• MicroRNAs (miRNAs) are known to regulate plant development, but have not been studied in gymnosperm seed tissues. The presence and characteristics of several miRNAs were examined in zygotic embryos (ZEs) and female gametophytes (FGs) of Pinus taeda (loblolly pine).
• Evidence for miRNAs was obtained using northern analyses and quantitative reverse transcription polymerase chain reaction (qRT-PCR) mediated with poly(A) polymerase. Partial sequences of two miRNAs were verified. Three regions of putative mRNA targets were analyzed by qRT-PCR to monitor the occurrence of stage-dependent miRNA-mediated cleavage.
• Five miRNAs were identified in ZEs and FGs along with partial sequences of Pta-miR166 and Pta-miR167. Both miRNAs showed differing degrees of tissue-specific and stage-specific modulation. Analysis of HB15L mRNA (a potential Pta-miR166 target) suggested miRNA-guided cleavage in ZEs and FGs. Analysis of ARF8L mRNA (a potential Pta-miR167 target) implied cleavage in ZEs but not in FGs. Argonaute9-like mRNA (ptAGO9L) showed stage-specific modulation of expression in ZEs that appeared to be inverted in the corresponding FGs.
• MicroRNAs and argonaute genes varied spatiotemporally during seed development. The peak levels of Pta-miR166 in FGs and ptAGO9L in embryos occurred at stage 9.1, a critical transition point during embryo development and a point where somatic embryo maturation often stops. MicroRNAs identified in FG tissue may play a role in embryogenesis.
One of the most distinctive features of conifer embryogenesis is the development of a diploid conifer embryo that is surrounded by the haploid FG (Fig. 1a). The FG functionally corresponds to the endosperm in angiosperms. The relatively large size of P. taeda embryos permits the study of stage-specific development in embryo and FG tissues. Figure 1(b) illustrates embryo development for both zygotic embryos (ZEs) and somatic embryos (SEs) (for description, see Pullman & Webb, 1994). Somatic embryogenesis is a useful method for studying conifer embryogenesis and has great potential to generate superior growing stock for plantation forestry (McKeand et al., 2007). During somatic embryogenesis, the culture consists of suspensor-like cells and SEs at various stages of development, but the extra-embryonal FG tissues are not present (Zoglauer et al., 2003). If the haploid FG produces compounds necessary for normal embryo development, these potentially important compounds would be missing in culture and must be added as medium supplements. The percentage of SEs that complete all stages of development is small (Pullman et al., 2003). Transcript analysis of P. taeda embryos has revealed over 3000 novel expressed sequence tags (ESTs) unreported in any other conifer tissue (Cairney et al., 2006; Cairney & Pullman, 2007). Molecular mechanisms regulating the development of P. taeda embryos are poorly understood.
One of the best characterized miRNA families in plants is miR165/166. Eight copies of DNA sequences encoding this family of miRNAs are present in the Arabidopsis genome. miR165/166 molecules are localized in shoot apical meristem (SAM), leaf primordia and vascular tissues in Arabidopsis. Five class III homeodomain leucine-zipper (HD-ZIP III) genes, namely Revoluta (REV), Phavoluta (PHV), Phabulosa (PHB), ATHB-8 and ATHB-15, are predicted target genes, and these mRNAs have been shown, in biochemical assays, to be cleaved at the miR165/166 target sequence (Tang et al., 2003; Kim et al., 2005; Jung & Park, 2007; Zhou et al., 2007). Phenotype analyses of plants with target genes mutated at the cleavage site reveal alterations in meristem functions, organ polarity, and vascular differentiation and patterning.
Another well-studied pair of miRNAs, namely miR160/miR167, regulate the mRNAs of auxin response factors (ARFs) that are associated with auxin homeostasis (Tian et al., 2004; Mallory et al., 2005; Sorin et al., 2005; Wang et al., 2005; Yang et al., 2006). In Arabidopsis, a transcription factor ARF17 mRNA is cleaved by miR160. The ARF17 represses downstream expression of the GH3 family that encodes auxin-conjugating proteins, resulting in an increase of the active indole-3-acetic acid (IAA) level in the cell. Cleavage of ARF17 mRNA, mediated by miR160, may reduce the amount of ARF protein, thus releasing the repression and decreasing the levels of free IAA (Mallory et al., 2005; Sorin et al., 2005; Wang et al., 2005). A recent report showed that regulation of ARF8L by miR167 is positively linked to the expression of GH3 (Yang et al., 2006). Thus, miRNAs that target positive regulators, such as ARF6 and ARF8, and negative regulators, such as ARF17, appear to fine-tune the expression of GH3 family genes, thereby regulating the availability of free IAA.
While a growing number of plant miRNAs have been found and predicted both by direct cloning approaches and by bioinformatics analyses, only two published reports include information on miRNAs in a gymnosperm (Axtell & Bartel, 2005; Lu et al., 2007). In the first study the presence of 11 well-conserved miRNAs was detected in needle tissues of the gymnosperm, Pinus resinosa (red pine) (Axtell & Bartel, 2005). In the second study, 26 miRNAs in xylem tissues were reported and their expression was investigated in tissue infected with a fungus that caused fusiform rust galls (Lu et al., 2007).
