Plant, Cell and Environment (2011) 34, 332–346
doi: 10.1111/j.1365-3040.2010.02247.x
Gene expression changes during short day induced terminal bud formation in Norway spruce pce_2247
332..346
DANIEL K. A. ASANTE1,5, IGOR A. YAKOVLEV2, CARL GUNNAR FOSSDAL2, ANNA HOLEFORS3,4, LARS OPSETH3, JORUNN E. OLSEN3, OLAVI JUNTTILA1 & ØYSTEIN JOHNSEN3 1 Department of Arctic and Marine Biology, University of Tromsø, N-9037, Tromsø, Norway, 2The Norwegian Forest and Landscape Institute, Høgskoleveien 8, N-1432, Ås, Norway, 3Department of Plant and Environmental Sciences, Norwegian University of Life Sciences, N-1432, Ås, Norway, 4Swedish University of Agricultural Sciences (SLU), Department of Plant breeding and Biotechnology, SE-750 07, Uppsala, Sweden and 5School of Biological Sciences, Department of Molecular Biology and Biotechnology, University of Cape Coast, Cape Coast, Ghana
ABSTRACT The molecular basis for terminal bud formation in autumn is not well understood in conifers. By combining suppression subtractive hybridization and monitoring of gene expression by qRT-PCR analysis, we aimed to identify genes involved in photoperiodic control of growth cessation and bud set in Norway spruce. Close to 1400 ESTs were generated and their functional distribution differed between short day (SD-12 h photoperiod) and long day (LD-24 h photoperiod) libraries. Many genes with putative roles in protection against stress appeared differentially regulated under SD and LD, and also differed in transcript levels between 6 and 20 SDs. Of these, PaTFL1(TERMINAL FLOWER LIKE 1) showed strongly increased transcript levels at 6 SDs. PaCCCH(CCCH-TYPE ZINC FINGER) and PaCBF2& 3(C-REPEAT BINDING FACTOR 2&3) showed a later response at 20 SDs, with increased and decreased transcript levels, respectively. For rhythmically expressed genes such as CBFs, such differences might represent a phase shift in peak expression, but might also suggest a putative role in response to SD. Multivariate analyses revealed strong differences in gene expression between LD, 6 SD and 20 SD. The robustness of the gene expression patterns was verified in 6 families differing in bud-set timing under natural light with gradually decreasing photoperiod. Key-words: bud set; dormancy; gene expression; Picea abies (L.) Karst.; qRT–PCR; subtracted libraries.
INTRODUCTION A major selection pressure on woody perennials of the temperate and boreal zones is to control timing of growth cessation and bud set in order to survive winter conditions. The ability of the trees to survive freezing temperatures in winter depends on mechanisms leading to entry into a state of dormancy and development of cold hardiness in the fall. Regulation of tree growth in this way is controlled Correspondence: I. A. Yakovlev. Fax: +47 64948001; e-mail:
[email protected] 332
genetically, in response to environmental cues, such as daylength, light quality and/or temperature that precede the onset of harsh winter conditions. Trees are thus able to perceive environmental variations and modulate their growth and development in synchrony with the changing seasons. The order of developmental events leading to establishment of dormancy and cold hardiness, also called inwintering processes, begins with cessation of elongation growth. This is followed by bud set or, in some cases, by shoot tip abscission, leaf senescence and induction of dormancy. The key role of apical growth cessation as the initial process leading to the inwintering events was provided by studies of hybrid aspen (Populus tremula ¥ tremuloides). Short days (SD) induce growth cessation and bud set, and subsequent exposure to low temperature induces senescence and abscission of leaves and development of cold hardiness (Olsen et al. 1997a). There are exceptions to these generalizations, however. For example, in apple (Malus pumila) and pear (Pyrus communis) low temperature, and not photoperiod, controls growth cessation and dormancy induction (Heide & Prestrud 2005). Also, in Norway spruce (Picea abies), bay willow (Salix pentandra) and hybrid aspen, the photoperiod control can be bypassed, and growth cessation can be induced even in long photoperiods through a combination of high day and low night temperature (Heide, 1974; Junttila 1980; Mølmann et al. 2005). Generally, photoperiod has a major role in initiation of inwintering events in trees with seasonally indeterminate growth patterns, such as Betula, Populus, Salix and young plants of Picea (Thomas & Vince-Prue, 1997; Nitsch, 1957). The various physiological processes which are initiated by cessation of apical elongation growth have different regulatory mechanisms;, e.g. the mechanisms for induction of growth cessation are not the same as those for the development of dormancy and cold hardiness (Mølmann et al. 2005). However, since light and temperature climates are important environmental cues in all processes, and the annual growth cycle in trees is a continuum, there is certainly an overlap between the mechanisms for the various © 2010 Blackwell Publishing Ltd
