MicroRNAs, the epigenetic memory and climatic adaptation in Norway spruce


Author for correspondence:
Igor A. Yakovlev
Tel: +47 48 146583
Email: igor.yakovlev@skogoglandskap.no


  • Norway spruce expresses a temperature-dependent epigenetic memory from the time of embryo development, which thereafter influences the timing bud phenology. MicroRNAs (miRNAs)are endogenous small RNAs, exerting epigenetic gene regulatory impacts. We have tested for their presence and differential expression.
  • We prepared concatemerized small RNA libraries from seedlings of two full-sib families, originated from seeds developed in a cold and warm environment. One family expressed distinct epigenetic effects while the other not. We used available plant miRNA query sequences to search for conserved miRNAs and from the sequencing we found novel ones; the miRNAs were monitored using relative real time-PCR.
  • Sequencing identified 24 novel and four conserved miRNAs. Further screening of the conserved miRNAs confirmed the presence of 16 additional miRNAs. Most of the miRNAs were targeted to unknown genes. The expression of seven conserved and nine novel miRNAs showed significant differences in transcript levels in the full-sib family showing distinct epigenetic difference in bud set, but not in the nonresponding full-sib family. Putative miRNA targets were studied.
  • Norway spruce contains a set of conserved miRNAs as well as a large proportion of novel nonconserved miRNAs. The differentially expression of specific miRNAs indicate their putative participation in the epigenetic regulation.


Norway spruce is an ecologically and economically important conifer having a wide geographic distribution and well adapted to a large range of environmental conditions. Conifers are masters of adaptation (Rohde & Junttila, 2008), despite exhibiting a very late sexual maturity and long generation intervals (> 30 yr). Yet conifers have been regarded as vulnerable to the rapid changes in temperature by classical evolutionary means (Rehfeldt et al., 1999; Rehfeldt et al., 2002). Thus, anticipated changes in global climate have been considered to represent a significant challenge for sufficiently rapid adaptation, especially for traits associated with the timing of the growth–dormancy cycles in such trees vital for their growth and survival. However, Norway spruce adjusts the adaptive performance by what appears to be an epigenetic response mechanism that is calibrated by the temperature conditions prevailing during embryo development. In both zygotic and somatic embryogenesis, the warmer the temperature conditions applied, the later the regenerated plants from these embryos formed terminal buds in a common environment. Despite being genetically identical the resulting adaptive change was large, equating to a provenance separation of 4–6 degrees of latitude. Moreover, the propagated plants from somatic embryogenesis displayed clonal variation in the memory (Kvaalen & Johnsen, 2008) and there is family variation in this trait (this work), strongly suggesting a genetic and heritable basis governing this epigenetic mechanism. Family and clonal materials are therefore well suited for identification of the genes and other regulatory mechanisms involved in this epigenetic memory mechanism.

The epigenetic memory could be understood as a type of adaptive phenotypic plasticity, which lasts in the following generation and realized through specific epigenetic patterns that established during development of the embryo and affect DNA replication, recombination, repair and gene expression. Previously, we have seen indications of correlation between the transcriptional regulation and the memory expression (Johnsen et al., 2005b). Our first step toward unravelling such a molecular mechanism is to identify the genes and other regulatory elements involved in the epigenetic regulation of adaptive traits controlling the growth–dormancy cycle in Norway spruce. Small RNAs are likely candidate regulatory elements that could play a part in this memory mechanism and thus were our target for investigation.

Small RNAs have specific regulatory roles and been implicated in epigenetic phenomena. They function in several pathways for gene regulation or silencing. These noncoding RNAs, which are 19–31 nucleotides (nt) long, behave as sequence-specific triggers for mRNA degradation, translation repression, heterochromatin formation and transposon control. Small RNAs can be classified into different groups based on their origin. In plants, small RNA groups include micro-RNAs (miRNAs) and small-interfering RNAs (siRNAs). miRNAs are noncoding RNAs of an average length of 22 nt, and derived from hairpin-structured single-stranded precursors, that facilitate translation repression in plants (Bartel, 2004; Kim, 2005; Yang et al., 2007; Axtell & Bowman, 2008; Morin et al., 2008; Carthew & Sontheimer, 2009) or additionally engage mRNA cleavage (Carthew & Sontheimer, 2009). siRNAs are derived from double-stranded (ds) RNA precursors and that silence genes by cleaving their target mRNAs (reviewed in Rana, 2007). Regardless of type and size, small noncoding RNAs share one unifying function in cellular physiology: regulation of gene expression (Chu & Rana, 2007) including epigenetic mechanisms.

Although significant progress has been made in identifying plant miRNAs and understanding their mechanism of action, the discovery of novel miRNAs in plants on a genome-wide scale is still at the early stage. Most plant miRNA studies have been done in angiosperms and few publications involving miRNAs in conifers and other gymnosperms exist. A total of 26 miRNAs from 11 families were identified in loblolly pine, possibly associated with the fusiform rust gall disease (Lu et al., 2007); seven miRNA families were loblolly pine-specific and four were conserved in other species. Recently, five additional conserved miRNAs were identified in loblolly pine seeds (Oh et al., 2008). In red pine, 11 conserved miRNAs were found in needle tissues, supporting the contention that many plant miRNA families have been conserved during land plant evolution (Axtell & Bartel, 2005). Sequencing of the Pinus contorta small RNA transcriptome allowed identifying 18 highly conserved and 51 novel miRNA families (Morin et al., 2008). The gymnosperm P. contorta have predominantly 21-nt long small RNAs and fail to produce significant amounts of the 24-nt small RNAs predominantly present in angiosperm species. Gymnosperms have specific Dicer-like family (DCL) genes not present in angiosperms (Dolgosheina et al., 2008). There appear to be no publications dealing with miRNA in spruce species until now.