In the current work, we explored the presence of miRNAs in embryos and female gametophytes of loblolly pine (P. taeda). We chose five well-studied miRNAs: miR159, Pta-miR166, Pta-miR167, miR171 and miR172. These miRNAs were identified in ZE and FG tissues and were examined for stage-specific modulation. We also examined several of the known mRNA targets for miRNA-guided cleavage in seed tissues as a function of stage of development and the expression of different argonaute genes possibly involved in miRNA metabolism. A polymerase chain reaction (PCR) validation method was used to reveal partial sequences of Pta-miR166 and Pta-miR167 (Lim et al., 2003).
Materials and Methods
Plant materials and RNA preparations
Cones from two open-pollinated loblolly pine (P. taeda L.) mother trees (7–56 and S4PT6) were collected weekly from approx. 15 June to 30 September in 1997, 2003 and 2007. Zygotic embryos and FGs were dissected from seeds, staged quickly using a dissecting scope (Pullman & Webb, 1994) and immersed in liquid nitrogen (Fig. 1a). Stage 9 embryos were categorized by the week of collection: 9.1 (stage 9, week 1), 9.2 (stage 9, week 2), etc. Needles were collected on 11 October 2005 from a seed orchard tree, and whole seedlings (GS), including root, hypocotyl, cotyledon and epicotyl, grown from surface-sterilized seed (Pullman et al., 2006) and germinated on medium 397 (Vales et al., 2007), were harvested after incubation in a light-controlled growth room for 5 wk. In addition, several different genotypes of liquid suspension culture (LSC) cells (including SE tissues) were maintained as described previously (Pullman et al., 2003), and cells were collected weekly and stored at −80°C. Total RNA was isolated from whole GSs, from stage-specific ZEs and FGs, and from SEs (a mixture of early stages 1 to 3), using TRI-reagent (Sigma-Aldrich, St Louis, MO, USA) according to the manufacturer's instructions or by the method using hexadecyl-trimethyl-ammonium bromide (CTAB) that was developed for the elimination of polysaccharides, phenolics and RNases associated with pine tissues such as FG, needle, xylem and seedling tissues (Chang et al., 1993).
Gel electrophoresis and northern analysis
Low-molecular-weight (LMW) RNAs were enriched using a column-based purification protocol (Protocol for isolation of low-molecular-weight RNA, RNA/DNA Kit, #14142; Qiagen Inc., Valencia, CA, USA) (Llave et al., 2002). Briefly, total RNA was mixed with QRL1 buffer along with β-mercaptoethanol (ME) and QRV2 buffer. After thorough mixing, the mixture was applied to the column that had been previously equilibrated with QRE buffer. After allowing the mixture to enter by gravity flow, the column was washed with QRW buffer and LMW RNA was eluted with QRW2 buffer. Low-molecular-weight RNAs were further purified with isopropanol precipitation, and the pellet was resuspended in nuclease-free water.
A precast gel system (15% TBE-Urea Criterion gels, #345-0091; Bio-Rad, Hercules, CA, USA) was employed to separate small RNAs for northern analyses. Both enriched LMW and total RNAs from various tissue samples were separated on the gels, transferred to nylon membranes and hybridized with different probes. RNA samples were mixed with RNA loading dye (Sigma-Aldrich) without ethidium bromide (EtBr), then heat-incubated and chilled on ice. After prerunning the precast gel for 30 min, samples were run at 150 V in 1× TBE buffer (Bio-Rad) until bromophenol blue dye was approx. 2.5 cm from the bottom of the gel. Gels were stained with EtBr for 5 min to visualize 5S rRNAs and tRNAs (Fig. 2), and were then destained before using the Semi-dry Trans-Blot SD (Bio-Rad) to transfer RNAs to Hybond-H+ nylon membrane (Amersham Biosciences, Piscataway, NJ, USA). After completion of the transfer protocol, the membranes were cross-linked and air-dried. Synthesized DNA oligomers (21 and 26 nucleotides) were loaded as size markers at 100 pmol. The probe sequences of the miR166 and miR167 oligonucleotides were 5′-GGGGAATGAAGCCTGGTCCGA-3′ and 5′-TAGATCATGCTGGCAGCTTCA-3′, respectively.
Detection of miRNAs on northern blots was performed using a North2South Chemiluminescent Hybridization and Detection kit (#17097; Pierce Biotechnology, Rockford, IL, USA). Briefly, probes were generated using a Biotin 3′-end DNA labeling kit (#89818; Pierce Biotechnology) to add a biotinylated dCTP to the 3′-end of DNA oligonucleotide complementary to several highly conserved miRNA sequences in Arabidopsis. Prehybridization and hybridization protocols were followed as described in the protocol (#17097; Pierce Biotechnology). Additional stringent washes were executed using 1× SSC (0.015 M trisodium citrate and 0.15 M sodium chloride, pH 7.0) and 0.1% sodium dodecyl sulphate (SDS) at 45°C, and the probes were targeted by Streptavidin–horseradish peroxidase (HRP) conjugate, producing light upon incubation with substrate buffers. The image was recorded using either X-ray films (Fujifilm, Stamford, CT, USA) or a luminescent image analysis system (LAS-1000; Fujifilm). In some cases, membranes were re-used after stripping the previous probes by incubating the membranes, twice, with 0.5% SDS solution at 60°C.
