Transcription in eukaryotic cells occurs in a stochastic [1
] and pulse manner [4
], which may generate transcriptional noise. Such transcriptional noise can be a phenomenon that concerns all expressed transcripts or can be limited only to some specific transcripts. In some cases, this phenomenon is beneficial for the cells, but usually, it is undesirable and must be suppressed [7
]. It was examined that gene-specific transcriptional noise is responsible for the phenotypic variability of isogenic cell populations living in homogeneous environments. In this situation, the variance of expression is a potential target of adaptation [8
]. Positive aspects of this process also occur in clonal populations of bacteria that can diversify cell fates. Transcriptional noise can also be enhanced to achieve competence in Bacillus subtilis
In contrast, transcriptional noise, which is caused by pulsed transcription, must constantly be buffering to allow the proper functioning of the cells [7
].It is diminished or cancelled out at the post-transcriptional level. The transcriptional burst can be regulated by elements such as changes in chromatin condensation, promoter architecture, or enhancer-promoter interactions [7
Periods in which a burst of transcriptional activity occurs are separated by phases with no or low transcription [6
]. The result is an average constant level of synthesis of specific proteins based on the mRNA pool constantly available in the cytoplasm. One of the main processes that influences the reduction of transcriptional noise is the process of pre-mRNAs and mRNAs retention in the nucleus. Due to this process, there is no simultaneous export of the whole pool of synthetized mRNAs to the cytoplasm. This prevents transcriptional chaos and significant fluctuations in the level of proteins in the cytoplasm. It was examined that post-transcriptional and translational auto-regulatory motifs are more effective at noise silencing than common negative-feedback [10
]. The role of mRNAs retention in the cell nucleus in the process of post-transcriptional reduction of transcriptional noise was discovered in the last decade and is now extensively studied. It appears to be common in all eukaryotes, both unicellular, such as yeasts [11
] and multicellular plants [12
] as well as animals [14
Recent studies suggested another role of nuclear mRNAs retention, namely strict temporal control of specific mRNAs export to cytoplasm [11
]. Furthermore, regulation of gene expression by nuclear mRNA retention occurs in mammalian metabolic tissues both under stress [15
] and normal physiological conditions [14
]. It was also found in generative cells that require the expression of specific proteins at a time strictly defined by their developmental stage [11
]. Retained transcripts are localized in the nucleoplasm, nuclear basket, and nuclear domains such as nuclear speckles, paraspeckles, Cajal bodies, and nuclear foci (“dots”) [13
]. mRNA retention may be due to an unspliced intron, a poly(A) tail with an altered length, cis-elements, or convert adenosine to inosine [13
]. Transcripts accumulated in nuclear foci in S. cerevisiae
have a hippo- or hyperacetylated poly(A) tail [19
]. Inhibited splicing via the attachment of spliceostatin A to the SF3b sub-complex in U2 snRNP causes the retention of transcripts with unspliced intron and its accumulation in the nuclear domains [20
]. Early spliceosome elements such as U1 snRNP and U2 snRNP are usually attached to the unspliced intron, resulting in the accumulation of these transcripts in nuclear speckles [18
]. mRNAs remain in the nucleus for a long time, and their export to the cytoplasm is slow. This slow export may be caused by a nuclear zip code, an RNA segment in ORF in mRNA [22
]. Removal of retention inducers results in rapid nuclear RNA export and translation [19
]. The increased speed of nuclear export may be caused by modifications of mRNA, e.g., N6-methyladenosine (m6A) and N5-methylcytosine [18
].In mammalian cells that show a high level of metabolic activity (liver, gut, and beta cells), transcripts are stored for several hours before being transported to the cytoplasm [13
]. In mouse liver cells, the sequence coding for mCAT2 protein is “enclosed” in a long non-coding RNA (CTN-RNA) [16
]. Stress causes cleavage in the 3′ UTR region and export of mRNA to the cytoplasm and, consequently, protein synthesis. As a result, the cells are able to respond very quickly to stressful conditions. Similarly, in plants under stress conditions such as hypoxia shock, mRNAs are temporarily retained in the nucleus and are released into the cytoplasm only when hypoxia subsides, when cells return to normal metabolism. Then, there is a rapid transport of mRNA to the cytoplasm and protein synthesis [24
]. The molecular basis of this phenomenon has not yet been understood.
In generative cells, the process of nuclear mRNAs retention appears to be associated with periods of genome silencing. Such a phenomenon was observed during meiosis in yeast [11
], spermatogenesis in animals [14
], and during meiosis and the development of microsporocytes in plants [12
]. In all these cases the mechanism involves intron retention. In pre-mRNAs usually only one or two introns are left and then post-transcriptionally spliced when a cell needs mature mRNAs.