SD induces gene expression in spruce 333 processes. These responses involve environmentally dependent modulations of the expression of specific genes and enzyme activities. Phytochromes are known to be involved in SD induced growth cessation and bud set in Populus (Howe et al. 1996; Olsen et al. 1997a). The PhyB2 gene has been mapped to a linkage group containing QTL for bud set and bud flush (Frewen et al. 2000; Chen, Howe & Bradshaw 2002). In Populus, a CONSTANS (CO)/FLOWERING LOCUS T (FT) regulatory module has been shown to be involved in photoperiodic control of shoot elongation (Böhlenius et al. 2006). In Norway spruce, two CO-like genes, showing decreased expression under SD were recently identified, but on the ′basis of their expression pattern, their function appears to differ from that of Populus CO (Holefors et al. 2009). Also, transcript levels of a Norway spruce FT homolog with high similarity to the Arabidopsis FT and TERMINAL FLOWER 1 (TFL1) gene was recently reported to increase strongly under SD (Gyllenstrand et al. 2007). Furthermore, it has been reported that photoperiod, acting through the phytochrome system, interacts with biosynthesis of plant hormones in woody species. SDs are detected by phytochrome and induce growth cessation through reductions in gibberellins (GAs) and indole acetic acid (IAA) in Populus and Salix (Olsen et al. 1995a, b; Olsen et al. 1997a, b). Also, abscisic acid (ABA) appears to be involved in the development of dormancy in trees (Quamaruddin et al. 1993; Welling, Kaikuranta & Rinne 1997; Rohde et al. 2002). Ruttink et al. (2007) reported a profound change in gene expression in Populus within the first 2 weeks after transfer from long day (LD) to SD, including light and ethylene signal transduction components. A second major transition was observed from 3 to 4 weeks of SD coinciding with terminal bud formation. Cell proliferation genes were then downregulated, and genes of the ABA pathway were upregulated. From 5 to 6 weeks of SD when dormancy was established, only 5 genes out of 945 showed a major change in expression (Ruttink et al. 2007). The environmental conditions during the initial phases of the inwintering processes in woody perennials affect the rate of development and depth of dormancy (Arora, Rowland & Tanino 2003; Junttila 2007; Søgaard et al. 2008; Olsen 2010). Thus, understanding the control of the initial stages of inwintering appears to be a key to understanding
dormancy. The findings of Ruttink et al. (2007) are consistent with this approach, since the final stages of bud development and dormancy establishment do not appear to be associated with gross transcriptional regulation, but rather with long-term fixation of the expressional state of critical components. Investigations by Holliday et al. (2008) on gene expression in field-grown Sitka spruce (P. sitchensis) during autumn provided many candidate genes for molecular dissection of the processes occurring in autumn. However, it was difficult to separate genes involved in growth cessation and dormancy induction from those associated with development of cold hardiness (Holliday et al. 2008). Thus, studies of SD-induced bud formation and dormancy induction under controlled conditions are also needed for an increased understanding of these processes. In this study, we aimed to identify Norway spruce genes that show changed expression after SD treatment as compared to LD. By combining suppression subtractive hybridization (SSH) and monitoring gene expression by qRT-PCR analysis, we successfully identified candidate genes to be targeted in further studies of the dormancyrelated processes. Strong differences in gene expression between LD (24 h photoperiod) and SD (12 h photoperiod) were observed and the robustness of these expression changes was verified in 6 different full-sib families differing in their bud-set timing under natural light conditions. Only PaSEP (SEPALLATA) was found to differ between families. Thus, in all these families, a photoperiod as short as 12 h results in similar changes in gene expression in preparation for bud set.
