- © 2012 American Society of Plant Biologists. All rights reserved.
Abstract
Mitochondria play a crucial role in germination and early seedling growth in Arabidopsis (Arabidopsis thaliana). Morphological observations of mitochondria revealed that mitochondrial numbers, typical size, and oval morphology were evident after 12 h of imbibition in continuous light (following 48 h of stratification). The transition from a dormant to an active metabolic state was punctuated by an early molecular switch, characterized by a transient burst in the expression of genes encoding mitochondrial proteins. Factors involved in mitochondrial transcription and RNA processing were overrepresented among these early-expressed genes. This was closely followed by an increase in the transcript abundance of genes encoding proteins involved in mitochondrial DNA replication and translation. This burst in the expression of factors implicated in mitochondrial RNA and DNA metabolism was accompanied by an increase in transcripts encoding components required for nucleotide biosynthesis in the cytosol and increases in transcript abundance of specific members of the mitochondrial carrier protein family that have previously been associated with nucleotide transport into mitochondria. Only after these genes peaked in expression and largely declined were typical mitochondrial numbers and morphology observed. Subsequently, there was an increase in transcript abundance for various bioenergetic and metabolic functions of mitochondria. The coordination of nucleus- and organelle-encoded gene expression was also examined by quantitative reverse transcription-polymerase chain reaction, specifically for components of the mitochondrial electron transport chain and the chloroplastic photosynthetic machinery. Analysis of protein abundance using western-blot analysis and mass spectrometry revealed that for many proteins, patterns of protein and transcript abundance changes displayed significant positive correlations. A model for mitochondrial biogenesis during germination is proposed, in which an early increase in the abundance of transcripts encoding biogenesis functions (RNA metabolism and import components) precedes a later cascade of gene expression encoding the bioenergetic and metabolic functions of mitochondria.
A distinguishing feature in the life cycle of many flowering plants is the production of seeds that are capable of remaining dormant for several years. Yet, upon perception of appropriate signals, the dormant seed transitions from a quiescent state to a highly active metabolic state. This transition requires a relatively rapid reestablishment of the various biochemical activities localized to a number of distinct cellular structures/organelles. These organelles can then utilize the stored reserves to establish the seedling before photosynthesis begins to support plant growth. This ordered and timely reestablishment of metabolic function is essential for successful seedling establishment, as the biochemical machinery required for autotrophic growth must be completed before seed reserves are exhausted (Bewley, 1997).
In dry seeds, mitochondria do not have the characteristic cristae structures, which represent infolding of the inner membrane that contains the ATP-producing respiratory chain (Logan et al., 2001; Vartapetian et al., 2003; Howell et al., 2006). Rather, they appear as more circular organelles where both membranes are close to each other (Logan et al., 2001; Vartapetian et al., 2003; Howell et al., 2006). This structure, which has been termed a promitochondrion structure (Logan et al., 2001), develops the more typical mitochondrial morphology in 12 to 24 h after imbibition (the passive uptake of water by the seed) in maize (Zea mays) and rice (Oryza sativa; Logan et al., 2001; Howell et al., 2006). This structural change is accompanied by an increase in metabolic activity. However, mitochondria do not arise de novo but rather from preexisting organelles, and the details of the molecular mechanisms and signals that initiate this process are unclear. Mitochondrial biogenesis requires energy and various building blocks that are produced in the mitochondria; thus, some endogenous mitochondrial activity must exist to prime subsequent mitochondrial biogenesis, which ultimately provides energy for a variety of cellular processes in the germinating tissues. The regulation and timing of mitochondrial biogenesis are of critical importance to the process of germination and early seedling establishment.