We report here the identification of 44 miRNA in Norway spruce for the first time. Using relative real-time reverse-transcription polymerase chain reaction (RT-PCR) we found 16 miRNAs with differential expression in transcript levels in the full-sib family expressing distinct differences in bud set, but not in the nonresponding full-sib family, thus indicating their putative participation in the epigenetic mechanism. Putative targets were found for 27 confirmed miRNAs. Most of target genes had unknown functions but we found that four selected genes PaLPT4, PaGaMYB, PaMYB10 and PaSPB13 are likely regulated by miRNAs pab-miR100, 159a, 858 and 156c, and may also be involved in or at least correlated with the epigenetic memory regulation.

Materials and Methods

Plant material, growth conditions and sample collection

We used progenies from the two extreme full-sib families of Norway spruce (Picea abies (L.) Karst) in regard to the epigenetic memory mechanism with known differences in timing of bud set. We used progenies from two full-sib families of Norway spruce with known differences in the timing of bud set, as assessed by growing the seedlings in a glasshouse experiment in the autumn of 2005 (Johnsen et al., 2005a). Within each family seeds from plants regenerated after embryogenesis in cold environment (CE) and seeds after embryogenesis in a warm environment (WE) were used, giving a total of four seed types. The seeds were sown at 22°C under continuous light (long day = LD) in eight chambers. After 8 wk growth, four of the chambers were programmed to give short days (12 h light + 12 h darkness; SD). Progeny of family 1 showed the lowest difference in bud set between CE and WE and were considered as ‘epigenetically indifferent’. Progeny of family 6 showed the greatest difference in bud set between CE and WE, was considered ‘epigenetically responsive’. Ten shoots were harvested from both SD and LD treatments for 6 d and 20 d from the onset of SD treatment and immediately frozen in liquid nitrogen. Collections were done 4–5 h after onset of light in the morning.

Small RNA isolation and library construction

Small RNAs were isolated from 80 mg of tissue using the mir-Premier microRNA Isolation Kit (SNC-50; Sigma-Aldrich, St. Louis, MO, USA), according to the manufacture’s instruction. The quality of small RNA was assessed by Agilent 2100 Bioanalyzer with RNA 6000 Nano Kit (Agilent, #5067-1511, Santa Clara, CA, USA). Small RNA preparations were stored at −80°C.

To identify small RNAs, we used a direct cloning and sequencing approach, allowing for high-efficiency identification of new miRNAs (Zhang et al., 2006a). We made two small RNA libraries using IDT’s miRCat Small RNA Cloning Kit (Integrated DNA Technologies, Coralville, IA, USA) by following the manufacturer’s instructions. For library construction we used small RNA extracts (nearly 1 μg) from family 6, which had the highest ‘memory’ response. The first library contained concatenated small RNAs expressed in progeny after embryogenesis in WE in response to 20 d of SD treatment (WEL) and the second contained concatenated small RNAs expressed in progeny after embryogenesis in CE in response to 20 d of SD treatment (CEL). After concatamerization, small RNAs were cloned into pDrive vector using PCR Cloning Kit (Qiagen).

Small RNA sequencing and discovery of novel miRNAs

The small RNA libraries were partly sequenced at the ABI-lab of the Departments of Biology and Molecular Biosciences (University of Oslo, Norway). Trace files were processed for removal of vector sequences and poor-quality regions as well as contigs assembly with the help of secman ii sequence analysis software (DNAStar Inc., Madison, WI, USA). The sequences obtained were manually scrutinized for removal of linker/connectors in concatamer units. The small RNA sequences were assumed to be the sequence between the 5′ and 3′ linker/connectors, and small RNAs obtained were grouped based on their sequences. The small RNAs identified were compared with all published mature miRNAs sequences from the miRBase Sequence Database, release 14.0 (http://www.mirbase.org/index.shtml) (Griffiths-Jones et al., 2008). Comparison was done using blastn procedure with word size 7. Similarities with a score > 32 or an E-value of −2 were considered a hit (allowing a maximum of 2 nt mismatches).

All the small RNAs identified were also searched against the National Centre of Biotechnology Information (NCBI) expressed sequences tag (EST) database restrained by Picea taxa. We allowed only up to one mismatch between the small RNA and matching site at the forward strain orientation (sense hit). Hairpin structures were predicted using mfold software (Zuker, 2003) (http://frontend.bioinfo.rpi.edu/applications/mfold/cgi-bin/rna-form1.cgi) and RNAfold web server (Gruber et al., 2008) (http://rna.tbi.univie.ac.at/cgi-bin/RNAfold.cgi). It is known that plant pre-miRNAs vary from approx. 80 to approx. 200 nt in length (Zhang et al., 2006b), so we select regions of different length, from 100 to 400 nt on either side of the small RNA sequence for folding analysis.

RNA sequences were thought of as miRNA candidates only if they fitted the following criteria: a RNA sequence could fold into an appropriate stem–loop hairpin secondary structure; a mature miRNA sequence site in one arm of the hairpin structure; miRNAs had < 6 mismatches with the opposite miRNA* sequence in the other arm; predicted secondary structures had higher folding free energy indexes (MFEIs), negative minimal folding free energies (MFE <–25 kcal mol−1) and 30–70% A + U contents (Zhang et al., 2006b). These criteria significantly reduced assignment of false positives (Ambros et al., 2003).

Discovery and analysis of conserved micro RNAs

To identify potential Norway spruce conserved miRNAs, we defined a total of 274 previously known wooden tree species miRNAs (37 from Pinus taeda and 234 from Populus trichocarpa) from the miRNA Registry Database (Release 14.0, September 2009: http://www.mirbase.org/) as reference set of miRNA sequences. In addition, 715 known miRNA sequences from Arabidopsis thaliana, Oryza sativa, and Zea mays were chosen. To avoid the redundant or overlapping miRNAs, the repeated sequences of miRNAs within the above species were removed and the remaining sequences were used as query sequences. As only mature miRNAs, rather than miRNA precursor sequences, were conserved in plants (Zhang et al., 2006b), mature miRNA sequences were the focus of blast search. We used blast search against the spruce ESTs and nr databases, which were obtained from the NCBI Genbank nucleotide databases (http://ncbi.nlm.nih.gov). Any sequences not encoding protein located in forward strain with 0–1 mismatch to analysed mature miRNA were considered as candidate miRNA genes and used for folding analysis. Candidate mRNAs were checked using the web-based computational software mfold as described earlier.