Previously, we demonstrated that during meiosis in larch microsporocytes, regulation of gene expression by nuclear mRNA retention occurred. During the first meiotic prophase we observed 5 cycles of transcriptional activity and 4 periods of transcriptional silencing related to chromatin contraction during the diplotene stage (lasting about 5 months) [25
]. We observed accumulation of mRNAs in the nucleoplasm and Cajal bodies (Cbs) in each of these cycles [24
]. The site of retention of pre-mRNAs with conserved intron sequences are CBs, and the introns are removed before mRNAs are exported to the cytoplasm.
In this study, we analysed the duration of a single cycle synthesis and flow of RNA poly (A) in microsporocytes. We also performed an analysis of the transcriptome of these cells. Based on the analysis, mRNAs related to various metabolic processes of these highly active generative cells were selected. We present different expression profiles for these mRNAs: patterns of nuclear retention and transport to the cytoplasm. We also try to explain the physiological sense of RNA pol II subunit 10 transcript retention and delayed expression during the subsequent stages of meiotic prophase.
The innovation of this study is related to the involvement of the CBs in the regulation of gene expression viamRNA retention. This phenomenon has not yet been described in any other model. Our research has also shown the existence of different patterns of mRNA nuclear retention and export. The results of conducted analyses can help better understand the processes involved in the maturation of germ cells in plants and post-transcriptional regulation of mRNA expression in eukaryotic cells.
Our previous studies have shown that mRNA synthesis in larch microsporocytes occurs in a pulsatile manner. Certain mRNAs remain in the nucleus for a long time after synthesis. They are then successively transported to the cytoplasm where they are translated [13
]. In previous studies, we have analyzed a single cycle of cellular synthesis and turnover of poly(A) RNAs, which allows us to distinguish successive stages characterized by changes in the level and distribution of poly(A) RNA [13
]. In the present study, we focused on determining the developmental synchronicity of microsporocyte development during meiosis and on the exact duration of each stage of the cycle. We observed that larch microsporocytes were developmentally synchronous both within a single flower bud (93.1%) (Table S1
, Figure S1
) and buds from a single branch (92.9%) (Table S2
, Figure S2
). Therefore, to determine how long each stage lasts, individual flower buds were harvested every day, protoplasts were isolated, and poly(A) RNA fluorescent in situ hybridization (FISH) was performed to identify the specific stage. A single cycle lasted in the range of10–11 days, of which the longest stage, lasting 8–9 days, was the first one, in which poly(A) RNAs were synthesized and retained in the cell nucleus (Figure 1
The export of poly(A) RNA to the cytoplasm starts at the beginning of the second stage and then proceeds very quickly. The next steps last only a few hours each. This indicates that when expression of mRNA starts to be needed, the next steps take place very quickly [13
Our research so far has mainly shown the fate of the general pool of polyadenylated transcripts during a single cycle of mRNA synthesis and turnover. Here, we made an attempt to check how such a cycle proceeds for individual mRNAs. To get information on which mRNAs were synthesized in a single cycle, we sequenced, assembled, and analyzed the transcriptome of larch microsporocytes. We obtained 220,400 transcript sequences over 150 nt in length. To assign putative functions, potential homologs were identified by searching NCBI nr or ntdatabases using BLASTX (for potential mRNA sequences) or BLASTN (for other sequences), respectively. GO-terms assignment was performed for protein coding sequences, for which 1024 transcripts with the highest expression over the studied period were selected. 214,112 GO terms were assigned to 426 sequences (Supplemental Dataset 1
). In the “biological process” domain at the sixth level (chosen to show detailed categories) the majority of transcripts were assigned terms from categories related to nucleic acids metabolism (Figure 2
We chose mRNAs involved in the most important biological processes (identified based on GO categories and enumerated below) to see how they were expressed relative to the total poly(A) RNA pool. We tracked thirteen mRNAs, encoding the following proteins: (1) RNA metabolism: SNRP27—homolog of small nuclear ribonucleoprotein 27, RPB10—polymerase II RNA subunit, PABP4—poly(A) tail 4 binding protein, SERRATE—involved in alternative splicing and involved in the microRNA biogenesis pathway, (2) translation related: RS6—S6 small ribosomal subunit protein, eF1a—translation elongation factor 1a, eIF5b—translation initiation factor 5b, (3) associated with the cytoskeleton: ACT—actin, CLATH—clathrin, (4) organelle organization related: PER40—peroxidase 40, SDH—succinate dehydrogenase, PEP—plastid RNA polymerase, (5) stimulus response related: DNJ—protein with DNaJ domain.