Studies on mitochondrial biogenesis during germination have been primarily carried out in starch-storing seeds, such as maize and rice. Initial studies in maize revealed that transcripts and proteins of the ATPase complex and cytochrome oxidase began to accumulate after 6 h of imbibition (Ehrenshaft and Brambl, 1990). A notable exception was observed with the mitochondrially encoded subunit 9 of the ATP synthase complex, which was detected in dry embryos and increased further in imbibition. A later study showed that the transcript and protein for mitochondrial HSP60 were present in dry embryos and increased upon imbibition (Prasad and Stewart, 1992). A detailed study in maize built on these earlier studies to propose that different populations of mitochondria existed and that the “light” or promitochondrial fraction matured into cristae-containing mitochondria with large increases in metabolic components (Logan et al., 2001). Studies in rice revealed that an ordered assembly of mitochondria occurs, that components required to build or import mitochondrial proteins are present in promitochondrial structures, and that they are active almost immediately upon imbibition to facilitate the rapid rate of mitochondrial biogenesis and associated increases in respiration observed in the first 24 h after imbibition (Howell et al., 2006, 2007).
No direct studies have been carried out in the oil seed Arabidopsis (Arabidopsis thaliana) to investigate mitochondrial biogenesis and determine how it proceeds relative to the other energy organelles: chloroplasts and peroxisomes. The importance of the mitochondrion in providing energy for germination and the interaction with photosynthesis position it as a critical organelle during germination and early seedling establishment (Nunes-Nesi et al., 2011). Previously, we have carried out an in-depth transcriptome analysis of germination and defined greater than 700 germination-specific genes (i.e. genes defined as displaying maximum expression during germination compared with all other development stages in Arabidopsis; Narsai et al., 2011). Thus, to define the role of mitochondria in germination and to study the earliest events of mitochondrial biogenesis during germination in Arabidopsis, we undertook a closer analysis of this germination transcriptome data set (GSE30223; Narsai et al., 2011) and combined it with a detailed analysis of mitochondria- and plastid-encoded organellar genes (using high-throughput quantitative reverse transcription [qRT]-PCR). To link these data with cellular development, we visualized mitochondrial profiles during germination using mitochondrially targeted GFP and compared this with chloroplast- and peroxisome-targeted red fluorescent protein (RFP) and cyan fluorescent protein (CFP), respectively. Protein abundance profiles during germination were also analyzed in this study, using antibodies and peptide mass spectrometry to link the timing of transcriptional changes to protein accumulation changes. The results show the status of mitochondria in fresh dry seeds and the changes that occur over the course of seed maturation, stratification (moist chilling of the seeds in darkness to alleviate dormancy), germination (the time from the start of water uptake to the point where the embryo emerges from the testa), and post germination. These results show some of the earliest changes in mitochondria that occur during germination.
RESULTS
Temporal Examination of Organelle Biogenesis
The use of a stably transformed line of Arabidopsis seed containing different fluorophores targeted to the mitochondria, plastid, and peroxisome facilitated an in vivo temporal examination of biogenesis for the energy organelles during germination (Fig. 1A). Throughout the germination time course, peroxisomes (cyan channel) were observed to be present in high densities, with negligible changes to morphology and size (approximately 2 μm). This is consistent with the pivotal role of glyoxysomes and peroxisomes in metabolizing the energy stores present in oil seeds prior to the establishment of photoautotrophism (Graham, 2008). In the red channel, small proplastids (1–2 μm) were observed at high densities in the first time point (dry seed; Fig. 1Ai). The plastids remain small until as late as 12 h SL (i.e. 12 h into continuous light, following 48 h of stratification at 4°C in the dark), after which they undergo a major increase in size (approximately 4 μm) and exhibit a globular morphology by 48 h SL. This organellar expansion is symptomatic of the transition from undifferentiated proplastid to etioplasts and chloroplasts, as photosynthesis is established in the greening cotyledons. In the green channel, mitochondria could not be visualized in dry seeds, and as the mitochondrial, peroxisomal, and plastid fluorescence-tagged proteins were all driven by the 35S cauliflower mosaic virus promoter, the lack of a mitochondrial signal reflects mitochondrial mass or protein, not a lack of expression of the marker. However, following 48 h of stratification (Fig. 1Aii), GFP-containing mitochondrial populations could be observed, with varied morphologies and sizes (0.5–2 μm). After 12 h SL in continuous light (Fig. 1Aiii), the mitochondrial morphology took on a more uniform rod shape, accompanied by an increase in mitochondrial density. This is consistent with studies using electron microscopy that indicate that the cristae structure of mitochondria develop 12 to 24 h after imbibition (Logan et al., 2001; Howell et al., 2006). This increase in density and uniform morphology remained consistent throughout the remainder of the time course. In the early time points (Fig. 1A, i–iii), peroxisomes and proplastids had a seemingly random distribution within cells and were not observed to interact with each other. However, in the final time point (48 h SL; Fig. 1Aiv), the peroxisomes were observed to colocalize in clusters around the chloroplasts. Colocalization of this nature has been observed in adult leaf tissue previously (Mano et al., 2002) and suggests the establishment of photorespiration, a hallmark of the transition from glyoxysome to leaf peroxisome.