Prediction of miRNA targets

When looking for potential mRNA targets of miRNAs, we used a blastn pattern search of obtained small RNA sequences against NCBI EST and nucleotide databases, confined to Picea taxa. We allowed only up to four mismatches between the candidate miRNA and miRNA target site at the reverse and compliment strain (antisense hit) in this prediction. The extracted sequences (ESTs) were combined into contigs to get full-length sequences when possible. We used a tblastx search of the ESTs nucleotide sequence against the NCBI database to identify putative gene homologues. Similarities with an E-value less than e−10 were considered a hit.

Relative real-time RT-PCR

Transcript abundance of the selected small RNAs were determined for the families 1 and 6 with relative real-time RT-PCR. cDNAs were synthesized from 600 ng of small RNA with the NCode miRNA First-Strand cDNA Synthesis Kit (MIRC-50; Invitrogen) following the manufacturer recommendations. Real-time RT-PCR amplification was performed using NCode SYBR GreenER miRNA qRT-PCR Kit (MIRQER-100; Invitrogen) in a 25 μl reaction volume, using 2 μl of a diluted cDNA solution already described as template and 200 nM of each primer. Reactions were conducted on the 7500 Fast Real-time PCR System (Applied Biosystems, Foster City, CA, USA) using the Invitrogen recommended cycling conditions. After PCR, dissociation curves were carried out to verify the specificity of the amplification. There were three biological replicates for each sample. All expression levels were normalized to geometric mean of three selected ribosomal and transfer RNA genes (Pa4.5S, Pa5S and PatRNA-R), showing most similar expression profiles among eight genes tested (see the Supporting Information, Table S1, Fig. S1). Forward primers were designed based on mature miRNA sequence. If Tm of mature miRNA was < 60°C, it had been adjusted by adding Gs and Cs to the 5′-end and/or As to the 3′-end of the miRNA sequence. The list of miRNAs studied and their primer sequences are shown in Table S2. To verify the specificity of the miRNA amplification, we analysed several PCR samples for each miRNAs on 2% agarose gels with ethidium bromide (EtBr) visualization of bands. Reverse primer was supplied with the NCode miRNA First-Strand cDNA Synthesis Kit (MIRC-50; Invitrogen).

Transcript abundances for the selected miRNAs target genes ESTs were determined in the full-sib family 6. The list of the ESTs and their primer sequences can be found in Table S3. Primers were designed using primer3 online software (Rozen & Skaletsky, 2000) available at http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi with calculated Tm of 70°C and amplification product not > 120 bp. cDNA was synthesized from 300 ng of total RNA with the TaqMan Reverse Transcription kit (Applied Biosystems) in 50 μl reaction volume and diluted three times. Real-time RT-PCR amplifications were conducted on the 7500 Fast Real-time PCR System (Applied Bio-systems) using the default cycling conditions. After each reaction, which included a no-template control, dissociation curve analyses were carried out to verify the specificity of the amplification. There were three biological replicates and all expression levels were normalized to actin.

Data acquisition and analysis were done using 7500-system SDS software for absolute quantification and MS Excel software as described previously (Johnsen et al., 2005a).


Small RNA sequencing and analyses

Two small RNA libraries were constructed from young seedlings of Norway spruce family 6 showing greatest differences in epigenetic response to the temperature conditions during embryogenesis. The first library contained small RNAs expressed in response to 20 d of SD treatment in the progeny after embryogenesis in WE (WEL) and the second contained small RNAs expressed in progeny after embryogenesis in CE (CEL). A total of 328 individual concatemerized clones were isolated and sequenced in both directions, yielding 323 small RNA sequences ranging in size from 17 to 24 nucleotides (Table 1 & Table S4). Sequence analyses revealed 191 distinct small RNAs; 103 different small RNAs were found only in WEL and 75 in CEL. Just 13 of the distinct small RNAs were shared between the two libraries. Most frequent small RNAs appear to be pab-smR02a (28 entries) and pab-smR02b (14 entries). Length distribution of small RNAs is summarized in Fig. 1. Both libraries contain predominantly small RNAs with 21 nt length, followed by small RNAs with 22 nt. Other length variants are represented by a few sequences.

Table 1.   Isolation and identification of conserved and novel Norway spruce specific microRNAs (miRNAs) and their putative targets (miRNA precursors are shown in the Supporting Information Table S5 and Fig. S2)
FamilyNameSequenceLength, ntArmmiRNA containing ESTTargetTarget gene functionScore/E-value
  1. n, No exact full-length matching in the NCBI spruces expressed sequence tags (ESTs) Database.

  2. 1Numbering just for internal reference and not corresponds to the miRNA family numbers at the miRBase.

  3. 2Obtained from sequenced libraries.