MATERIALS AND METHODS Plant materials, growth conditions and sample collection To investigate the effect of photoperiod on gene expression and to test the robustness of any expression differences, we used six full-sib families of Norway spruce (Table 1) with known timing of bud set (Johnsen et al. 2005a). To generate these materials, Norway spruce family material was crossed in 2004 at Biri nursery and Seed Improvement Center (Norway – 60.9°N) and selected in 2005 after performance
Table 1. Origin of parents of the studied Norway spruce full sib families and the proportions of plants with a visible terminal bud at three different dates in August 2005, including standard error of mean (SE), observed on seedlings under natural day length with gradually decreasing photoperiod in a greenhouse (59.7°N latitude) Terminal bud set (%) and (SE) Full sib family number (씸 ¥ 씹)
Female origin (씹, latitude, North/altitude, m)
Male origin (씸, latitude, North / altitude, m)
12.08.05
18.08 05
22.08.05
1 (5433 ¥ 7292) 2 (1641 ¥ 1960) 3 (2474 ¥ 7292) 4 (6172 ¥ 1957) 5 (6170 ¥ 1960) 6 (1957 ¥ 7436)
60.2° / 320 60.0° / 230 59.8° / 415 65.9° / 125 66.0° / 80 60.7° / 200
60.1° / 235 60.7° / 160 60.1° / 235 60.7° / 200 60.7° / 160 60.2° / 330
50 16 9 34 51 3
79 53 50 68 86 17
97 96 97 99 99 84
© 2010 Blackwell Publishing Ltd, Plant, Cell and Environment, 34, 332–346
(3.3) (3.3) (3.3) (3.3) (3.3) (3.3)
(4.6) (4.6) (4.6) (4.6) (4.6) (4.6)
(3.4) (3.4) (3.4) (3.4) (3.4) (3.4)
334 D. K. A. Asante et al. evaluation in a greenhouse at the Norwegian Forest and Landscape Institute experimental farm (Hoxmark, Norway – 59.7°N) under natural daylight conditions with gradually decreasing photoperiod (see Johnsen et al. 2005a for full details of growing conditions). Recording of terminal bud formation in the six different full-sib families was done during August 2005. The proportions of plants with a visible terminal bud are shown for 12, 18 and 22 August (Table 1). The difference between families in terminal bud formation was highly significant (P < 0.0001), both for original proportion data and for arcsine transformations of the data. The timing of bud set under natural conditions was earliest in families 1, 4 and 5, intermediate in families 2 and 3, and latest in family 6. In light of these differences, we also aimed to test whether these families differed in gene expression in response to SD. For the gene expression studies, seeds from these preselected families were germinated and seedlings raised at 22 °C under continuous light (24 h photoperiod; LD) from fluorescent tubes (Osram 36W20, Osram, Munich) in growth chambers, as previously described (Kohmann & Johnsen 1994). Very long photoperiods up to 24 h correspond to the natural situation during at least part of the summer at high northern latitudes such as those where the family materials originate. Thus, a LD of 24 h photoperiod was used to ensure a photoperiod longer than the critical one, and thus sustained growth for all families. The photon flux density (400–700 nm) ranged from 87–91 mmol m-2 s-1. Eight weeks after germination, 50% of the seedling population was transferred to SD treatment (12 h light + 12 h darkness), whereas the remaining plants were retained in LD. Shoots were harvested in the period from 4 to 5 h after the onset of light in the morning at day 6 and day 20 from the onset of the SD and at the same time of the corresponding days for LD exposed plants. Three replicate samples, each consisting of 10 shoots, were harvested for each of the six full-sib families in each of the two SD or LD durations and immediately frozen in liquid nitrogen. The experimental material comprised altogether 72 samples.
