The link between transcript regulation and de novo protein synthesis in the retrograde high light acclimation response of Arabidopsis thaliana

Reorganization of the leaf proteome in light acclimation

Sun and shade acclimation depends on structural and functional reorganization of photosynthetic organs [24, 25]. Total leaf protein amount related to fresh weight differed between plants grown under L- or N-light conditions more than twofold. Two possible reasons might exist, namely either a similar protein complement at lower level or a profound qualitative difference that explains the lower level. Since plasmatic compartments such as cytosol, matrix and stroma contain about 25% (w/v) protein, e.g. 10 mg protein/40 μl chloroplast volume [26, 27], a twofold difference clearly indicates that the volumes of plasmatic compartments is strongly decreased after the 10 d L-light acclimation [11]. But in addition to a general decrease in volume, polypeptide composition also changes qualitatively. The best established example of light acclimation-dependent differences in protein composition concerns the increase in D1 protein and the decrease in light harvesting complex proteins (LHCII) with increasing growth light [28]. Changes in the photosynthetic apparatus are instrumental to adjust energy conversion and growth and are also important for optimized resource allocation, e.g. in dependence on light and nitrogen availability [29]. Protein patterns of silver-stained electropherograms differed between L- and N-acclimated plants. Many polypeptides appeared to be less abundant in N-light plants than in L-plants. This may be explained by normalization of each spot on total intensities in the gels. Due to the high RubisCO amount in extracts from N-plants, the intensities of most other bands will appear to be lower. But considering the low fresh weight-related protein contents of L-plants it becomes clear that the polypeptide abundance in silver gels would need some correction if polypeptide abundance should be related to fresh weight. Abundance of only few proteins changed during the 6 h period of H-light treatment. RubisCO was among the significantly accumulating proteins in the L→H-plants. It should be noted that the combined evaluation of both light shift treatments appeared justified despite in some cases different starting points due to the mostly similar response of protein abundance (82% similar response) and transcript regulation (100% similar response). This regulation leads to a highly similar transcriptome state after 6 h H-light [20].

Strengths and drawbacks of in vivo labelling of de novosynthesized proteins

Acclimation responses to environmental conditions are most frequently analysed at the level of specific transcripts or of genome-wide transcriptomes [30]. The matching of annotated silver-stained or Coomassie-stained 2D gels with autoradiograms was expected to allow for protein assignments of de novo synthesized polypeptides. But the labelling method needs some discussion. Labelling of intact plant tissue with ³⁵S-methionine requires time for uptake and incorporation, and in some studies it was achieved by wounding [31], in others by feeding via the transpiration stream [18] or by application to tissue surfaces. We chose the application to the cuticular surface of the youngest fully expanded leaves because neither application to the transpiration stream e.g. by injection or wounding, appeared suitable for our purpose of undisturbed but sensitive labelling of newly synthesized proteins. Labelling de novo synthesized leaf proteins by feeding the labelled amino acid to roots unlikely would allow for sufficiently strong incorporation within 2 h, but this could be compared in the future. The experimental design required incubation time for sufficient incorporation. Nevertheless, radiolabelling still is the only method at hand that allows for rapid, sensitive and reliable labelling of the de novo synthesized protein. It may be expected that with further advancement of mass spectrometric analysis, stable isotopes will offer alternative methods to study protein turnover also for eukaryotic multicellular organisms similar to unicellular organisms that can easily be labelled in suspension [32]. A recent review summarizes the strategies to label de novo synthesized proteins by modern proteomics [33]. The here employed method should be added to the portfolio of potential options that can be employed. Starting 4 h after transfer to H-light appeared suitable because many transcriptional changes had been shown to reach a new steady state at this time, e.g. sAPX [11] or monodehydroascorbate reductase, ABA-dependent cold regulated 47 (COR47), pyruvate kinase related protein (PKRP) [20]. Thus, the labelling that starts after translocation of ³⁵S-methionine through the cuticle to the mesophyll reflects a transcriptional state similar to 6 h after transfer to H-light for which the transcript analysis has been performed.

Apparent absence of coupling between transcript regulation and de novoprotein synthesis

The comparison of transcript regulation with differences in de novo synthesized protein demonstrates the flexible coupling between transcript regulation and translation (Figure 6). Piques et al. [34] compared transcript levels, ribosome occupancy, enzyme protein amount and activity at different times of day. Their scatter analysis revealed a poor dependency of ribosome loading on total amount of investigated transcripts. The Pearson’s correlation coefficient was 0.065 in the dark period and 0.102 in the light period [34]. Here, transcript analysis revealed efficient regulation following transfer to H-light. In sum 27 out of 42 transcripts of identified proteins, i.e. 64%, had log₂-fold differences ≥|0.5| between N- and L-light grown plants prior to H-light treatment. The size of this group of differentially regulated transcripts decreased to only 2 genes after 6 h of H-light. Thus, transcript regulation within this selected set of identified proteins was entirely in line with the global regulation of the transcriptome after 6 h of H-light [20]. Thus transcriptional regulation in response to H-light was almost completed after 6 h H-light.