Poly(A) polymerase-mediated reverse transcription polymerase chain reaction for miRNA amplification and partial miRNA priming
Because of the limited amounts of tissues available, particularly at early developmental stages, we employed the approach of Shi & Chiang (2005) to detect the presence and level of miRNAs at different developmental stages of P. taeda during embryogenesis. This method is based on bacterial (Escherichia coli) poly(A) polymerase (PAP)-mediated reverse transcription polymerase chain reaction (RT-PCR). It is capable of detecting miRNAs with as little as 100 pg of total RNA and is able to distinguish miRNAs with 1 nucleotide difference. Total RNA, including miRNAs, was polyadenylated with E. coli PAP (#1350; Ambion, Austin, TX, USA) and reverse-transcribed with a poly(T) adapter (the 3′ SMART CDS primer in the SMART RACE cDNA Amplification Kit, #634914; Clontech Laboratories, Mountain View, CA, USA) into cDNAs for both semiquantitative RT-PCR (sqRT-PCR) and quantitative RT-PCR (qRT-PCR). Tests screened for the following five miRNAs: miR159, miR166, miR167, miR171 and miR172. The PCR primers were a miRNA-specific forward primer (Supplementary material Table S1) and a reverse primer with a sequence complementary to the poly(T) adapter segment (NUP, #634914; Clontech Laboratories). For sqRT-PCR, PCR products were collected every five cycles after the 15th cycle and run on gels. The qRT-PCR procedure is illustrated below.
The melting temperature (Tm) values of primers were determined experimentally from the thermal dissociation curves generated from the target primers and their corresponding antisense sequences (Supplementary material Table S1) using either the Bio-Rad iCycler iQ Real-Time PCR Detection System (Bio-Rad) or the ABI Prism 7000 Sequence Detection System (Applied Biosystems, Foster City, CA, USA). Briefly, 100 pmol of the primer and its antisense oligonucleotide were mixed with 12.5 µl of SYBR® Green PCR Master Mix (Bio-Rad or Eurogentec, San Diego, CA, USA) or Power SYBR Green Master Mix (Applied Biosystems) in a total volume of 25 µl with a programmed temperature ramping from 55 to 95°C to generate dissociation curves where the Tm was calculated.
For each PCR, 1 µl of template cDNA (equivalent to approx. 1 ng of total RNA) was mixed with 12.5 µl of 2× SYBR Green PCR master mix or Power SYBR Green Master Mix and 5 pmol each of the forward and reverse primers in a final volume of 25 µl. The amplification programs were the standard protocols of the cyclers – 15 s at 95°C and 1 min at 60°C for 40 cycles – and were followed by the thermal denaturing step to generate the dissociation curves to verify amplification specificity. All reactions were performed in triplicate, and the sizes of the PCR products were validated by electrophoresis on a 3% agarose gel and by cloning followed by sequence analyses.
Internal normalization of the reactions from different ZE and FG stages employed the DNA product amplified using a primer specific for P. taeda 5S rRNA. The quantity of miRNAs, relative to the reference gene (5S rRNA), was calculated using the formula , where ΔCT = (CT sample − CT Reference RNA) and sample is a specific miRNA. CT represents the threshold cycle. Comparison of miRNA expression at different stages was based on a comparative CT method (ΔΔCT). The relative miRNA expression can be quantified according to the formula , where ΔΔCT = (CT sample − CT reference RNA) − (CT calibrator − CT reference RNA) and sample is a specific miRNA. The product generated using the miR159 specific primer at stage 9.1 FG was used as a reference value and set to a value of 1 (Fig. 3). This calibrator sample enabled a comparison of all samples to a single standard. A calibrator was chosen for each data set and is indicated in the figure legend.
A method we call partial miRNA priming, which is a modification of a similar method described by Lim et al. (2003) to validate predicted miRNA sequences in Caenorhabditis elegans, was used to generate partial miRNA sequences from pine. Instead of using a cDNA library generated with small RNA as templates and determining the 5′ terminus of miRNAs, our template was cDNA derived directly from total RNA without additional size fractionation and purification. The PCR products synthesized in the reactions determined the final 6 or 10 nucleotides (3′-terminus) of loblolly pine miRNA sequence. The first 11 or 15 nucleotides of the Arabidopsis miRNA sequences (Supplementary material Table S1) were used as a forward primer and the NUP primer in the SMART RACE cDNA Amplification Kit (Clontech Laboratories, #634914) was used as a reverse primer.