RNA extraction Total RNA was extracted from ground shoots using an RNAqueous small-scale total RNA isolation kit with RNA Isolation Aid (#1911, Ambion, Austin, TX, USA), according to the manufacture’s instruction. Leftover DNA was removed by using the DNA-Free® DNase treatment and removal kit (Ambion, #1906).Total RNA preparations were stored at -80 °C. The integrity and quantity of total RNA was assessed by an Agilent 2100 Bioanalyzer with RNA 6000 Nano Kit (#5067-1511, Agilent Technologies, Palo Alto, CA, USA).
SSH and cDNA-library construction Both forward and reverse subtracted libraries were prepared from 1 mg of total RNA from two samples of family 6 (see Table 1) harvested at LD and after 6 d of SD treatment
using SuperSMART cDNA Synthesis Kit (#635000, Clontech, Mountain View, CA, USA) and PCR-Select cDNA Subtraction Kit (Clontech, #637401), following the manufacturer recommendations. In order to obtain a cDNAlibrary enriched for transcripts present in Norway spruce in response to SD, forward SSH was made in which the SD sample was used as the tester and the LD samples as driver. The reverse subtraction was performed using the LD sample as the tester and the SD sample as the driver, to identify transcripts normally present under LD conditions. The differentially expressed cDNAs were cloned into the pDrive vector using a PCR Cloning Kit (#231124, Qiagen, Hilden, Germany). The library arrays were sizefractionated using CHROMA SPIN-400 Columns (Clontech, #636076) and non-directionally cloned in a pDrive vector using PCR Cloning Kit (Qiagen, #231124), according to the manufacturer’s protocols.
cDNA sequencing and EST analyses The subtracted libraries from SD and LD treated plants were partially sequenced at the DNA Sequencing Laboratory at the Institute of Biotechnology, University of Helsinki, Finland. A total of 2034 sequences from randomly chosen clones were obtained. Sequences were processed for removal of vector sequences and poor quality regions as well as contigs assembly using SecMan II sequence analysis software (DNAStar Inc., WI). 1400 sequences were selected for further analyses and assembled. The obtained sequence data comprising consensus sequences and singletons were manually scrutinized for open reading frames (ORFs) using the NCBI ORF-finder (http://www.ncbi.nlm.nih.gov/gorf/ gorf.html), and the BLASTP analysis of the largest ORFs against the National Center of Biotechnology. Information (NCBI) database (NIH, USA) was also performed. Concurrently, TBLASTX and BLASTN analyses of the nucleotide sequence against the NCBI database was carried out. Similarities with a score more than 46 or an E-value of 1 e-4 were considered as a hit. Contigs were grouped into functional categories based on the Munich Information Center for Protein Sequences (MIPS) schemes (Schoof et al. 2002) and on literature searches. From the obtained list of gene models we selected 20 candidate genes among the most abundant contigs in the libraries and also among the annotated genes we guessed might be involved in bud set regulation. Because of a very limited number of annotated genes among those found in the subtracted libraries, on basis of literature searches we also selected a variety of genes considered to be involved in sensing and responding to environmental changes, regulation of bud set, cold acclimation and vernalization pathways in Arabidopsis and Poplar. To identify spruce homologs of these genes, we queried spruce EST databases with Arabidopsis protein sequences using TBLASTN. Selected spruce ESTs were then combined into contigs and best ORFs were used to query the protein database using BLASTP to confirm proper contigs homology. In total, we then detected 22 gene homologs for analysis in addition to
© 2010 Blackwell Publishing Ltd, Plant, Cell and Environment, 34, 332–346
SD induces gene expression in spruce 335 the 20 candidate genes selected from the subtracted libraries. A list of the studied genes is presented in Table 2.
Relative quantitative real-time RT-PCR Transcript levels of selected contigs for genes were determined in the 6 full-sib families with quantitative real-time RT-PCR. Primers were designed using the Primer3 online software (Rozen & Skaletsky 2000) with calculated Tm of 70 °C and amplification product no longer than 120 bp. The list of studied gene models and their primer sequences are shown in Table 2. The cDNAs were synthesized from 300 ng of total RNA with the TaqMan Reverse Transcription kit (Applied Biosystems, Foster City, CA, USA) in 50 mL reaction volume, followed by threefold dilution. Real-time PCR amplification was performed in a 25 mL reaction volume, using 2 mL of a threefold diluted cDNA solution and 250 nm of each primer. Reactions were conducted on the 7500 Fast Real-time PCR System (Applied Biosystems) using the default cycling conditions (2 min at 50 °C 10 min at 95 °C, 40 cycles of 95 °C for 15 s and 60 °C for 1 min). After each reaction, which included a no-template control, dissociation curves were carried out to verify the specificity of the amplification. All samples were analysed in duplicate and the transcript levels were normalized to actin.