Novel miRNAs1
1pab-miR02aUCACAUCUGGGCCACGAUGGUU225′ES261905DR475593No significant similarity found
2pab-miR02mUCUGGGCCCCGGUGGUUUAUGA225′ES261905DR543641BAC83199.1 – TIR/P-loop/LRR disease resistance protein-like protein (Oryza sativa Japonica Group)59 (142)/9e-08
3pab-miR029AGAGGGUGCUCAUGAACUGCUC225′EX326419DR524266 (3 mismatches)AAY78890.1 – CC-NBS-LRR resistance-like protein (Pinus lambertiana)311 (796)/5e-83
7pab-miR065UAGCCCCUGACUUCAACAUGAG225′EX427311EX417859 (3 mismatches)AAM28917.1 – putative TIR/NBS/LRR disease resistance protein (Pinus taeda) 147 (372)/2e-42
8pab-miR070CUUGCAACUCUGCCUUGGCUUA225′EX404488DR451014 (3 mismatches)No significant similarity found
9pab-miR078GACAGAAGAUAGACUUUGGUC213′DR469293EX385049 (4 mismatches)Q03682 – BIP2_TOBAC Luminal-binding protein 2 (BiP 2) (78 kDa glucose-regulated protein homolog 2-Hsp70 family)366 (939)/2e-99
10pab-miR080GACGCCCAAAACUGAAGGUCA213′EX369017EX412904; CO484875 (3 mismatches)NP_172027.1 – thiF family protein (Arabidopsis thaliana) molybdopterin biosynthesis protein MoeB335 (860)/2e-90
DR474361 (1 mismatch);
EX370463 (2 mismatches)
No significant similarity found
12pab-miR100AAUCUCUUGGUGCUUAUUCGC215′DR571316EX358490 (4 mismatches) EX315011 (3 mismatches)NP_568201.1 – positive transcription elongation factor – Spt4/zinc ion binding (Arabidopsis thaliana) EAY72686.1 – hypothetical protein OsI_000533 (Oryza sativa (indica cultivar-group))185 (470)/4e-45 194(494)/6e-48
13pab-miR105UAGACUCUAUCAGCCUUGUCC215′EX312498EX406540 (2 mismatches)No significant similarity found
15pab-miR119GUAAGUGGUUAUGAUCUGGAC215′DR499063DR491334 (1 mismatch)AAM28917.1 – putative TIR/NBS/LRR disease resistance protein (Pinus taeda)212 (539)/3e-53
18pab-miR132UCACACAACAUUGCUCGUACA215′CO213440 (1 mismatch)
19pab-miR144aUCAGAUGCUUUAAAUUCCCGA215′DR567751; CO212009EX370466 (3 mismatches)CAN67425.1 – hypothetical protein (Vitis vinifera)114 (285)/1e-23
20pab-miR144bUUUAAAUGCCUUAAAUUCCCGA225′DR551675EX343057; (2 mismatches)CAO63167.1 – unnamed protein product (Vitis vinifera)119 (299)/2e-42
Conserved miRNAs
252pab-miR159a/pta-miR159aUUGGAUUGAAGGGAGCUCCA20No hairpin structureEX379646; CO226626EX411402 (1 mismatch)BAB85242.1 – transcription factor GAMyb (Oryza sativa Japonica Group)83 (205)/2e-14
26pab-miR160.1/ptc-miR160aUGCCUGGCUCCCUGUAUGCCA21 CO484471EX444394; EX442742 (2 mismatches)ACI13681.1 putative auxin response factor ARF16 (Malus × domestica)192 (487) 3e-47
27pab-miR160.2/ptc-miR160aUGCCUGGCUCCCUGUAUGCCA21 DR468340
28pab-miR165a/ath-miR165a/pta-miR166a UCGGACCAGGCUUCAUUCCUC21 GH284233DR519274 (3 mismatches)ABD75310.1 class III homeodomain-leucine zipper protein C3HDZ2 (Pseudotsuga menziesii)535 (1378) 2e-150
29pab-miR166a/pta-miR166aUCGGACCAGGCUUCAUUCCUU21 GE475070ES874991No significant similarity found
30pab-miR395/ptc-miR395aCUGAAGUGUUUGGAGGAACUU21 EX396486
31pab-miR396a/pta-miR396UUCCACAGCUUUCUUGAACUA21 EX431337CO481346No significant similarity found
32pab-miR396b.1/ptc-miR396aUUCCACAGCUUUCUUGAACUG21 EX431337CO481346No significant similarity found
33pab-miR396b.2/ptc-miR396aUUCCACGGCUUUCUUGAACUU21 CO480982 (1 mismatch)
34pab-miR397/ptc-miR397aUCAUUGAGUGCAGCGUUGACG21 DR451001EX356746; DV996417; BT070419; EF677723AAK37828, AF132124 laccase (Pinus taeda)475 (1223)/2e-132
35pab-miR482a/pta-miR482aUCUUCCCUACUCCUCCCAUUCC22 EX443975
36pab-miR482e/pta-miR482a UCUUCCCUAUUCCUCCCAUUCC22 EF087815
37pab-miR482f/pta-miR482aUCUUUCCUACUCCUCCCAUUCC22 GH281854
382pab-miR529b/osa-mir529bAGAAGAGAGAGAGUACAGCCU21nnEX357165 (2 mismatches)CAN69561.1 – hypothetical protein (Vitis vinifera)51 (121)/9e-05
39pab-miR535/osa-miR535 UGACAACGAGAGAGAGCACGC21 ES663083
40pab-miR947a/pta-miR947UAUCGGAAUCUGUUACUGUUUC22 EX312403EX312737No significant similarity found
412pab-miR949/pta-miR949UCUCCGGGAAUCCAAUGCGCCU22nnDR547733 (2 mismatches)No significant similarity found
EX417324AAM28917.1 – putative TIR/NBS/LRR disease resistance protein (Pinus taeda)157 (396)/6e-37
43pab-miR1311/pta-miR1311UCAGAGUUUUGCCAGUUCCGCC22 ES248264DR565328 EX311606No significant similarity found
44pab-miR1863a/osa-miR1863AGCUCUGAUACCAUGUUAGAUU22 EX352990EX357058ABK21509 unknown (Picea sitchensis)73 (178)/8e-12
Figure 1.

 Length distribution of unique small RNA sequences from Picea abies (light bar, 193 sequences from WEL; dark bar, 130 sequences from CEL)

We checked all of the small RNAs for similarity to the miRNAs described earlier, placed into miRBase Database (release 14.0). Only four small RNAs had sequences identical, or highly similar, to those of conserved plant miRNAs that belongs to miR159, miR529, miR949 and miR951 families. All the other 195 novel small RNAs had no matches in the miRNA database.

Discovery of novel P. abies miRNAs using spruce ESTs

We further searched with the small RNA sequence patterns obtained against the NCBI est_others ESTs and nucleotide collection databases confined to Picea (taxid:3328), allowing up to four mismatches between the small RNA and matching site. We found hits for 98 of the small RNAs among spruce ESTs, while 101 small RNAs showed no matches with known transcripts (Table S4). These 99 small RNAs with no matches (excluding two small, RNAs which were identified among conserved miRNAs) were not further analysed owing to lack of any information about their origin sequences or putative targets. For the conserved pab-miR949/pta-miR949 and pab-miR529/osa-miR529b we found no matching sequences among the Picea sequences publicly available, so designation of these as miRNAs is based solely on their sequence matching known miRNAs.