In most cases regulation of transcript amounts was more pronounced than regulation of de novo protein synthesis. Regulation of 6 proteins occurred much stronger at the level of de novo protein synthesis. Several translation factors have been identified as target of posttranslational regulation including thiol-disulfide transitions [35], glutathionylation [36], phosphorylation [37] and S-nitrosylation [38]. Among the targets researchers identified several ribosomal proteins (RPL S1, S6, L13, L30), elongation factors (EF-Tu, EF-G, EF-2, EF-1α) and enzymes such as nucleoside diphosphate kinase III and tRNA synthetases which all are involved in translation. Redox changes, ROS production and activation of phosphorylation cascades have been implicated in retrograde signalling. The protein kinases STN7 and STN8 mediate light-dependent reorganization of the photosynthetic apparatus [39]. ROS waves adjust nuclear gene expression in excess light acclimation [40]. ROS and redox feed into the mitogen activated protein kinase pathway [41]. Translational activity is strongly altered by ROS in yeast [42]. Thus, translation in plants is a prime but hitherto not sufficiently explored target of retrograde signalling as underlined by the data presented in this paper. The reader is also referred to the metaanalysis by Schwarzländer et al. [43] who observed that transcripts encoding for proteins involved in protein synthesis are significantly affected by retrograde signals released from the mitochondrion.

Functional implications of translational control of identified targets

Control of posttranscriptional processes accelerates the speed and versatility of stress acclimation. The high significance of specific transcript recruitment to ribosomes in plants has best been demonstrated for acclimation to hypoxia [44]. The authors showed hypoxia-specific changes of transcriptome and translatome at the global, organ- and cell-specific level. Preferential ribosome association was observed for sucrose transporters, heat shock factors and transcription factors [45]. Here, expression of six genes was more strongly regulated at the level of protein synthesis than of transcript accumulation. It may be assumed that the gene product functions are needed after transfer to H-light. Despite down-regulation at the transcript level, ³⁵S-methionine incorporation into HSP70-1 still occurred at high rates. In a converse manner, HSP70-2 was synthesized at similar rates despite a large increase in transcript amount. Chloroplast HSP70s facilitate protein import into the chloroplasts, a function which is of eminent importance during environmental transition such as exposure to excess excitation energy [45]. High chlorophyll fluorescence HCF136 was identified in a screen for genes with function in assembly of functional photosystem II [46]. FBP aldolase as part of the Calvin cycle, O-acetyl serine thiol lyase with its function in cysteine synthesis, carbonic anhydrase which facilitates equilibration between carbonate and CO₂ as substrate of the Calvin cycle and 3-ketoacyl CoA thiolase 3 involved in fatty acid synthesis showed stimulated de novo synthesis. This type of regulation may easily be reconciled with their metabolic functions which are important for H-light acclimation. Arguments appear less straight forward when it comes to explain the low level of de novo protein synthesis observed for 16 genes. They mostly function in metabolism such as seduheptulose-1,7-bisphosphatase which is suggested to limit Calvin cycle activity [47], large and small subunits of RubisCO, RubisCO activase, phosphoglycerate mutase, phosphoglycerate kinase and ribose-5-phosphate isomerase. Others are involved in redox homeostasis and antioxidant defence (malate dehydrogenase, dehydroascorbate reductase, superoxide dismutase, stromal ascorbate peroxidase and the regulator of chloroplast cysteine synthase complex cyclophilin Cyp20-3 [12]. It may be hypothesized that these proteins are present at sufficient amounts prior to H-light treatment and that the low ratio of de novo synthesis-to-transcript amount merely reflects such mechanisms of yet un-understood feedback control. It should be noted that photoreceptor-dependent signaling might contribute to the transcriptional and translational responses described in this paper, albeit previous work largely excluded a major role of photoreceptors in this particular experimental setup [7, 11].

Plant growth and treatment

Arabidopsis thaliana was grown in a growth chamber in a mix of 50% soil, 25% Perlite and 25% Vermiculite, supplemented with one dose of Lizetan (Bayer, Germany). Following seed stratification for 2 d at 4°C, plants were grown for 30 d in 80 μmol quanta^.s^-1.m^-2 (N-light) with a 14 h light and 10 h dark phase. Subsequently, plants were transferred to 8 μmol^.s^-1.m^-2 (L-light) for 10 d prior to the experiment with transfer to 800 μmol^.s^-1.m^-2 (H-light; 100-fold light increase). The L-plants have been shown to be entirely shade acclimated [11]. Another set of plants was grown in N-light for the whole period of 40 d and also transferred to 800 μmol^.s^-1.m^-2 (10-fold light increase). Control plants were kept in L- and N-light, respectively, and harvested in parallel to the H-light rosettes. Harvest time was always at 3 pm. Chlorophyll, protein and RNA contents and effective quantum yield of photosystem II by pulse amplitude modulation (PAM) were determined as described in Oelze et al. [11].

