically functional and have specific roles in plant cells [70].
The obtained results showed the presence of isomiRs differing from the canonical form only in length. They are most likely formed from a single pre-miRNA and are only length variants of the canonical form [71]. However, the data currently available in miRBase for barley are too limited to unequivocally state that they are not shared by several members of particular miRNA families.
All identified isomiRs were shorter than the canonical form. Their formation may therefore be related to partial degradation of miRNA molecules by exonucleases or to variable processing by Dicer [19,72,73]. The absence of 24 nt miRNAs may indicate a lack of Dicer-like protein DCL3 and DCL5 activity. DCL3 is involved in the biosynthesis of 24 nt long siRNAs derived mainly from transposons and repetitive DNA fragments and is involved in transcriptional gene silencing (TGS) [74][75][76]. In contrast, DCL5 reported in monocots is expressed in developing inflorescences and is responsible for the formation of 24 nt reproductive phasiRNAs [77,78]. Among the observed isomiRs, the 3 trimmed ones predominated, which is most likely due to nucleotide trimming at the 3 end of the AGRONAUTE (AGO)-associated miRNA [79]. As shown by previous crystallographic studies of AGO, the 5 end of the miRNA is located within the MID domain, whereas the 3 end overhangs the PAZ domain and is thus more vulnerable to 3 -5 exonuclease activity [80,81]. Trimming the 3 end decreases the length of the miRNA molecule and increases its stability in the AGO protein [82,83]. The length of the miRNAs sequence is also an important factor determining sorting them into particular AGO complexes. Most miRNAs with a length of 21 nt bind to AGO1 and AGO2 while 24 nt miRNAs bind to AGO4 [84]. Therefore, it is suggested that the isomiRs length variation may be related to the strategy of target differentiation by assigning isomiRs to different AGO complexes [85].
Higher levels of 20 nt and 18 nt isomiRs of miR156 than its canonical form were also observed in the studied material. This indicates that isomiRs may regulate key processes at this developmental stage. Notably, previous studies have shown differential expression of miR156 isomiRs induced by drought stress in barley. The isomiRs regulated the expression of several genes including dehydration-responsive element-binding proteins [86]. Other studies confirm that indeed differential expression of isomiRs occurs tissue-specific and is associated with specific regulatory functions [87][88][89].
In the study, the most numerous group was 21 nt miRNAs, both for known and novel miRNAs. Many canonical 21 nt miRNAs have a high level of expression in plants [90]. The study of Ding et al. [91] showed 21 nt miRNAs constituted the major fraction. Generally, in plants, the common length distribution of miRNA sequences were between 21 and 24 nt, which was related to miRNA repetition [92]. Based on the differences in miRNA expression, it was found that there were characteristic expression profiles for dry and imbibed seeds, which is consistent with the results obtained by Ding et al. [91]. Previous studies concerning soybean shown that 22 nt small RNA class was highly abundant and was associated with repetitive sequences like rDNA and transposons. This class also played other functions than 21-and 24-nt small RNAs [92]. Such results were obtained for wild barley [14], wheat [9], corn [93], and soybean [92]. The length distribution in the Deng et al. [14] study on wild barley varied between 20-24 nt and the most abundant group was 21 nt (77.78%).
Six of the identified miRNA families, i.e., miR156, miR159, miR166, miR168, miR171, and miR397 occur both in mono-and eudicots, which may indicate that miRNAs are involved in regulating key genes for the development and functioning of plants [5]. Furthermore, the miR156, miR159 and miR166 take part in regulating of transcription of genes encoding activators and germination and dormancy repressors [5,13,94,95]. The families miR444, miR5049, and miR6201 were identified only in monocots, which suggests that they evolved as early as, at least, 90 Mya [96,97]. The families miR5051 and miR6200 were identified only in H. vulgare, which confirms that they are species specific. According to Wang et al. [98], most of the identified miRNAs in barley were highly conserved in monocotyledonous species, especially in the Poaceae family. Some miRNAs were present in mono-and eudicots plants, while others were specific to barley. Differences in miRNA occurrence may be affected by gains and losses due to evolution [13]. Data analysis has shown that some identified miRNAs had been sequenced only a few times, while others had been sequenced thousands of times. Among the identified conserved miRNA families, miR159, miR156, miR166, and miR168 showed the highest expression in all studied groups. Deng et al. [14] explained miRNAs high expressions by their critical role in growth and development of wild barley. The miR159 family, which showed the highest expression in the study regardless of the sample, regulates GAMYB transcription factor, which interacts in response to giberelic acid (GA) [99]. The miR156 was identified as one of the most expressed families in H. vulgare subsp. spontaneum (K. Koch) Thell. [14]. It regulates expression of a transcription factors family SPL (Squamosa promoter binding proteins like) genes [100,101]. The miR166 family that had the highest abundance in the early phase of maize seeds germination targeted DNA-binding transcription factor homeobox-leucine zipper in wild barley [5,14]. The miR168 family, that targets AGO1 in A. thaliana and regulates RISC was also one of the most frequently expressed in the wild barley [14,102]. In general, the novel miRNAs identified here, were at low or even very low level. According to Wang et al. [5], differences in the number of readings associated with the level of expression are caused by the activity of various physiological and biochemical processes in the seeds at the moment. However, in dry seeds where metabolism is suppressed, the differences in the level of individual miRNAs will indicate which metabolic pathways will be triggered in the very initial germination phase. They may also be related to processes associated with the last phase of seed ripening and dormancy.