A method we refer to as regional amplification quantitative RT-PCR (RA-PCR) was developed to monitor the miRNA-directed cleavage of mRNAs. Potential mRNA targets of miRNAs were determined by sequence analysis of The Institute for Genomic Research (TIGR) and Sanger databases (http://plantta.tigr.org/, http://compbio.dfci.harvard.edu/tgi/cgi-bin/tgi/gimain.pl?gudb=pine and http://microrna.sanger.ac.uk/), as shown in Table 1. A signature of an mRNA with a target site cleaved by a miRNA-guided process is a decrease in the RT-PCR amplification of any fragment that contains a region which is upstream of the target site. The reverse transcription of the miRNA-cleaved mRNA will not generate a cDNA beyond the cleaved site. Thus, for an mRNA regulated by miRNA-mediated cleavage, the abundance of a PCR-amplified segment of the cDNA that is upstream of the cleavage site can be expected to be less than a segment that is downstream of the cleavage site. The RA-PCR technique for monitoring miRNA-guided cleavage of mRNAs has an advantage of including control reactions in the same experiment. Other approaches, such as 5′ rapid amplification of cDNA ends (RACE), require separate reactions for controls and may be skewed by the presence of false positives. However, RA-PCR also may produce uncertainties in data interpretation as a result of the presence of additional primer-binding sites, self-priming caused by RNA secondary structures, or unusual patterns of mRNA degradation (Haddad et al., 2007).
Table 1. Five different orthologs of potential mRNA targets found in Pinus taeda (loblolly pine) expressed sequence tags (ESTs)1
Name of miRNA
Annotation potential target
Contig # LP EST database (TIGR)
Target sequences (5′→3′)
Mismatch b/w miRNA and target (nt)
Polymorphisms at target site b/w AT and LP (nt)
Accession # AT target genes (GenBank)
Potential microRNA (miRNA)-regulated loblolly pine (LP) mRNAs were searched using known Arabidopsis (AT) mRNA targets with sequences complementary to miRNAs. Annotated contigs were retrieved from the DFCI database (http://compbio.dfci.harvard.edu/tgi/cgi-bin/tgi/gimain.pl?gudb=pine) and compared with the sequences of miRNAs and target sites. Mismatches between Arabidopsis miRNAs and P. taeda target genes were counted and recorded as shown in lower case in the target sequences. Polymorphic bases were also identified at target sites between AT and LP, and accession numbers for the AT target genes, which were used for the search, were indicated.b/w, between; nt, nucleotide.
For each potential mRNA target, three sets of primers were designed to amplify three different regions: a 5′ region, a middle region and a 3′ region (Supplementary material Table S2). The middle region includes the target site. Primers for RA-PCR were designed using the beacon designer software (Bio-Rad) to produce fragments that would be amplified at similar rates by quantitative RT-PCR. The same protocols described above for the PAP-mediated quantitative RT-PCR of miRNA were used for RA-PCR.
The quantities of the three different regions, relative to the cDNA of the reference gene 18S rRNA, were calculated using the formula described above for miRNAs. The relative amount of DNA from each region was quantified according to the formula , where ΔΔCT = (CT sample − CT reference gene) − (CT calibrator − CT reference gene) and sample represents a particular region. The sample produced using the primers for the 3′ region at stage 4 of ZE or FG was selected as the calibrator.
Tissue-specific detection of miRNAs in P. taeda
Based on the results of Axtell & Bartel (2005), we screened RNA from several loblolly pine tissues using oligonucleotides complementary to the sequence of Arabidopsis miR166 and miR167. Small RNA hybridizing with the miR167 probe was clearly observed in needles, GS and FG (pooled stages 1–9.12) but was not detected in SEs from LSC at stages 1–3 (Fig. 2b). The miR166 probe detected a small pine RNA in SEs, but not in RNA isolated from needles (Fig. 2a). We were unable to examine ZEs for these miRNAs or to expand studies to FGs for miR166 because the quantities of tissues (harvested annually) were insufficient for these techniques. The sizes of the small pine RNAs are consistent with those expected for miRNA species and suggest expression of miRNAs in SE and FG tissues. The presence or absence of these putative miRNAs in different tissues also suggests that they are differentially expressed in the pine tissues examined.
Poly(A) polymerase-mediated miRNA detection
The PAP-mediated RT-PCR method was initially employed to quantify the two putative miRNAs detected by northern analyses as well as several other miRNAs in FG tissue from developmental stage 9.2. Figure 3(a) shows the results for miR166, miR167, miR172, miR159 and miR171 monitored in a semiquantitative manner. Samples were removed every five cycles following the 15th PCR cycle. All PCR reactions yielded a single band migrating between 70–100 bp. The quantity of PCR product, as judged by the intensity of bands in Fig. 3(a), suggests that different amounts of each of these pine miRNAs were present in the tissue examined.
Some of the products for miR166 and miR167 were cloned and sequenced. The results verified the expected structure and sequence of the PCR products, including the Pta-miR166 and Pta-miR167 sequences as well as a poly(A) tail with an anchor priming sequence (data not shown). Partial miRNA priming confirmed that the amplified PCR product was derived from pine miRNAs. Sequencing showed that the last 6 and 10 nucleotides of Pta-miR166 and Pta-miR167 are the same as in Arabidopsis miR166a and miR167d, respectively (data not shown, T. J. Oh et al., unpublished). The sequence for Pta-miR167 was not included in the report of Lu et al. (2007). Partial miRNA priming is useful for identifying the presence and partial sequence of a miRNA.