Data and statistical analysis Data acquisition and analysis was carried out using the 7500 real-time PCR system SDS software for absolute quantification and Microsoft Excel.The relative amounts of transcripts were subjected to statistical analysis, as described previously (Johnsen et al. 2005a). The raw data from real time PCR consist of the critical threshold cycle values (dCt) for the endogenous transcript (actin – AT) and the corresponding threshold values for the transcript to be quantified (TARGET). The higher the transcript levels, the lower are the dCt values. A subtraction variable X = AT - TARGET was made for each sample and each target gene. To obtain a better presentation of higher- and lower transcript levels, the raw data X were transformed. Firstly, a mean value (Q) was calculated from all experimental observations of X for a specific target gene. Then a new transformed variable was calculated using the formula Yi = Xi - Qi for each sample and transcript. Note that Q is a scalar specific for each target transcript, and Y and X represent variables (varying by samples) specific for each target gene. Y is symmetrical around its mean value, which is 0 for each target gene. Positive values represent higher and negative values – lower transcript levels than average, and the unit of Y is expressed as the number of PCR cycles above or below zero (see Fig. 2).The new variable was analysed with the general linear model (GLM) procedure in the SAS software version 9.3 for Windows (SAS Institute Inc., Cary, NC, USA) according to Yijk = μ + Fi + Tj + Yk + FTij + TYjk + FYik + eijk, with Fi = family effect (i = 1,2,6); Tj = treatment effect (j = short or long days); Yk = day effect (6 or 20 d after onset of SD); FTij = interaction between family and treatment effect;
TYjk = interaction between treatment and day effect; FYik = interaction between family and day effect; eijk = residual random error (that is, F, T, Y, FT, TY, and FY were regarded as fixed effects). We focused only on differences with strong significance (P < 0.001) to avoid false rejections of the null hypotheses (Type I errors). In addition, trying to avoid overstatements as to the biological significance, we only discussed differences higher than one cycle in this paper, even if differences less than one cycle could be significant. To obtain an overall analysis of expression patterns of the 16 transcripts showing significant differences among F, T, Y at P < 0.001, Principal Component Analysis (PCA) was performed on the normalized data set (i.e. all Yijk) of these 16 chosen genes, with the PRINCOMP procedure in the SAS software. In this way, we were able to test for differences or patterns amongst treatments and similarities amongst observations within treatments, and to draw conclusions based on both variance and covariance patterns among transcript variables. The input data comprised 72 observations (two daylengths, two sampling points, six families, replicated three times) of each of the 16 transcripts included in the statistical analysis. The first two principal components retained 58% of the original variation, and constituted an output data set that could be analysed statistically, using exactly the same statistical model given above. Least square means of treatment and family combinations were plotted to highlight overall differences in the gene expression data (Fig. 3). Due to missing data observations in one of the replicates for one family at day 20, least square means had larger standard error of mean in this case.
Sequences deposition The P. abies EST sequences obtained were submitted to GenBank dbEST-database at the National Centre for Biotechnology Information (http://www.ncbi.nlm.nih.gov) with accession numbers ES227739-ES228292 (SD library) and ES228293-ES228855 (LD-library). The annotated sequences have been submitted to the GenBank with the following accession numbers EU332972–EU332991 (Table 2).
RESULTS EST analysis SD-cDNA and LD-cDNA libraries were generated by SSH to obtain ESTs representing genes predominantly expressed under SD or LD, respectively. For these two libraries a total of 1009 and 1025 clones were obtained, respectively, by DNA sequencing. After pDrive vector and quality end trimming, as well as removal of ambiguous (polyA-tails, short, noise,