miRNA in Regulating Gene Expression Based on in Silico Analysis
Contrary to animals, in plants, almost perfect complementarity of miRNAs to their target transcripts is required. The target sites are located within the open reading frames (ORFs) of target gene [103]. Therefore, an efficient and accurate prediction method is based on a basic matching of plant miRNA and mRNA sequences [104]. The in-silico analysis performed here identified a total of 817 putative targets for miRNA found in dry seeds.
According to GO results, it is clear that targets' biological functions are largely related to the metabolism of macromolecules. Both known and novel miRNAs regulate the catalytic activity associated with hydrolysis. It is well documented that in barley seeds both the scutellum and the aleurone layer are involved in the production and secretion of hydrolytic enzymes that catalyze the degradation of starch and protein reserves of the endosperm during germination [105]. In the initial germination phase of barley seed, amylolytic activity was detected almost exclusively in the region of epithelial cells of the scutellum [106]. Therefore, miRNAs derived from embryonic part of barley seeds may be responsible for regulating hydrolytic processes during initial germination phase. Among others, miR159, miR168 and miR171 can be involved in regulating starch hydrolysis. However, cells walls of endosperm constitute a potential barrier that limits access of hydrolytic enzymes to their substrates within endosperm cells. Thus, cell wall degradation is an important event in the early germination phase. Barley endosperm cells walls consist of about 70% (w/w) of (l-3,1-4)-p-glucans [107]. The (1-3,1-4)-beta-D-glucanase (EC 3.2.1.73) secreted during barley germination is responsible for their depolymerization [108]. The hvu-new41, that had significantly higher expression in seeds with the lowest viability might be involved in regulation of 1-3,1-4-beta-D-glucanase synthesis. During barley germination glucanase is synthesized de novo, mRNA encoding these enzymes increase during germination [109]. Thus, the increased content of hvu-new41 might be responsible for the removal of (1-3,1-4)-beta-D-glucanase transcripts and lowering or completely blocking the synthesis of this key enzyme for seed germination. However, experimental target verification for this novel miRNA is needed. It should also be stressed that in the case of dry seeds, in which metabolism is quiescent, the identified miRNAs do not regulate gene expression exactly at the moment of performing the analysis. Their role is to regulate genes which expression will be initiated only at the moment of the environmental conditions favorable for germination occurrence, i.e., at the moment of seed rehydration. Thus, miRNAs accumulated in dry seeds represent a guarantee of proper regulation of gene expression at the next developmental stage. The phenomenon is that it may take place even over several decades.