In vivo labelling of de novo synthesized proteins

L-[³⁵S]-methionine (NEG009T, Perkin Elmer, MA, USA) was supplemented with 0.1% (v/v) Triton X-100 and applied to leaf surfaces with a radioactivity of 20 μCi per leaf. For each treatment 20 μCi were administered to fully expanded leaves from three different rosettes 4 h after transfer to H-light. After 6 h, the leaves were excised from the rosettes, washed first with 0.1% (v/v) Triton X-100 and then with 0.5 mol/L Tris-Cl, pH 6.8.

2D-gel electrophoresis

Leaves were ground with a pestle in 1 mL acetone/trichloroacetic acid/β-mercaptoethanol (89.93:10:0.07% v/v) according to Méchin et al. [49]. Following precipitation at -20°C for at least 1 h and subsequent centrifugation, the pellet was washed and sedimented thrice with ice-cold acetone/β-mercaptoethanol, dried and resuspended in lysis buffer [50]. For radioactive samples, incorporated ³⁵S was quantified by precipitating aliquots on Whatman filter followed by scintillation counting. For silver-stained gels, protein amounts were quantified at 595 nm with the BioRad protein assay. Separation in the first dimension was achieved with Immobiline™ DryStrips (pH range 3-10 NL, 18 cm, GE Healthcare, Uppsala, Sweden). Sample equivalent to 100 μg protein or 10⁶ cpm was dissolved in 340 μL complete rehydration buffer (8 mol/L urea, 2% (w/v) CHAPS, 0.002 bromophenolblue, 0.3% ampholyte, 1.4% (w/v) dithiothreitol) and applied to the Immobiline strips. The rehydration and isoelectric focusing protocol consisted of the steps as follows: 1 h 0 V, 12 h 30 V, 2 h 60 V, 1 h 500 V, 1 h 1000 V, 1000-8000 V for variable time to reach 42000 Vh. Separation in the second dimension was performed on a 12% (w/v) SDS-PAGE of 18 cm length at 40 mA. Silver staining was performed according to Blum et al. [51] and autoradiography as described in Dietz and Bogorad [52].

Analysis of 2D-gels and heat map construction

Delta 2D software (Decodon, Greifswald, Germany) with its SmartVectors Technology was used to align the gel images to each other to allow for efficient and reliable spot matching. A fusion image was generated containing all spot positions. Each gel was matched with this master gel. Spot boundary detection, pixel intensity quantification and statistical analysis (one way ANOVA) were performed with the built in TIGR MeV tool. Before constructing the heat map, the data set was standardized to zero mean and unit variance. Clustering was achieved using the eucledian distance and complete linkage- default settings of the delta 2D software (DECODON, Greifswald, Germany).

ATH1-genome array hybridisation and analysis

Isolated total RNA was sent to KFB-company (Competence Centre for Fluorescence Bioanalytics, Regensburg, Germany), processed, and derived fluorescent probes hybridized against the 25mer oligonucleotide ATH1-genome array (Affimetrix, Santa Clara, USA). Glyceraldehyde-3-phosphate dehydrogenase, actin and ubiquitin were used as reference transcripts. The raw data were fed into ROBIN (MPI Golm, Germany). Statistical evaluation of the data was based on the corrected p-value [22, 23].

Protein identification by mass spectrometry

Corresponding areas of interest were excised from the 2D gels and washed with (a) two times a solution containing trifluoroacetic acid (0.1% w/v) and acetonitrile (60% v/v), (b) acetonitrile (50%), (c) acetonitrile (50%)/50 mM NH₄HCO₃ for 0.5 h, and (d) acetonitrile (50%)/10 mM NH₄HCO₃ at 21°C for 0.5 h each. Dried gel slices were resuspended in trypsin solution (0.013 mg sequencing quality trypsin (Promega, Mannheim, Germany) in 10 mM NH₄HCO₃ pH 8.0) at 4°C for 0.5 h and afterwards at 37°C for about 15 h. Digestion solutions were supplemented with cyano-4-hydroxy-cinnamic acid at a 60:40% ratio. Mass spectra were determined using a Biflex III matrix-assisted laser desorption/ionisation-time of flight mass spectrometer (MALDI-TOF)-MS (Bruker, Bremen, Germany) (previously described [53]). The peptide mass fingerprints (PMF) obtained by tryptic digested proteins were analyzed by MALDI-TOF-MS and proteins were identified by MASCOT (Multiple-Access Space-Time Coding Testbed) software and the National Center for Biotechnology Information (NCBI) protein database. The program compares the peptide masses obtained from experimental digestion to the predicted peptide masses from the theoretical digestion of proteins.

Correlation of de novoprotein synthesis and transcript regulation during H-light treatment

The obtained values of the spot intensities for the autoradiograms by Delta 2D were used to calculate the ratios between the different treatments (N/L, N→H, L→H, L→H/N→H). The ratios were recalculated as log₂-fold change values, to be easily comparable to the obtained log₂-fold change values of the microarray experiments by ROBIN.

Therefore, the maxima of regulation would fit in the range between -2 and 2. To evaluate the relationship between de novo synthesis and transcriptional regulation, the calculated values were plotted in a diagram where deviation from the diagonal ≤0.5 was set as a cutoff (gray shaded area) and only larger deviations (outside this area) were accepted to indicate distinct regulation between transcript and de novo protein synthesis.