Developmental profiling of mitochondria, plastids, and peroxisomes in an Arabidopsis embryo/seedling over a germination time course. A, Seeds from a stably transformed line of Arabidopsis plants expressing mitochondria-targeted GFP, plastid-targeted RFP, and peroxisome-targeted CFP were examined at a number of time points during seed germination. i, The cotyledon of an embryo dissected from a dry seed. Peroxisomes (cyan) and proplastids (red) are visible in high numbers. ii, The cotyledon of an embryo dissected from a seed after 48 h of stratification (48 h S; prior to transfer to continuous light). Peroxisomes (cyan) and proplastids (red) are visible in high numbers; mitochondria (green) can begin to be identified, although these are not as distinct as in later time points. iii, The cotyledon of an embryo dissected at 12 h in continuous light (12 h SL). All three organelles are clearly visible. iv, The cotyledon of an embryo dissected at 48 h in continuous light (48 h SL). All three organelles are more clearly visible. The plastids (red) have quadrupled in size and are colocalizing with the peroxisomes. Bars = 15 μm. B, Total protein was extracted from Arabidopsis seeds collected during the germination time course and separated by SDS-PAGE (30 μg). Separated proteins were transferred to a polyvinylidene difluoride membrane and subjected to western-blot analysis. Following quantitation of band intensities using ImageQuant TL software (GE Healthcare), values were normalized to the highest level of intensity over the time course and graphed.
Protein abundance changes for several organelle proteins were analyzed across a time series from freshly harvested seeds (upon removal from siliques) as well as dry seeds after 15 d of ripening (0 h), across 48 h of dark stratification (S), and then following 48 h of exposure to light (SL) at 22°C. Western-blot analysis using antibodies raised to markers of different compartments showed that the pattern of protein expression over the time course differed substantially (Fig. 1B). For some proteins, the protein abundance did not dramatically change in abundance over the time course, with less than a 50% change in abundance seen from dry seeds to 48 h SL (Fig. 1B). This was true for the cytosolic marker UGPase, which maintained a relatively consistent abundance, being highly abundant (70% or 0.7) in harvested and dry seed to maximum abundance (100% or 1.0) at 48 h SL (Fig. 1B). The mitochondrial marker ATPβ varied in abundance from 60% (0.6) at harvest and dry seed to maximum abundance at 48 h SL (Fig. 1B). Larger increases in protein abundance were observed in mitochondrial voltage-dependant anion channel (VDAC) and the peroxisomal 3-ketoacyl-CoA thiolase (Kat2; 40% [0.4] at harvest and dry seed to 100% [1.0] at 48 h SL). The largest increases in protein abundance were observed for the two subunits of plastid Rubisco, denoted SSU and RbcL (30% [0.3] at harvest and dry seed and 100% [1.0] at 48 h SL), and the vacuole V-ATPase (20% [0.2] at harvest and dry seed to 100% [1.0] at 48 h SL; Fig. 1B).