Figure 3(a) shows a PAP-mediated sqRT-PCR analysis of five miRNAs present in stage 9.2 FG from tree 7–56 (2003). Figure 3(b) shows the results from a PAP-mediated qRT-PCR analysis of the same five pine miRNAs found in stages 9.1 and 9.2 FG of the same tree. The relative amounts of the different miRNAs measured in Fig. 3(b) generally followed the hierarchy exhibited in the semiquantitative data in Fig. 3(a). In addition, Fig. 3(b) shows that the relative expression levels for these five miRNAs are similar in stages 9.1 and 9.2 FGs. Figure 3(c) shows the results from a second loblolly pine genotype, S4PT6 (2007), using the same approach as used for Fig. 3(b). The different miRNAs show similar patterns of relative abundance in FG tissues from both trees. We note, however, that expression levels of Pta-miR166 and Pta-miR167 are dramatically reduced for S4PT6 and the amount of miR171 is now lower than miR159.
Variation of miRNA expression in FG tissues during seed development was examined from stages 6 to 9.2 for the most highly expressed miRNAs: Pta-miR166 and Pta-miR167. Results from qRT-PCR are shown in Fig. 3(d). The average level of Pta-miR167 expression showed a slight decrease between stages 6 and 7 followed by a modest increase from stage 8 to stage 9.2. These changes are slightly greater than the measured standard deviation, but were not statistically significant. Stage-specific modulation is considerably larger for Pta-miR166. The level of this miRNA changed little between stage 6 (when cotyledon primordia begin to emerge) and stage 8 (where elongated cotyledons crown the SAM; Fig. 1b). However, it increased by more than three-fold at stage 9.1, when expanded cotyledons enclose the SAM. Differences were statistically significant (P = 0.05, analysis of variance (ANOVA) and multiple range test). The level of miRNA decreased, but remained at a relatively high level, in stage 9.2 FG. At this point the FG is sustaining a maturing embryo that continues cotyledon development and storage-product accumulation (Pullman & Webb, 1994). Figure 3(e) shows the results for a second tree, S4PT6 (2007). The dramatic increase observed in Fig. 3(d) for Pta-miR166 at stage 9.1 FG was diminished, but was still significantly greater (P = 0.05, ANOVA) than expression in stages 8 or 9.2 FGs. Pta-miR167 also revealed significant stage-specific modulation, reaching a maximum level in stage 9.1 FG. A cDNA sample generated by reverse transcription of the total RNA from stage 6 FG without prior treatment of PAP was used as the negative control. No amplification was recorded (data not shown).
A similar experimental approach was applied to investigate the presence and change of miRNA expression in zygotic embryos. This could not be carried out for the first tree, 7–56, but was performed using a limited series of developmental stages for the second tree, S4PT6. Figure 3(f) demonstrates that the five miRNAs studied in FG tissues are present in ZEs, but their expression patterns are noticeably different in the two tissue types (Fig. 3b,c). Although Pta-miR166 is the most abundant miRNA species in both FG and ZE tissues, Fig. 3(g) shows that stage-specific variation of Pt-miR166 and Pta-miR167 in ZEs is different from the patterns shown with FG tissues (Fig. 3e). The modest increase of Pta-miR166 expression in stage 9.1 ZE was statistically significant compared with stage 8 and remained elevated at stage 9.2. A 0.2-fold increase of miR167 expression was noted during embryo development in Fig. 3(g), but this should not be overemphasized because it was not statistically significant (Fig. 3f).
Monitoring expression of potential mRNA targets of miRNAs
Because the above analysis showed modest changes in Pta-miR167, whereas a larger stage-specific modulation was observed for Pta-miR166, we wished to determine whether their potential mRNA targets exhibited similar expression patterns over loblolly pine embryo development. In seeking potential miRNA-regulated pine mRNAs, we scanned pine mRNAs homologous to known Arabidopsis mRNA targets for sequences complementary to miRNAs, especially the 11 nucleotides of the 5′ side of miRNA sequences (positions 2 to 12). The 11 nucleotides of the 5′ side of miRNA (11 nucleotides of 3′ of the target site in mRNA) have been shown to be extremely well conserved for known miRNAs. We assumed that potential mRNA target sites in P. taeda differed by no more than two nucleotides from previously characterized mRNA targets in Arabidopsis (Schwab et al., 2006). An exception was made for a potential target of miR172 with six mismatches because an ortholog of an Arabidopsis target found and cloned in red pine was 95% identical with a P. taeda ortholog. The target sequences found in P. taeda (our research) and red pine (Axtell & Bartel, 2005) showed higher complementarity to the miR172 in Arabidopsis compared with its own mRNA target in Arabidopsis. Table 1 shows the different mRNA sites of five miRNAs that were included in the study. However, no mRNA target for miR159 was available at the time our research was carried out. Later, Lu et al. (2007) reported finding three contigs for miR159.