Based on the time spent in the nucleus and the initial moment of export, we identified two main patterns of mRNA synthesis and export.
The first of the patterns was characterized by a long retention time, until the third/fourth stage of the cycle and export beginning in the first/second stage. This category included mRNAs encoding: SNRP27, PER 40, ACT, DNJ, PEP, PABP4, RPB10 (Figure 3
and Figure S3
The second pattern was characterized by a shorter retention period (until second/third stage) and later beginning of export (second stage). This expression profile was characteristic of mRNAs coding for translation factors: RS6, eF1a, eIF5b, Serrate (Figure 4
and Figure 6, Figure S4
). In both patterns the cytoplasmic concentration of a given mRNA increased over time until the very end of the cycle.
In addition to these two dominant patterns, we also observed others deviating from them, which suggested the possibility of precise adaptation of the system to the demand for a specific protein at the right time (Figures S5 and S6
Our previous studies have shown that, in the case of mRNAs encoding Sm proteins, there is nuclear retention of not fully spliced transcripts with one or more introns unspliced. We performed a detailed analysis of the mRNA encoding the RNA polymerase II subunit 10, the key protein for the expression of protein-encoding genes, and checked if it was subject to the same type of regulation. As mentioned before, this mRNA belongs to the group displaying the first synthesis and export pattern. The transcript with an unspliced second intron underwent nuclear retention (Figure 5
Unspliced pre-mRNAs are stored in nucleoplasm and, in particular, in Cajal bodies for a long period of time. Fully mature transcripts (after splicing of the second intron) appeared in the nucleus in the second stage of the cycle (Figure 5
B). Concentration of the transcript at the periphery of the nucleus was visible, which suggested preparation for export to the cytoplasm. In subsequent stages, the amount of RPB10 transcript in the cytoplasm increased until the fifth stage, when the signal was the strongest.
The appearance of a large amount of the transcript in the cytoplasm correlated with the appearance of a large pool of non-phosphorylated form of the protein in this compartment (Figure 5
C). These data suggested that this was the point at which the individual subunits were translated and the ready-to-use enzymes were assembled.
A similar pattern was observed with the expression of a SERRATE protein. mRNA encoding this protein represents the second pattern of synthesis and export. Mass export of mRNA encoding the SERRATE protein to the cytoplasm in the 3rd and 4th stages of the cycle (Figure 6
A) preceded the appearance of the SERRATE protein in the cytoplasm at the end of the cycle (Figure 6
4. Materials and Methods
4.1. Plant Materials and Isolation of Meiotic Protoplasts
Anthers of European larch (Larix decidua
Mill.) was collected from the same tree (Toruń, Poland, coordinates 53.021155 N, 18.570384 E) in the diplotene stage of prophase of larch microsporocytes from December to January at weekly intervals. They were fixed in 4% paraformaldehyde solution in phosphate-buffered saline (PBS) buffer pH 7.2 for 12 h and squashed to obtain free meiocytes. Meiotic protoplasts were isolated from these cells according to the method of Kołowerzo et al. [32
]. Isolated protoplasts were next used for fluorescent in situ hybridization (FISH).
4.2. FISH in Multiplex Reactions
Prior to the assay, the cells were treated with 0.1% Triton X-100 in PBS for 10 min. For the hybridization, the probes were resuspended in hybridization buffer (30% v/v
formamide, 4× SSC, 5× Denhardt’s buffer, 1 mM EDTA, 50 mM phosphate buffer) at a concentration of 50 pmol/mL. Hybridization was performed overnight at 27 °C. The sequences of all oligo probes used in this work are summarized in Table 1
. The antisense DNA oligonucleotides (Genomed, Warsaw, Poland) were dissolved in the hybridization buffer at a concentration of 10 pmol/mL.Digoxygenine (DIG) probes were detected after hybridization using rabbit anti-DIG antibody (Thermo Fisher Scientific, Waltham, MA, USA) diluted in 0.01% acetylated BSA in PBS (1:100) in a humidified chamber kept at 11 °C overnight and secondary anti-rabbit antibody labelled with Alexa 488 (Invitrogen) diluted in 0.01% acetylated BSA in PBS (1:100) in a humidified chamber kept at 33 °C for 1.5 h. After labelling, the slides were stained for DNA detection with Hoestch (1:1000) and mounted in ProLong Gold Antifade reagent (Life Technologies, Carlsbad, CA, USA).