Transcript and Protein Correlation in Organelle Biogenesis during Germination
To study the process of organelle biogenesis during germination, transcript abundance changes were analyzed using the previously published microarray data (GSE30223; Narsai et al., 2011), and protein abundance was measured by shotgun mass spectrometry detection of peptides. Of the 15,789 genes expressed in at least one time point during the time course (Narsai et al., 2011), quantitative data for 178 of the corresponding proteins were obtained by counting peptide mass spectra (Ishihama et al., 2005; see “Materials and Methods”). For each of these, the corresponding transcript abundance was normalized to maximum (for the equivalent time points as the proteins) and visualized as a heat map in order of functional category (Fig. 2; Supplemental Table S1). In this way, it was possible to identify any concordance between transcript and protein levels during germination.
Concordance/discordance in transcript and protein abundance data during germination. Transcript and protein abundance data were normalized to maximum and visualized as a heat map in order of functional category. Pearson’s correlation coefficient was calculated for each transcript/protein abundance pair. Correlation coefficients greater than 0.625 were considered statistically significant (P < 0.05) and are indicated by the gray line. Annotation and localization details for each gene are shown in Supplemental Table S1, in accordance with the displayed order.
Examination of the heat map showing transcript and protein abundance, and the corresponding correlation coefficient for each gene revealed a pattern of concordance relating to function, with 81 of the 178 transcripts/proteins showing significant positive correlations of r > 0.62, while only 15 of the 178 transcripts/proteins showed significant negative correlations (r < −0.62; Fig. 2). Notably, a number of genes encoding mitochondrial (green bars), plastid (red bars), and peroxisomal (blue bars) components from this overlapping set revealed strong correlations (Fig. 2). For example, for genes encoding translation functions, particularly ribosomal proteins, a strong concordance was observed between transcript and protein abundance, with positive correlations observed for all but two of the genes encoding translation functions (Fig. 2). Also, nine of the 14 mitochondrial proteins displayed significant positive correlations (r > 0.62); for example, two ADP/ATP carrier proteins (At3g08580 and At5g13490) that have a role in transport were seen to have significant positive correlations (Fig. 2; Supplemental Table S1). Additionally, 10 of the 34 plastid proteins also showed strong positive correlations (r > 0.62; Fig. 2). The pattern of expression for these organellar proteins (and transcripts), largely encoding major metabolic components, showed an increase over time, sometimes showing a lower/stable expression by 48 h SL (Fig. 2). In contrast, embryo development and other seed-related proteins, such as the late embryogenesis abundance proteins (LEA) and two abscisic acid-responsive proteins, were most highly expressed in dry seed (0 h) and decreased in abundance in both transcript and protein to varying degrees upon imbibition. However, despite the difference in expression pattern, these functions were also seen largely to show strong transcript and protein correlations. Thus, using microarray analysis of transcript abundance and mass spectrometry-based quantitation of proteins (Fig. 2), in a number of cases, it is clear that strong concordance is evident.