We reasoned that evidence of transcript cleavage around these target sites would strongly suggest miRNA-mediated mRNA regulation. A signature of an mRNA target site cleaved by a miRNA-guided process is a decrease in the RT-PCR amplification of a region flanking the target site (Navarro et al., 2006). After miRNA-induced cleavage, only the 3′ fragment of the target mRNA possesses a poly(A) tail. When poly(T) adapters are used to prime reverse transcription, only the 3′ end of the cleaved mRNA should be copied into cDNA. Intact mRNA molecules will be copied in their entirety. Previous studies of mRNAs with miRNA target sites suggest that the 5′ region of RNA (upstream) of the cleavage site is less stable than the 3′ region (downstream) of the cleavage site (Llave et al., 2002; Oh et al., 2006). One can generally expect the RT-PCR product from the upstream region to be present at levels no greater than the downstream region because reverse transcription with poly(T) adapters would not generate cDNA beyond a cleaved site.
Quantitative RT-PCR was employed to assess the abundance of three regions of several mRNAs. Figure 4(a) illustrates the relative positions of the three primer sets for a given mRNA for the RA-PCR method. The first set (F1, R1) amplifies a region upstream from the potential cleavage site. The second set (F2, R2) synthesizes a fragment that contains the target site and flanking regions, and the third set (F3, R3) amplifies a fragment downstream of the target site. The relative positions of these amplicons are illustrated for the actin2-like (ACT2L), auxin response factor8-like (ARF8L) and class III homeodomain-leucine zipper transcriptional factor AtHB15-like (HB15L) mRNAs (Fig. 4a). Figure 4(b) shows the results of the RA-PCR for the actin mRNA at seven stages of ZE development. This mRNA does not have a known miRNA target site and was used as a negative control for potential mRNA targets of miRNAs. The results indicate that the stability of the middle region is similar to that of the 3′ region for this mRNA. The 5′ region appears to decay faster than the 3′ region. This may be a result of preferential decay from the 5′ side (Murray & Schoenberg, 2007).
Expression of targeted mRNAs within embryos and FGs of developing seeds
In the foregoing experiments we obtained evidence for the presence of particular miRNAs in two different seed compartments: female gametophyte and zygotic embryo. The endosperm of angiosperms, the compartment corresponding to the FG of gymnosperm, feeds the embryo and the embryo signals its level of development to the endosperm to elicit a cascade of responses (Berger et al., 2006). The molecular basis of the gymnosperm embryo–FG signaling is not known. We wished to monitor the apparent miRNA-mediated regulation of the same target mRNAs in these different parts of the seed. We initially examined all four potential target mRNAs listed in Table 1 by using an end-point PCR method with three primer pairs for each mRNA. The amount of PCR products for the 5′, middle, and 3′ regions of the scarecrow-like (SCRL) and apetala2-like (AP2L) mRNAs were similar for all tissues and stages tested (data not shown). Differences were noted for these regions in the HB15L and ARF8L mRNAs, which merited more detailed analyses. Figure 5(a) shows the products of the RA-PCR reactions for the HB15L mRNA, a potential target of Pta-miR166, in stages 4 to 9.2 of ZEs. The corresponding Arabidopsis mRNA encodes a transcription factor regulating many aspects of meristem initiation, organ and vascular development and bilateral symmetry (AT1G52150; http://www.arabidopsis.org/). Unlike Fig. 4(b), where the amount of product of the target region was similar to or greater than that for the 3′ regions, the target region was decreased relative to the 3′ regions. The amount of product of the middle region was 0.28 to 0.6 of the amount observed for the 3′ region for stages 4 to 9.2, but similar to the amount observed for the 5′ region for all stages. This is consistent with miRNA-induced cleavage of this mRNA. Products of the RA-PCR reactions for HB15L mRNA from FG tissues are shown in Fig. 5(b). The middle region and the 5′ region products are 0.45 to 0.78 of the 3′ end product. Whereas miR166 was expressed at a three-fold higher level at stage 9.1 in FG (Fig. 3d), no corresponding reduction was observed in the target region of HB15L mRNA relative to its 3′ region product. This comparison will be considered in the Discussion.
Figure 5(c) shows the results from RA-PCR analysis of the three regions amplified from ARF8L mRNA from stage 4 to stage 9.2 ZE. The ratio of product of the middle region to the 3′ region varied from 0.18 to 0.37. This is again in agreement with miRNA-induced cleavage of this mRNA. Figure 5(d) shows the relative amounts of DNA products amplified from ARF8L mRNA from the corresponding stages of FG tissues. Here, the middle region was higher in amount than either the 3′ or 5′ DNA products. The stability of the three regions of ARF8L mRNA is apparently regulated in a different manner in FGs when compared with the ZEs at corresponding stages of development. If the decreased amount of target region in ZE stages is indeed caused by a miRNA, this result indicates that within the same seed, ARF8L is regulated by miRNA in the embryo but not in the FG. The Pta-miR167 itself is clearly present in FG tissue (Figs 2b, 3a–e), and thus one or more components involved in miRNA-induced cleavage may be missing or not functional in the FG.