Among the contigs with matches to small RNAs, we selected candidate miRNA genes among sequences not coding for a protein and containing a mature small RNA sequence located in forward strand orientation (sense hit), allowing up to one mismatch. In total, we found 37 such candidate miRNA genes. Sequences upstream and downstream of the matching region (≈ 200 bp) for the 37 candidate miRNAs were used to predict the secondary structure using mfold and RNAfold web servers. Stable stem–loop miRNA precursor structures were identified for 25 of the candidates. The EST ES261905 was predicted to be a precursor for both pab-miR002a and pab-miR002m situated on different shoulders of stem–loop structure. Contig FD735299 was predicted to be a precursor for two copies of novel pab-miR154ns located opposite each other on 5′ and 3′ arms of the precursor. All the other ESTs were predicted precursors for single miRNAs. These small RNAs were designated as pab-miRNAs (Tables 1, S5, Fig. S2). For conserved pab-miR159a/pta-miR159a we identified several ESTs with matching region, but stem–loop hairpin structure was not confirmed, therefore we define pab-miR159a, pab-miR949 and pab-miR529 as miRNAs, based solely on their sequence matching to previously described miRNAs despite the lack of in silico precursor structure confirmation.

The designated 28 miRNAs could not be combined into miRNA clusters and miRNA families. Each miRNA is likely a single representative of their miRNA families. Only pab-miR144a and b showed similarity to each other (but still 7 bp differences) and we considered them tentatively as representatives of the same novel miRNA family. Four miRNAs – pab-miR159a, miR949, miR951 and miR529b – are likely single representatives of four miRNA families conserved in other plant species (loblolly pine, Arabidopsis and rice). All other putative families obtained were novel (Table 1).

Discovery and analysis of conserved microRNAs in Norway spruce

We searched the defined reference set of plant miRNAs by blastn against known spruce ESTs set at the NCBI EST database. The ESTs having high similarities (allowing with 0–1 mismatches) to the reference miRNA on the forward strand (sense hit) were selected as candidate miRNA genes, while transcripts (ESTs) containing up to four mismatches to miRNA sequence on reverse strand were designated as putative miRNA targets. In total, by screening of nearly 1000 described miRNAs from other plant species, we identified 60 spruce EST contigs containing miRNA matching regions. We selected 32 of these ESTs in which miRNA-matching region was located in the forward strand, as microRNA gene candidates. A careful evaluation of the predicted secondary structures, using the criteria described in the method section, allowed us to confirm stem–loop hairpin-folding patterns for 16 of these candidates (Table 1). These predicted precursor sequences and microRNA homologues were further considered as Norway spruce microRNAs. For the remaining 15 contigs, no indications of hairpin structures were found.

The sixteen such identified Norway spruce miRNAs were found to fall into 11 miRNA families: miR396 and miR483 each has three members, miR160 and miRNA165/166 has two members, while for other miRNA families, such as miR395, miR397, miR535, miR947, miR951, miR1311 and miR1836, only one member was predicted. Most of these microRNAs (14 of 17) were similar to known loblolly pine and Populus miRNAs and only few matched other plant species miRNAs.

Prediction of the potential miRNAs and small RNAs targets in Norway spruce

To find the putative target genes regulated by the novel and conserved miRNAs, we made pattern blast search against NCBI EST and nucleotide databases, confined to Picea taxa. We allowed only up to four mismatches between the candidate miRNA and miRNA target site at the reverse and compliment strain (antisense hit) in this prediction. The mature sequences of the 44 confirmed pab-miRNAs (Table 1) and 45 unconfirmed conserved miRNAs (Table S6) were analysed.

Putative functions of the potential targets were assigned based on similarity to putative annotated homologues by obtained contigs blastx against NCBI protein database. Among the confirmed pab-miRNAs, five of the putatively targeted ESTs were homologous to TIR(CC)/NBS/LRR disease-resistance proteins from pine species. The predicted targeted ESTs for 11 of the pab-miRNAs showed no significant similarity to known genes and four of the pab-miRNAs corresponded to hypothetical or unnamed protein products. For 18 of the pab-miRNAs we did not find any putative targets. Thus, 29 pab-miRNAs may target Norway spruce genes that are not yet sequenced (Table 1). Only six pab-miRNAs putatively targeted to annotated functional genes and transcription factors.

Among the predicted (but unconfirmed) conserved spruce miRNAs, 22 of the putative target genes were homologous to structural genes, as well as transcription factors. For 10 conserved miRNAs, no target was found, and 8 more matched genes with unknown function.

Expression of selected Norway spruce miRNAs for epigenetically different samples

With real-time RT-PCR we characterized the expression of all defined pab-miRNAs in the Norway spruce seedling from the Norway spruce full-sib family 1 (F1, having low ‘epigenetic memory’ response) and family 6 (F6, having a distinct ‘epigenetic memory’) after embryogenesis in CE and WE at LD conditions and after SD treatment. The expression of seven conserved and nine novel miRNAs showed significant differences in transcript levels in the full-sib family expressing distinct differences in bud set, but not in the nonresponding full-sib family.

Among novel miRNAs only pab-miR144a showed large differences in transcript abundances between WE and CE from F6 samples at SD6 and small differences for F1 (Fig. 2). The largest group of miRNAs – pab-miR080, 100, 105, 119, 122, 132, 144a,b and 157 – showed large differences in amounts of transcript in the SD20 samples for F6 and showed small differences between WE and CE for the SD20 samples of F1. Thus, these miRNAs showed transcription patterns compatible with being putatively involved in, or affected by the epigenetic memory mechanism. Among these, the most interesting were pab-miR105, 119, 122, 132 (greatest difference between WE and CE SD20 samples at F6) and miR144a (significant difference between WE and CE samples for both SD6 and SD20 at F6); These were considered as being the more likely candidate regulators, which could possibly involved in the molecular epigenetic memory mechanism.