Comparative Analysis of Organelle Transcripts Reveals a Transient Suite of Genes Unique to the Mitochondria
Of the 15,789 genes expressed in at least one time point during the germination time course, the genes encoding organelle-targeted proteins were analyzed to compare transcriptomic profiles observed over the time course of stratification and germination (Supplemental Fig. S1; Supplemental Table S2). Using strict and consistent criteria of localization predication databases and publications showing experimentally determined localization (see “Materials and Methods”), genes were defined as encoding proteins localized to the plastid (2,135), peroxisome (447), and mitochondrion (980) in an equivalent manner (Supplemental Fig. S1; Supplemental Table S2). By defining localizations in this manner, bias for a single organelle was reduced to a minimum, allowing maximum comparability between these gene lists (see “Materials and Methods”). For each gene set (e.g. the mitochondrial set; Fig. 3), expression values were hierarchically clustered, and four distinct clusters were observed: cluster 1, transcripts that increased in abundance from 6 h SL and remained at high levels across the time course; cluster 2, highly expressed in dry seed and up to 12 h S before decreasing over time; cluster 3, transiently expressed with low/absent expression levels in dry seed and up to 12 h S before increasing dramatically in abundance at 48 h S, but then decreasing after 6 to 12 h into the light (SL); and cluster 4, largely unchanging in abundance over the time course. Upon examination of the genes encoding plastid-localized proteins, it was evident that these were significantly overrepresented in cluster 1 profiles (up-regulated after 6 h SL; 48% plastid versus 30% across the whole transcriptome; Supplemental Fig. S1Ai). This overrepresentation was not unexpected, given that the plastid proteins largely function upon exposure to light. In contrast, genes encoding peroxisome proteins were not different in expression pattern from that seen in the genome (Supplemental Fig. S1Aii). Interestingly, it was seen that for the mitochondrial gene set, there was an overrepresentation of transcripts showing transient expression (cluster 3, 25% versus 15% across the whole transcriptome; Fig. 3). The decrease seen in the abundance of transcripts in this transient cluster coincides with the time point(s) when transcripts in cluster 1 begin to increase (6–12 h SL) and also coincides with the time point(s) when the morphological observations reveal that mitochondria take on the typical rod-shaped appearance (Fig. 1Aiii). Thus, the appearance of transcripts in cluster 3 correlates with a transitional event from the dormant seed to active mitochondrial biogenesis. The correlation between mitochondrial function and this transitory burst in expression led us to examine putative regulatory elements in the promoters of these genes (Fig. 3).
Distribution of overrepresented motifs (left panel) and transcription factor (TF)-binding sites (right panel) and in the promoters of mitochondrial genes (center panel). All transcripts for genes encoding mitochondrial proteins called present at a minimum of one time point were normalized to the highest expression value of each gene and hierarchically clustered. Cluster 3 of the mitochondrial set was observed to be significantly enriched compared with the clustered distribution of the genome, indicated with an asterisk. On the left, putative regulator-binding motifs calculated to be overrepresented (by z-score analysis; P < 0.05) are indicated for clusters 1 and 3 (i.e. where transcripts were observed to increase during the time course of the experiment). On the right are listed known transcription factor-binding sites enriched in clusters 1 (pink) and 3 (green), as determined by Pscan (Zambelli et al., 2009), with the expression of the profile of the transcription factors indicated according to cluster (e.g. the binding site for PIF3 is enriched in genes present in clusters 1 and 3, and the expression of PIF3 is assigned to cluster 1 by hierarchical clustering).
In order to gain insight into the regulatory processes that drive the observed changes in transcript abundance, the presence of cis-acting regulatory elements or transcription factor-binding sites was analyzed. First, the promoter regions of genes in clusters 1 and 3 were searched for 6-mers (six-nucleotide motifs) to find significantly overrepresented motifs (Fig. 3, left side). In addition, Pscan was used to search for motifs that are known binding sites of specific transcription factors (Zambelli et al., 2009), and their overrepresentation is shown with a variety of elements defined in clusters 1 and 3 compared with the genome representation, and while there was some overlap (GATA, Heat Shock Factor binding site, and Heat Shock Element), there were also distinct groups in each cluster (Fig. 3). Notably, the ABRE-like and DRE elements, which have been previously associated with germination, were present in cluster 1 but not in cluster 3 (Nakabayashi et al., 2005). Analysis of known transcription factor-binding sites may be more informative, as the regulators are known. Here, it was also observed that both clusters 1 and 3 shared overrepresented binding sites for PIF3, ANT, FLC, ABI4, and CDC5, but cluster 3 was also unique in that it contained RAV1, DAG1, and DAG2 binding sites, while cluster 1 contained bHLH binding sites that were not enriched in cluster 3. The binding sites for all these transcription factors have been previously associated with germination, but with reference to the expression of plastid-encoded genes or proteins found in other locations (Peterson, 1977; Ikeda et al., 2009; Groszmann et al., 2010; Lau and Deng, 2010; Taylor et al., 2010; Rizza et al., 2011). Thus, it would appear that the regulation of genes encoding mitochondrial proteins during germination is likely to be under these same mainstream regulatory pathways. Notably, the finding that genes encoding mitochondrial proteins are overrepresented in this transient expression pattern suggests that one of the earliest targets of these previously characterized regulators of germination is in fact genes that encode mitochondrial proteins, which are activated very early during germination.