4.3. Immunodetection of Non-Phosphorylated RNA Pol II and SERRATE Proteins
Prior to the assay, the cells were treated with 0.1% Triton X-100 in PBS for 25 min to induce cell membrane permeabilization. Nonspecific antigens were blocked with PBS buffer containing 2% acetylated BSA for 15 min. Non-phosphorylated CTD domain of RNA pol II and SERRATE proteins was detected by incubating respectively with rat anti-RNA pol II (Chromotek, Planegg-Martinsried, Germany; 1:100) and rabbit antibodies (Agrisera, Vännäs, Sweden; 1:250) in 0.01% acBSA in PBS, pH 7.2 (overnight incubations). After rinsing with PBS, the samples were incubated for 1 h at 37°C with the following secondary antibodies: anti-rat Alexa 488 (Thermo Fisher Scientific, Waltham, MA, USA) or anti-rabbit Alexa 488 (Thermo Fisher Scientific, Waltham, MA, USA) 1:500 in 0.01% acBSA in PBS, pH 7.2.
4.4. Design of Multi-Labelling Immunofluorescence-FISH Reactions
Doublelabellingof immunofluorescence-FISH reactions (non-phosphorylated RNA pol II proteins + poly(A) RNA + U2 snRNA; SERRATE proteins + poly(A) RNA + U2 snRNA) wasperformed as described above. In the doublelabelling of immunofluorescence-FISH reactions, the immunocytochemical methods always preceded the in situ hybridization methods because when in situ hybridization was applied first, the subsequent levels of immunofluorescence signals were very weak.
4.5. Confocal Microscopy
The images were captured with a Leica TCS SP8 or Olympus FV3000 confocal microscope. The optimized pinhole, long exposure (400Hz), and 63X (numerical aperture 1.4) Plan Apochromat DIC H oil immersion lens were used. The images were collected sequentially in the blue (Hoechst 33342), green (Alexa 488), red (Cy3), and far red (Alexa 633, Cy5) channels. To minimize bleed-through between the channels, we employed low laser power (1.5% of maximum power) and sequential collection. For all antigens and developmental stages, the obtained data were corrected for background autofluorescence, as determined by negative control signal intensities. ImageJ [33
] was used for image processing and analysis. Each reaction step was performed using consistent temperatures, incubation times and concentrations of probes and antibodies.
4.6. Protoplasts of a Single Bud and Single Branch, Analysis of Synchronicity and Duration of Stages
To investigate the developmental synchronicity of larch microsporocytes within one flower bud, protoplasts were prepared in accordance with the procedure described earlier. In this case, only one bud was used. Cells were resuspended in 80 μL of PBS buffer. The entire volume of the suspension was placed on two slides (40 μL for each slide). To investigate the developmental synchronicity of larch microsporocytes within one branch; protoplasts were prepared in accordance with the procedure described earlier. In this case, the cells were isolated from 10 flower buds coming from one branch. Cells were resuspended in 40 μL of PBS buffer. The entire volume was placed on one slide. In both cases on such prepared protoplasts, double-labelling was performed. The hybridisation reaction was carried out in accordance with the previously described procedure. Directly labelled probes were used: poly(A) RNA probe with Cy3 and U2 snRNA probe with Cy5. Images were collected using the Leica Sp8 confocal microscope.
The collection of data was carried out by counting the cells at different stages of the poly(A) RNA cycle. To ensure reliable results, multiple counting of one cell was prevented by marking counted ones on a slide map.
4.7. Libraries Preparation and Sequencing
RNA from microsporocytes was isolated with aPicoPure RNA Isolation Kit according to the manufacturer’s protocol. After qualitative (Bioanalyzer 2100; Agilent Technologies, Santa Clara, CA, USA) and quantitative (NanoDrop 2000, Qubit; Thermo Fisher Scientific, Waltham, MA, USA) evaluation, the RNA was used for library preparation. Libraries wereprepared with the use of a Total RNA Seq Library Preparation Kit (Lexogen, Vienna, Austria) according to the manufacturer’s protocol, but with displacement stop primers (DSPs) concentration decreased by half (0.5 µL per reaction), which resulted in longer library fragments. Quantification of the libraries was done viaqPCR with a KAPA Library Quantification Kit for Illumina (KAPA Biosystems, Wilmington, MA, USA). The analyses were performed in accordance with the manufacturer’s protocol. Next the libraries were sequenced on MiSeq (Illumina, San Diego, CA, USA) with MiSeq Reagent Kit v3 (600 cycles, Illumina, San Diego, CA, USA).
4.8. Bioinformatic Analysis
The reads were filtered in terms of quality via PRINSEQ-lite [34
]. The reads from all the libraries (we have created libraries for 16 biological repetitions) were pooled and assembled with Trinity [35
]. Next the sequence of transcripts wasanalysed with BLASTX (for putative mRNA) and BLASTN (for other RNA) with the use of the NCBI databases. Putative functions and ontologies were found using Blast2GO [36
] based on BLAST results.