Plant Material
Barley seeds of cv. 'Damazy' harvested in 1972, were dried to 2.96% water content and seed germination tests were carried out. Seeds which showed viability above 95% were placed in two air-filled glass flasks and hermetically sealed. All flasks were stored under the same conditions, i.e., in a basement with limited light, at an ambient temperature (10-25 • C) and a relative humidity in the room of about 80-90%. Germination tests were carried out again in 2015 and it turned out that seed viability differed between flasks. A sample of the high viable seeds was taken, and the remaining seeds were then stored further. The high viable sample was regenerated in the field (in 2017/2018) according to the gene bank standards and provided the reference material after drying them to a low water content (approx. 7%). In 2018, seed viability and moisture content (Figure 8) were evaluated and three groups were selected for the study: seeds stock with high viability (Hv) 86.7% and moisture content 3.58%, seeds with low viability (Lv) 2% and moisture content 12.5% and regenerated seeds (Rc), which were a reference sample. Three biological replicates of 25 dry seeds each were used at every stage of the study.
water content (approx. 7%). In 2018, seed viability and moisture content (Figure 8) were evaluated and three groups were selected for the study: seeds stock with high viability (Hv) 86.7% and moisture content 3.58%, seeds with low viability (Lv) 2% and moisture content 12.5% and regenerated seeds (Rc), which were a reference sample. Three biological replicates of 25 dry seeds each were used at every stage of the study.

miRNA Extraction and Library Preparation
The initial step in miRNA isolation was the dissection of the part containing the embryo and the scutellum from dry seeds. This procedure was performed to reduce the amount of starch, which negatively affects the effectiveness of miRNA isolation. Parts of the seeds prepared in this way were crushed into fine powder in a mortar under liquid nitrogen. miRNA was isolated using a Micro RNA kit (A&A Biotechnology, Gdynia, Poland). The miRNA was quantitatively evaluated using Nanodrop 1000 spectrophotometer

Bioinformatic and Statistical Analysis
The quality of the readings from the Illumina sequencing was evaluated using FastQC [110]. The raw data was filtered: adapters, low-quality sequences (Q < 30), and sequences shorter than 17 bp and longer than 25 bp were removed. The UEA Small RNA Workbench: Adapter Removal and Filter software was used to remove the adaptor sequences, sequences containing ambiguous nucleotides, and t/r RNA contamination [111]. High-quality reads were forwarded for further analysis. Using the Sequence Alignment UEA Small RNA Workbench [112], the miRNA sequences were mapped to the Hordeum vulgare reference genome (Hv_IBSC_PGSB_v2) obtained from Ensembl Plants database release 40 [113] accessed 16.03.2020. To find known miRNAs deposited in the miRBase version 22.1 database [11] and their isoforms, analyses were performed using the UEA Small RNA Workbench miRProf with no mismatch allowed. Identification of novel miR-NAs was performed in the miRCat UEA Small RNA Workbench tool. The software was used with plant parameters (genome hits = 16, hit dist = 200, max gaps = 3, max overlap

miRNA Extraction and Library Preparation
The initial step in miRNA isolation was the dissection of the part containing the embryo and the scutellum from dry seeds. This procedure was performed to reduce the amount of starch, which negatively affects the effectiveness of miRNA isolation. Parts of the seeds prepared in this way were crushed into fine powder in a mortar under liquid nitrogen. miRNA was isolated using a Micro RNA kit (A&A Biotechnology, Gdynia, Poland). The miRNA was quantitatively evaluated using Nanodrop 1000 spectrophotometer (Thermo