To gain a detailed insight into the mitochondrial functions activated as one of the primary events of germination, a functional analysis was carried out. The first step involved further curation of the mitochondrial gene list, which was generated largely upon collation of localization data from 40 relevant publications (Supplemental Table S2D; Supplemental Materials and Methods S1). In this way, the expression of a refined list of 945 genes encoding mitochondrial proteins was examined, and a custom pathway image was used to compare and contrast patterns of transcript abundance for genes encoding mitochondrial proteins that were assigned to distinct functional categories (Fig. 4; Supplemental Fig. S1B). Stable and low transcript abundance was observed in all functional categories at the first four time points (H, 0 h, 1 h S, and 12 h S; Fig. 4), corresponding with the lack of visible mitochondria during this time (Fig. 1Ai). The first suite of transcripts to undergo a discernible and dramatic increase in abundance were those encoding proteins belonging to the protein import apparatus of the outer mitochondrial membrane (TOMs, SAM, and metaxin), which were seen to peak in expression at 48 h S and 1 h SL (Fig. 4). The early increase in the abundance of these transcripts complies with previous findings in rice that have also shown these genes to be among the first to increase in transcript abundance upon imbibition (Howell et al., 2006, 2009). Interestingly, however, this intensive examination of the mitochondrial subset revealed that this distinct expression pattern was also observed to a greater extent for transcripts encoding proteins involved in the transcription of mitochondria-encoded genes as well as RNA editing and splicing factors (Fig. 4, light green shading). Notably, when the protein abundance for one of these factors (a mitochondrial pentatricopeptide repeat [PPR; At5g46460]; Fig. 2) was quantitated, the protein abundance also corresponded with the transient expression seen for the respective transcript (Fig. 4, purple asterisk; Supplemental Table S1). This transient expression was also seen for RNA polymerase and splicing factors as well as mitochondrial members of the mitochondrial transcription termination factor family (Fig. 4), which have been shown to have a role not only in transcription termination but also in transcription initiation (Hyvärinen et al., 2010) and even in modulating mitochondrial DNA replication (Hyvärinen et al., 2007; Roberti et al., 2009). Interestingly, it was seen that chloroplast RNA polymerase did not show this transient expression pattern (data not shown) and instead peaked in expression at 48 h SL. Coincident with this early transient induction of genes involved in mitochondrial RNA metabolism was an analogous increase in the abundance of transcripts encoding proteins involved in cytosolic nucleotide synthesis and a number of mitochondrial carrier proteins previously defined as nucleotide carriers (Palmieri et al., 2011; Fig. 4, nucleotide synthesis and nucleotide carriers).
Transcript abundance profiles of genes encoding mitochondrial proteins during germination. Transcript abundance was normalized to the maximum expression level during the time course, classified into nontrivial BINS, and displayed on a custom pathway image. A method of color-coding was utilized to distinguish subtle variations between the profiles. Three shades of green were used to denote transcript abundance peaking at 48 h S/1 h SL (light green), 6 h SL (medium green), and 12 h SL (dark green). Two shades of red were used to denote transcript abundance increasing steadily, with maximal abundance in the final time point (dark red) and transcript abundance increasing steadily until 24 h SL, followed by a small decrease/stabilization in abundance at 48 h SL (light red). For a number of transcripts, one or more proteins were quantitated. Purple asterisks have been used to indicate when transcript and protein abundances peak at the same time point.