Figure 2.

 Transcript profiles of selected Norway spruce novel(n) and conserved microRNAs (miRNAs) putatively involved in epigenetic regulation. Samples are seedlings from families 1 and 6 with low and high epigenetic memory response (correspondently) after cold (CE) and warm (WE) environment treatment at days 6 and 20 under short day (SD) and long day (LD) photoperiod conditions. Transcript level was measured as the difference between geometric average of three reference genes –Pa4.5S, Pa5S and Pa-tRNA_R (endogenous control) and the chosen transcripts relative to the mean value for the genes targets (−dCt). Bars indicate standard error of means. Numbering of novel miRNAs just for internal reference and not corresponds to the miRNA family numbers at the miRBase.

Among the conserved pab-miRNAs we found several miRNAs (pab-miR395, 396a, 396b, 535, 947 and 951) that showed significant difference in expression between WE and CE samples for F6 family and virtually no difference for F1 family progeny (Fig. 2). In addition, pab-miR160 and 395 had difference in transcript levels between WE and CE samples of family F6 for both day 6 and day 20, but were found to have significant differences for F1 under SD20 conditions and no differences under LD for all F1 samples. These pab-miRNAs could potentially be involved in epigenetic memory regulation. The other pab-miRNAs studied did not reveal any consistent patterns of expression and are probably not related to the epigenetic phenomenon.

The conserved pab-miRNAs found tended to have higher expression (Ct –9 to 23) than the novel miRNAs (Ct –20 to 33). The distinct peak after melting curve analysis and consistent expression patterns of pab-miRNAs observed strongly support that they were correctly defined as miRNAs.

Study of pab-miRNAs and their putative target coexpression and verification of regulation

To test and support regulation of putative target genes by miRNAs, we tried to confirm antagonistic expression of the miRNA to its predicted target mRNA (i.e. establish if there was a relationship between small RNA transcript abundance and transcript-level changes of their putative target gene mRNA). For this purpose, we selected conserved and novel miRNAs (even if we did not confirm stem–loop precursor structures in some cases) for which known functional gene targets were predicted. We studied this, using relative RT-PCR transcript-level differences of miRNAs and target gene mRNAs in parallel for the same samples from the seedlings of Norway spruce family 6 (epigenetically responsive), from CE and WE embryogenesis at LD and after 6 d and 20 d of SD. In total, we tested 10 conserved and 2 novel miRNAs (Table S7).

Finally, we picked four miRNA-mRNA pairs that were significantly different in transcript abundances of both miRNAs and mRNAs using the same samples (Fig. 3). We expected that the abundance of target mRNA became reduced when the presence of small RNAs transcripts increased and vice versa.

Figure 3.

 Expression of conserved microRNA (MiRNA) and novel miR100n (a) and their putative targeted mRNAs (genes)(b) in Norway spruce family 6 samples originated after embryogenesis in cold (CE) and warm (WE) environment at long day (LD) and after 6 and 20 d of short day (SD) treatment using relative RT-PCR. Expression level of mRNAs was measured as the difference between PaAct (endogenous control) and the chosen transcripts relative to the mean value for the target genes. Bars indicate standard error of means. Transcript levels of miRNAs were normalized to geometric mean for three selected ribosomal and transport RNA genes (Pa4.5S, Pa5S and PatRNA-R).

PaGaMyb (gibberellic acid MYB transcription factor) seemed to be regulated by pab-miR159a (pta-miR159a), as shown in Fig. 3. Significantly decreased amounts of miR159a transcripts at day 6 to day 20 coincided with the accumulation of targeted mRNAs for WE samples, both under SD and LD. For the CE samples, there was no significant difference in transcript abundances of miRNAs and mRNAs. A similar pattern was found for transcription elongation factor PaSPT4 mRNA, putatively regulated by pab-miR100 (Fig. 3). For pab-miR858 targeted PaMyb10 (transcription factor MYB10) the results were less clear (Fig. 3). A significant decrement of pab-miR858 transcripts from day 6 to day 20 under SD for CE samples was accompanied by a higher abundance of putative targeted mRNAs. However, no interrelation in levels was found for WE samples. At LD conditions, the increment of siRNAs did not lead to a decrease of mRNAs. PaSPB13 (Squamosa promoter-binding SBP-domain like protein 13) could be putatively regulated by small RNA pab-siR156c. PaSPB13 mRNA had significant differences in expression between WE and CE for day 20 samples from seedlings grown under SD only. Under SD20 conditions, a significant decrease in the transcript abundance of pab-miR156c coincided with accumulation of PaSPB13 transcripts, while nonsignificant differences in miRNAs transcript levels correspond to nonsignificant differences in mRNA levels under other conditions (Fig. 3).


It is well known that miRNAs play an important role in regulating a variety of biological processes. The regulation mechanisms include repression of translation and cleavage of targeted mRNAs (Carthew & Sontheimer, 2009). The miRNAs may directly target transcription factors which affect plant development, and specific genes which control metabolism. Many mature miRNAs are evolutionarily conserved in the plant kingdom, which provides a powerful approach to predict the existence of miRNA orthologues in other plants (Carthew & Sontheimer, 2009; Zhang et al., 2009).

One of the challenges in the study of plant miRNAs is to identify novel miRNAs. For wooden plants with very long generation times and limited genome sequence information, such as most conifer tree species, direct isolation, cloning, and transcript sequencing and similarity search among known ESTs, are currently the best alternative methods for miRNA discovery. We aimed to identify Norway spruce miRNAs that were differentially expressed in full-sib progenies from two families, where one expressed the epigenetic memory and the other did not. In this way we hoped to identify the most pertinent candidates, which could act in the adaptive epigenetic memory mechanism affecting the bud phenology and frost hardiness of Norway spruce (Johnsen et al., 2005a,b; Kvaalen & Johnsen, 2008). Toour knowledge, our report is the first one dealing with miRNAs in spruce.