Bioinformatic and Statistical Analysis
The quality of the readings from the Illumina sequencing was evaluated using FastQC [110]. The raw data was filtered: adapters, low-quality sequences (Q < 30), and sequences shorter than 17 bp and longer than 25 bp were removed. The UEA Small RNA Workbench: Adapter Removal and Filter software was used to remove the adaptor sequences, sequences containing ambiguous nucleotides, and t/r RNA contamination [111]. High-quality reads were forwarded for further analysis. Using the Sequence Alignment UEA Small RNA Workbench [112], the miRNA sequences were mapped to the Hordeum vulgare reference genome (Hv_IBSC_PGSB_v2) obtained from Ensembl Plants database release 40 [113] accessed 16.03.2020. To find known miRNAs deposited in the miRBase version 22.1 database [11] and their isoforms, analyses were performed using the UEA Small RNA Workbench miRProf with no mismatch allowed. Identification of novel miRNAs was performed in the miRCat UEA Small RNA Workbench tool. The software was used with plant parameters (genome hits = 16, hit dist = 200, max gaps = 3, max overlap percentage = 80, max percent unpaired = 50, max unique hits = 3, maxsize = 25, min abundance = 1, min energy = −25, min gc = 20, min hairpin length = 60, min paired = 17, min size = 18, orientation percentage = 80, hairpin extension = 100, p-value = 0.05).
The total number of mapped reads from given library was used for normalization to reads per million (RPM) of novel miRNA read abundance. The sequence was considered as a miRNA candidate, when its secondary structure met the criteria reported by Axtell and Meyers [20]. Novel miRNAs were assigned to the known miRNA families in miRBase 22.1 using miRclassifity [114]. Prediction of target genes for the found miRNAs was imple-mented in psRNATarget software (https://plantgrn.noble.org/psRNATarget/, accessed on 12 February 2021) [21] with default Schema V2 (2017 release) with an expectation score up to 5 and the length complementarity scoring was 17. Quantitative analyses were carried out using DESeq2 from SARTools R package [115]. The results were presented in a graphical form using ggplot2 [116] package for R. ANOVA analysis (p < 0.05) with post hoc Tukey's HSD test by XLSTAT ECOLOGY 2017.1.1 (Addinsoft) were used to determine the relevant differences between the samples and the miRNA in the family. In order to define potential roles of H. vulgare miRNA in the biological and molecular process, Gene Ontology (GO) annotations of miRNA target genes were downloaded from UniProt identifiers GO categories represented by predicted targets of miRNA were demonstrated using g:Profiler toolset [22]. Raw data is available in the Open Science Framework repository under identifier HDG78 and NCBI GEO record GSE164512.

RT-qPCR
A RT-qPCR (Reverse transcription qPCR) reaction was performed to confirm the results obtained during deep sequencing for statistically significant miRNAs [117]. miRNAs occurring in each of the samples and showing a stable level of expression in the samples were chosen as the reference miRNA using RefFinder [118]. Primers for sequences of mature miRNA were designed using miRprimer software [119] and the primer sequences are presented in Supplementary Table S7. miRprimer is an automatic method for design of functional primers for miR-specific RT-qPCR. The cDNA synthesis reaction with miRNA was carried out using the Mir-X™ miRNA First-Strand Synthesis Kit (Takara Bio, Kusatsu, Japan) and 400 ng miRNA, the reactions were conducted according to the manufacturer's protocol. Real Time PCR reactions were performed with the following composition: 2 µL 1× HOT FIREPol EvaGreen qPCR Mix Plus (Solis BioDyne Tartu, Estonia), 0.25 µM primers, 5.5 µL water and 2 µL 20-fold diluted cDNA. The reaction was carried out under the following conditions: initial denaturation: 60 • C/15 min, 40 cycles: denaturation 95 • C/25 s, hybridization of primers 60 • C/25 s, elongation 72 • C/25 s and final elongation 72 • C/5 min, the reaction was completed in 4 • C. The analysis was performed in Rotor-Gene 6000q series (Corbett Life Science, Mortlake, Australia) thermocycler. All analyses were performed in three biological and three technical replicates. The expression of the investigated miRNAs was calculated based on the ∆∆ Ct analysis. Three miRNAs, namely miR159, miR166, and miR168, were used as references.

Conclusions
The results presented here clearly indicate that miRNAs are not destroyed during the long storage of barley seeds in a dry state. The level of the majority of known and novel miRNAs remained stable regardless of whether the seeds were fully viable or had very low germination capacity, as well as whether they were long-term stored or nonstored. Therefore, it can be concluded that this is the only RNA class which does not suffer deterioration due to the aging of the seeds. The novel miRNA (hvu-new41) was the only one that had a significantly higher expression in seeds with low germination ability. Further research is needed to verify whether it actually regulates 1-3,1-4-beta-D-glucanase synthesis. The analysis of the degradome and detailed analysis of the transcriptome of the same plant material are ongoing and the results will be published soon.
Supplementary Materials: Supplementary Materials can be found at https://www.mdpi.com/ article/10.3390/ijms22094315/s1. Table S1. An average level of expression of known miRNAs identified in the tested seed samples; Table S2. An average level of expression of novel miRNAs identified in the tested seed samples; Table S3. Hairpin structures of novel miRNA; Table S4. Results of Differential Expression (DE) analysis; Table S5. Targets for known and novel miRNAs; Table S6. Results of GO analysis; Table S7. Primers used in the RT-qPCR of miRNA; Figure S1. Novel miRNA hairpins structure.