At 6 h SL, transcripts of genes encoding ribosomal proteins, proteins involved in protein folding, and the inner membrane protein import machinery peaked, as did a number of proteins associated with mitochondrial DNA replication (Fig. 4, medium green shading). The final transient peak in transcript abundance was observed for genes encoding tRNAs and the protein constituents of the VDAC, which occurred at 12 h SL (Fig. 4, dark green shading; Supplemental Fig. S2, dark green shading). Quantitative western blotting also revealed that the protein abundance of VDAC accumulated in a similar way to the corresponding transcript expression pattern from dry seed to 48 h S, and protein abundance only peaked at the final time point (Supplemental Fig. S3).
Transcripts encoding the metabolic functions of mitochondria, such as components of the electron transport chain, ascorbate glutathione metabolism, and photorespiration, all peaked later during germination, at either 24 h SL or 48 h SL (Fig. 4, light red and dark red shading, respectively). Quantitative western blotting for components of complex I, complex III, and complex V also revealed that these proteins peaked in abundance coincident with the respective transcript abundances (Fig. 4; Supplemental Fig. S3). Interestingly, tricarboxylic acid (TCA) cycle components also peaked in transcript abundance at 24 h SL or 48 h SL (Fig. 4). It is possible that this late induction of the TCA cycle and glycolysis components reflects the change in energy pathways, from the mobilization of storage lipids via β-oxidation and the glyoxylate cycle during germination (Bewley, 1997), which could provide the energy for early RNA and protein metabolism, to the use of newly fixed carbon through photosynthesis and the glycolytic degradation of sugars as seedling establishment occurs. Interestingly, genes encoding proteins of the glycolytic pathway, located in the cytosol but previously reported to be associated with the outer mitochondrial membrane (Graham et al., 2007), peaked in expression at the same time (48 h SL) as the transcripts encoding mitochondrial metabolic functions (Fig. 4, glycolysis). Notably, these transcripts (encoding glycolytic functions) showed a very slow increase in abundance during stratification and up to 12 h SL (likely representing a minor response to imbibition) before sharply increasing in transcript abundance at 24 h and 48 h SL, when the energy demand increases with seedling establishment (Fig. 4, glycolysis).
The Mitochondrial Transcriptome, Nucleotide, and RNA Metabolism during Germination
The changes in transcript abundance patterns described above suggest that mitochondrial RNA metabolism is activated very early during germination, just after imbibition. As ATH1 microarray GeneChips use polyadenylated mRNA to measure transcript abundance, this is not a suitable platform to measure the abundance of the nonpolyadenylated transcripts of mitochondria-encoded genes. Thus, a qRT-PCR platform analyzing random-primed cDNA was employed to measure transcript abundance for genes encoded in the mitochondrial genome (de Longevialle et al., 2007). We used four time points to determine the pattern of expression for mitochondrial genes encoding proteins found in complexes I, III, IV, and V and for three genes, ccmB, ccmC, and ccmF, that are required for cytochrome c biogenesis (Giegé et al., 2008). The profiles for these mitochondrial transcripts were compared with the transcriptomic profiles observed for the nucleus-encoded counterparts of these complexes at the same time points: 0 h, 6 h SL, 24 h SL, and 48 h SL. The 0-h and 6-h SL time points reveal which mitochondrial transcripts (indicated in gray) increased significantly in abundance before the nucleus-encoded counterpart transcripts (indicated in red). The latter were seen to significantly increase in abundance only at or after 6 h SL (Fig. 5).