Argonaute family genes in pine seed tissues
Members of the argonaute (AGO) gene family are important for plant development, the production of small RNA molecules, miRNA function and RNA silencing (Bonnet et al., 2006). Argonaute genes are themselves regulated by miRNA. Because the above results show a difference in the processing of the ARF8L mRNA between ZEs and FGs during development, we sought argonaute genes in loblolly pine with the objective of comparing their expression in these two tissues. Several loblolly pine EST libraries have been generated from various tissue types (Allona et al., 1998; Kirst et al., 2003; Pavy et al., 2005; Cairney et al., 2006; Lorenz et al., 2006). From these libraries, several similar clones, annotated as P. taeda argonaute-1-like (ptAGO1L) and P. taeda argonaute-9-like (ptAGO9L), were recovered and aligned with similar genes found in Arabidopsis, white spruce and human. Phylogenic analyses and alignment of these argonaute-like sequences characterized the probable function of P. taeda argonaute-like genes (Fig. 6a,b). Figure 6(a) shows that the contig ptAGO1L from the loblolly pine EST database has a strong similarity (85.6% identity, 92% similarity) to Arabidopsis AGO1 (AtAGO1). Another gene from the argonaute family, ptAGO9L, has a very strong similarity (94.6% identity, 97.6% similarity) to white spruce (Picea glauca) pgAGO and good similarity (60.6% identity, 78% similarity) to AtAGO9 in Arabidopsis (Tahir et al., 2006). Most AGO proteins have a conserved histidine at the equivalent of AtAGO1 position 798 that is required for mRNA cleavage (Baumberger & Baulcombe, 2005). This histidine is observed in ptAGO1L. The histidine at this position is replaced with a proline (P) in ptAGO9L, as observed with some argonaute genes in Arabidopsis. One may note from Fig. 6(a,b) that argonaute genes lacking histidine at position 798 are grouped at the top of the tree. The substitution is mainly with proline, but arginine (R) and serine (S) are also observed in AtAGO9 and AtAGO4, respectively. It was originally speculated that such proteins do not function as slicers, but a recent study of AtAGO4 shows involvement in RNA-directed DNA methylation (chromatin remodeling) and cleavage of target RNA transcripts for efficient siRNA production (Qi et al., 2006). AtAGO9 has also been speculated to be involved in similar processes as AtAGO1 in reproductive organs (Scutt et al., 2003).
Reverse transcription PCR of loblolly pine argonaute genes
To examine the expression of the argonaute genes in seed tissues, amplified DNA products of stage-specific and tissue-specific RT-PCR were investigated. Figure 7(a) shows EtBr-stained gels of amplified DNA products using two different sets of primers targeting both ptAGO9L and a normalizing control (albumins) (Supplementary Table S3). The albumin control is strongly expressed and remains relatively constant across the developmental stages 4 to 9.2 of ZEs and FGs. However, ptAGO9L expression does not remain constant in both tissues, but appears complementary. The ptAGO9L is expressed strongly early in development within the FGs, while at that time expression of ptAGO9L in the embryo is low. As seed development continues, expression of ptAGO9L gradually diminishes in the FGs, while it rises, essentially to the similar degree and at the same rate, in the ZE tissues. Unlike the ptAGO9L gene, the ptAGO1L gene shows similar expression at all stages (Fig. 7b). The data strongly suggest that expression of these genes, which are putatively involved in miRNA metabolism, are regulated differently in two compartments of the seed during development.
In this study, five different P. taeda miRNAs, corresponding in sequence to Arabidopsis miR159, miR166, miR167, miR171 and miR172, were detected in zygotic embryos and female gametophytes of seed tissues. This is the first report of miRNAs in a seed gametophytic tissue. Pta-miR166 was detected in SEs, and Pta-miR167 was found in FGs, needles and seedlings, of loblolly pine using northern blots. All five miRNAs were also detected in ZEs and FGs using PAP-mediated RT-PCR. Northern blots for Pta-miR167 showed that this miRNA was highly expressed in needle tissues and in germinating seedlings. The authors noted that the amounts of total RNA loaded were different – 130 µg and 70 µg for SEs and needles, respectively – and both blots were run with the same amounts and sources of total RNAs. Pta-miR167 signal was not observed in SE tissues when almost twice as much total RNA was employed (Fig. 2b). Pta-miR166 was not detected in needles but gave a signal in SE tissues using the same RNA sources probed for Pta-miR167. These observations suggest that these two miRNAs are differentially expressed in these P. taeda tissues. We cannot exclude the possibility that Pta-miR166 is also expressed in needles, but at levels below the detection limits of this study. It may also be worth noting that the total RNA isolated from FGs was obtained from a mixture of developmental stages 1 to 9.12.