The small RNAs were isolated and sequenced from two libraries. In total, we obtained 199 distinct small RNA sequences. The length distribution of small RNA sequences showed the absolute prevalence of 21 nt small RNAs; considerably less well represented were the 22 nt small RNAs and the occurrence of the 24 nt length small RNAs was rare. This distribution has been shown for gymnosperms (Dolgosheina et al., 2008). Search at the miRBase revealed just four conserved miRNA, which could be categorized as orthologues of the loblolly pine pta-miR159a, miR949, miR951 and rice osa-miR529b miRNA genes, suggesting that our Norway spruce small RNA collection likely includes novel miRNAs not previously described. However, for three conserved miRNAs homologous to pta-miR159a (pab-miR159a), miR949 (pab-miR037) and osa-miR529b (pab-miR038), we could not confirm precursor with hairpin structure among available spruce ESTs. We suspect that these miRNAs could originate from introns of genes and be expressed during mRNA maturation (Carthew & Sontheimer, 2009). We had similar results when analysing the reference set of miRNA sequences from other species. The ESTs of miRNA genes could not be found, and neither did we manage to confirm hairpin folding patterns for 42 from 60 conserved miRNAs that we found to be homologous to sequences in the spruce EST Database. Thus, EST collections were not the only source for the miRNAs search because they could only help us to identify miRNAs originated from exons, together with expressed product, or from 3′ UTR (Bartel, 2004; Axtell & Bowman, 2008). Full genome sequence of any of the spruce species will promote considerably the discovery and confirmation of miRNAs. Based on our data we anticipate that part of the novel small RNAs obtained could also be confirmed as miRNAs in the future.

We identified stem–loop precursor structures for 25 novel small RNAs and 17 conserved miRNAs, which met all the criteria listed, and these were designated as true miRNAs. Three small RNAs were assigned as miRNAs based on homology to conserved miRNAs. All ESTs that coded for miRNA genes were homologous to unknown genes. This indicates that a large portion of such kind of ESTs, with no significant similarities in the Databases, could be an excellent source for identifying new miRNA genes in future research.

We consider that direct cloning and sequencing, with the construction of concatemerized small RNA libraries, is a promising avenue for identifying new miRNAs. We sequenced only a very limited number of clones (328) from WEL and CEL, but still we obtained quite large number of distinct small RNAs (193), from which we further identified 24 novel and 4 conserved miRNAs. These became founders of 20 miRNA families. Contemporary next-generation sequence approaches will greatly facilitate the identification of new miRNAs. In particular, such sequencing would allow changes in small RNA expression levels to be monitored by the sequencing process itself, greatly speeding up the discovery of small RNAs that change expression owing to developmental condition.

The lack of conserved miRNAs among sequences we found could indicate novel elements specifically associated with ‘epigenetic memory’ regulation, which in turn could affect developmental processes in Norway spruce. Perhaps they also could serve as a source for novel epigenetic elements in general, and these newly identified miRNAs will be studied further by us to find their target genes and their functions. In the absence of the full transcriptome and genome sequence for spruce, the microRNA genes reported here likely represent an important part of the genes that produced the mature miRNAs but probably there is a significantly larger number of such genes present in this species, but not in the EST collections available.

Identification of miRNA target genes has been a great challenge. Currently, there is no clear consensus as to what criteria we should follow to determine miRNA targets and to confirm their biological efficacy (Kuhn et al., 2008). For plants, miRNAs are not strictly located at 3′ UTR, but could be placed in UTRs, exons and nontranscribed part of the genome (Fahlgren et al., 2007; Carthew & Sontheimer, 2009). Most of the miRNAs have, however, been shown to be nearly perfectly complementary to their targets (Meyers et al., 2006). Thus, we used a computational approach for searching of putative targets, looking for matching sites at the reverse strand of ESTs.

Unfortunately, we did not find target genes for most of the novel and conserved miRNAs. In addition, a considerable amount of pab-miRNAs’ putative target genes were unknown or without significant similarity to other genes at the Databases. Lack of target genes for confirmed miRNAs in the quite comprehensive spruce ESTs Database could imply the existence of unknown and specific epigenetic regulation pathways, which implies a necessity to make specific EST libraries related to the epigenetic memory regulation transcriptomes.

Another big share of miRNA target genes were homologous to TIR(CC)/NBS/LRR disease-resistance proteins, which could play an important role in the plant defence system or possibly the extracellular domains of such proteins could act as receptors for sensing other extracellular cues. Moreover, from the 67 small RNAs with identified putative target genes among the spruce ESTs, 12 distinct small RNAs had exact matching to ESTs that were homologous to different loblolly pine TIR(CC)/NBS/LRR disease-resistance proteins. Disease resistance LRR genes could be among the genes often targeted by miRNA genes. It is shown for Brassica that miRNA genes can originate through inverted duplication events from TIR-NBS-LRR disease-resistance protein-coding gene sequences (He et al., 2008) and participate in their regulation. Relatively small amount of miRNAs were targeted to known structural genes and transcription factors.

Relative real time RT-PCR was used to investigate the expression of 24 novel and 21 conserved miRNAs in Norway spruce seedlings originated after embryogenesis on CE and WE of family F1 (having low ‘epigenetic memory’ response) and F6 (having high ‘epigenetic memory’ response) under LD conditions and after SD treatment, which initiated growth cessation and bud set. As selection criterion of miRNAs involvement into epigenetic regulation we considered significant differences in transcript levels between WE and CE originated samples of family 6 and lack of differences in family 1 or vice versa.

We found several candidates between novel and conserved miRNAs that correspond to our selection criterion. The candidates among novel miRNAs were miRNAs pab-miR029, 080, 100, 105, 119, 122, 132, and 144a and b. We found it interesting that pab-miR061 showed significant differences in transcript abundances at SD samples for F1, but not for F6. Among the conserved miRNAs, the best candidates were pab-miR156c, 159a, 160, 395, 396a,b, 535, 858, 947 and 951, despite the fact that we did not find miRNA genes or did not have confirmed hairpin structure for several of them (pab-miR156c, 159a, 858). The precise roles of these candidate pab-miRNAs and their genetic interactions with target transcripts in epigenetic regulation of bud set need to be examined further.

For the most novel miRNAs, significant differences between WE and CE origins (when detected) lasted until day 20 after SD treatment. This could reflect the fact that we used SD20 samples for constructing the miRNA libraries, and mainly miRNAs that differentiated at SD20 were cloned, or that some miRNAs regulate genes in the late SD-response, reaching their high levels of accumulation or suppression > 6 days after SD treatment. Only pab-miR029 showed significant differences between both WE and CE samples of F6 at SD6, and pab-miR144a were significantly different both for SD6 and SD20. pab-miR144a targeted hypothetical protein gene and should be studied further.

We found only 12 miRNAs (including both novel and conserved) that targeted genes with putative known function. For the conserved miRNAs, their target sites should possibly also be conserved across different plant species. As expected, the targets for conserved miRNAs in spruce were similar or functionally related to previously validated plant miRNA targets. For example, pab-miR397 could target laccase and laccase precursor (Table S3). Laccase has also been predicted to be the target of miR397 in Arabidopsis and rice (Abdel-Ghany & Pilon, 2008; Xue et al., 2009). pab-miR160 targeted AUXIN RESPONSE FACTOR10, similar to what was described in Arabidopsis (Liu et al., 2007), and pab-miR173 targeted PaAGO1 and pab-miR396a targeted PaGRF, which was also known from studies with Arabidopsis (Montgomery et al., 2008; Liu et al., 2009).

Putative targets were also predicted for some newly identified miRNAs. We used antagonistic presentation of miRNA-mRNA transcript abundances, but should mention that these observations are not an absolute measure as other RISC (RNA-Induced Silencing Complex) components and family members with different sequences may come to play for actual regulation. We showed that pab-miR159a, orthologous to pta-miR159a, likely regulates expression of PaGaMYB in Norway spruce. The GaMYB gene could respond to GA signals and thus bei involved in GA transduction pathways, participating in numerous developmental processes, including seed germination (Tsuji et al., 2006; Gong & Bewley, 2008) and flower development (Achard et al. 2004; Tsuji et al., 2006). miR159 is a conserved miRNA family (Allen et al., 2007). miR159 is a phytohormonally regulated homeostatic modulator of GAMYB activity, and hence of GAMYB-dependent developmental processes (Achard et al., 2004; Millar & Gubler, 2005; Tsuji et al., 2006; Allen et al., 2007). Further studies of pab-miR159a/miR159a and PaGaMYB involvement in the epigenetic regulation of development should be conducted, along with measurements of GA regulation in relation to growth cessation (Olsen et al., 1997).

Another spruce gene homolog, which is likely regulated by miRNA is PaSPB13. SQUAMOSA promoter-binding proteins (SBP) is a family of transcription factors possessing a SBP-domain, which plays important roles in plant development, including regulation of shoots development and floral transition. The SBP genes are targeted by the highly similar miRNAs miR156 and miR157 (Wu & Poethig, 2006; Gandikota et al., 2007; Riese et al., 2007). Our results generally confirm miRNA (pab-miR156c) targeted regulation of PaSPB13 gene expression. Putative participation of these transcription factors and pab-miR156c may be involved in the regulation of the epigenetic memory Norway spruce, as has been described for tomato (Manning et al., 2006). pab-miR100 could regulate the expression of PaSPT4. SPT4 is a transcription elongation factor, which affects elongation by Pol II and influences growth and rRNA synthesis rates (Schneider et al., 2006). We found significant accumulation of PaSPT4 transcripts from SD 6 to SD20 in WE spruce seedlings of family 6 and drastic decrease of pab-miR100 transcript abundance, but no PaSPT4-transcript changes in CE spruce seedlings of family 6 and family 1. Thus, this gene and the regulatory pab-miRNA expression patterns are consistent with a putative involvement in epigenetic developmental regulation.

We expected that the expression of another MYB transcription factors could be regulated by miRNAs. The miR159 family miRNAs have been described as regulators of MYB factor genes in Arabidopsis (Reyes & Chua, 2007), and miR399 in bean roots (Valdés-López et al., 2008). Some lack of interrelation between transcript abundance of miRNA858 and the targeted PaMyb10 gene could possibly be caused by a delay between changes in small RNA abundance and changing amounts of target mRNA transcript. Alternatively, in the case of post-transcriptional repression instead of mRNA cleavage, there could appear to be no direct interrelation. Despite some discrepancies, all the chosen elements for deeper analysis of the miRNA–mRNA pair, we found significant differences in transcript abundances among WE and CE samples both for miRNAs and mRNAs. Thus, we considered them as good candidates for future studies in relation to the epigenetic memory regulation.

There are still a great number of questions that remain to be answered regarding Norway spruce miRNA functions and targets, but our study demonstrates the existence of a set of conserved miRNAs and a large proportion of novel nonconserved miRNAs with relatively low expression levels. Nearly all genes targeted by miRNA were unknown, so unknown and hypothetical genes could be the main ‘players’ in epigenetic regulation. By analysing putative miRNA target genes, we confirm the differential expression of several genes as a result or initiator of epigenetic regulation. The miRNAs could also help us in finding the candidate genes. A considerable proportion of the novel and conserved miRNAs were differentially expressed in relation to whether the siblings originated from CE or WE. These findings imply that both kinds of miRNA might be involved or at least could be affected by the molecular mechanisms that regulate the temperature-sensitive epigenetic memory. We believe that we are at the beginning of a very important endeavour, where further studies of miRNAs and their target genes will help us to acquire a better understanding of this exciting phenomenon.


We thank Monica Fongen (Norwegian Forest and Landscape Institute) for excellent technical help during small RNA extraction and libraries construction. This work was supported by the Norwegian Research Council (Grants # 191455) and the Norwegian University of Life Sciences (UMB).