Poly(A) polymerase-mediated qRT-PCR was used to investigate the expression of five different miRNAs in zygotic embryo and female gametophyte tissues. Female gametophyte tissue from two different trees showed relatively similar expression patterns for the miRNAs examined. Zygotic embryos and FGs showed very different miRNA expression patterns during seed development. The statistically significant increase in the level of Pta-miR166 in stage 9.1 FG (Fig. 3d) strongly suggests that this miRNA is triggering an event required for embryo and/or FG development. However, a corresponding stage-specific change was not observed in the degradation pattern of its potential mRNA target HB15L. The ratio of RA-PCR products for the target region of HB15L mRNA did not show significant modulation from stages 4 to 9.2 for either ZE or FG tissues. HB15L may be controlled by multiple factors. These factors may be supervening miRNA-mediated cleavage of this gene in pine embryos (Ohashi-Ito & Fukuda, 2003). Although an extensive embryo-expressed P. taeda EST database is available, HB15L mRNA was the only potential target we identified for Pta-miR166. A more comprehensive analysis of Pta-miR166 target mRNAs is needed to gain an understanding of the consequences of the rise in levels of Pta-miR166.
The potential mRNA target of Pta-miR167, ARF8L, was found in the P. taeda EST database based on the sequence of a 5′-RACE product from red pine (Axtell & Bartel, 2005). The sequences of ARF8L mRNA in loblolly pine and red pine are highly conserved (96% nucleotide identity). It was annotated as ARF8L after analysis using clustalw with other ARF genes in Arabidopsis. Our RA-PCR approach showed different degradation patterns for ARF8L mRNA in ZE vs FG. Significant reductions in the target region compared with the 3′ region were observed in ZE tissues, whereas in FG tissues this did not occur. Multiple restriction digestions of the three DNA fragments confirmed the homogeneity of the cDNA sequence in both tissues. These results clearly show that ARF8L mRNA stability is regulated in a different manner in ZE and FG tissues. The observations are consistent with Pta-miR167-mediated cleavage of ARF8L mRNA in ZEs but not in FGs; however, additional studies are needed to confirm this interpretation.
One aspect of the regional amplification results of ARF8L mRNA in ZEs does not appear to be consistent with a simplified picture of miRNA-mediated cleavage and mRNA stability. If miRNA-induced cleavage occurs to a messenger RNA, mRNA fragments upstream of the cleavage site will no longer have the poly(A) tail and would not be subject to reverse transcription using the oligoT-anchored primer. Thus, the amount of 5′ region amplified by qRT-PCR should be less than or equal to the amount amplified from the target region. Unlike the results shown in Figs 4(b) and 5(a,b), Fig. 5(c) shows that the amount of 5′ region product is higher than the amount of target region product. Unknown rates of 5′ and 3′ degradation, or aberrant secondary structures introduced by miRNA-guided cleavage of this specific mRNA, may cause the RA-PCR data to vary from expectations. This observation suggests caution in interpreting the RA-PCR results for the ARF8L mRNA in terms of the miRNA-mediated degradation. We are currently exploring further for evidence of self-priming caused by RNA secondary structures or unusual degradation patterns of this mRNA (Haddad et al., 2007).
Two homologs of argonaute family genes were profiled as a function of developmental stage in embryo and female gametophyte tissues. Other studies of argonaute genes have shown that they are involved in many different functions of cell metabolism such as regulation of gene expression with small RNAs including miRNAs, schematic silencing of genes on the chromosomes and defense mechanisms against invasion of foreign agents. The mutation studies of Baumberger & Baulcombe (2005) indicate that the catalytic Asp-Asp-His (DDH) triad and additional histidine of the PIWI domain are necessary to function as an RNA slicer. In addition, they illustrated AtAGO4, AtAGO6, AtAGO8 and AtAGO9 as producing proteins having a different amino acid residue when aligned with AGO1 position 798. They therefore suggest that AtAGO4, AtAGO6, AtAGO8 and AtAGO9 do not produce active slicer proteins. However, Qi et al. (2006) tested this speculation and they have shown that AtAGO4 with serine instead of histidine at position 798 along with the triad, still has slicer activity. This suggests that ptAGO9L with proline at position 798 along with the conserved triad may still function as a slicer, and possibly this gene is a tissue-specific functional homolog of a miRNA/siRNA carrier (with a PAZ domain) and the slicer for reproductive tissue, as suggested by Scutt et al (2003). Our observation of inverted or complementary expression of ptAGO9L in ZE and FG tissues may indicate cross-talk between tissues through uses of small RNAs.
While a number of questions remain to be answered regarding targets and function of miRNAs in P. taeda embryos, the results from this study demonstrate stage-specific and tissue-specific modulation for some miRNAs and miRNA processing components in seed tissues. Given the importance of miRNAs to general developmental processes in plants, it is likely that the miRNAs identified in the FGs play a role in embryogenesis. Our observation of differential Pta-miR167-mediated cleavage in FG vs ZE tissues suggest that miRNAs expressed in FG may be involved in regulating embryo development in loblolly pine. An understanding of the role of miRNAs in FGs on the regulation of embryo development may lead to advancements in clonal technology for improved sustainable forestry.
We thank the member companies of IPST at Georgia Tech for financial support and Temple Inland, Texas Forest Service, and Weyerhaeuser Company for cones. We are grateful for the help of J. Grabowski, D. Gluck, B. Lee, S. Johnson and A. Skryabina for the collection of SE, ZE and FG.