Coordination of nucleus- and organelle-encoded components of the mitochondrial electron transport chain. Expression values derived from microarray analysis of nucleus-encoded components of the mitochondrial electron transport chain were profiled in parallel to corresponding mitochondria-encoded components quantitated by qRT-PCR. For each gene displayed, gene names, annotations, and abundance levels are detailed in Supplemental Table S3.
It was evident that the pattern of accumulation of transcripts for mitochondrially encoded proteins differed significantly compared with the accumulation of transcripts for nucleus-encoded proteins in the same respiratory complex (Fig. 5). While transcript abundance for both nuclear and mitochondrial genes was at its lowest at 0 h, several mitochondrial transcripts increased in abundance during stratification and peaked in expression between 1 h SL and 12 h SL, while the nucleus-encoded counterparts increased in abundance later and peaked at 24 h SL (Fig. 5; Supplemental Table S3). Furthermore, the protein abundance for nucleus-encoded electron transport chain components of complex I and complex III also peaked at 24/48 h SL, as did their corresponding transcripts (Fig. 4; Supplemental Fig. S3). Thus, it appears that the early increase in expression observed for nuclear genes encoding mitochondrial proteins associated with RNA transcription, splicing, and editing has functional significance, as the transcript abundance of mitochondrial transcripts increases in a similar manner and pattern over the same time period. Interestingly, the protein abundance of a mitochondrial PPR protein (At5g46460) also showed this transient expression, peaking in abundance at 48 h S before decreasing over time (Fig. 2). Notably, the mitochondria-encoded transcripts generally did not decrease in abundance, unlike the nucleus-encoded transcripts in cluster 3 (Figs. 3 and 5; Supplemental Table S3).
A corresponding survey of the chloroplast-encoded components of the photosynthetic machinery and their nucleus-encoded counterparts was also carried out for comparison (Supplemental Fig. S4). It was observed that the nuclear and plastid transcripts differed in expression at 0 h, as the nuclear transcripts were all lowly expressed or absent, while most of the chloroplast transcripts were moderately to highly expressed (Supplemental Fig. S4). However, by 6 h SL, the abundance of plastid transcripts was reduced to levels similar to those of their nuclear counterparts and then proceeded to increase in abundance in a very similar manner to the nucleus-encoded components, with all transcripts peaking in expression at 48 h SL (Supplemental Fig. S4).
In order to further investigate the relationship between the nucleus-encoded mitochondrial proteins involved in RNA metabolism and their organelle-encoded targets, 261 PPR proteins known or predicted to be targeted to the mitochondria were compiled (Supplemental Table S4). Hierarchical clustering of these revealed that a significant percentage of this set of PPR proteins showed transient expression, peaking in expression between 48 h S and 6 h SL (62% versus 15% in the genome; P < 0.01; Fig. 6A). A number of studies have characterized the interactions between specific Arabidopsis PPR proteins and their organelle-encoded targets (Chateigner-Boutin et al., 2011; Hammani et al., 2011; Hölzle et al., 2011; Jonietz et al., 2011; Verbitskiy et al., 2011; for review, see Delannoy et al., 2007; Fujii and Small, 2011). These studies were used as a basis to investigate specific regulatory events of mitochondrial RNA processing during the germination time course (Fig. 6B). When the transcript abundance for the PPR proteins and their targets were plotted together, it became clear that the PPR transcripts increased in abundance transiently before or coincident with their targets, but never after (Fig. 6B). Interestingly, the majority of PPR proteins known to bind just one transcript had a profile characterized by a single peak (e.g. MEF8, MEF22, and MEF18; Takenaka et al., 2010), while PPR proteins with multiple known targets had profiles punctuated with multiple peaks (e.g. MEF11 [Takenaka et al., 2010] and MEF1 [Zehrmann et al., 2009]; Fig. 6B). These concordant profiles were also observed when transcripts encoding chloroplastic PPR proteins were profiled alongside their organelle-encoded transcripts (Supplemental Fig. S5). Thus, it appears that the expression of nuclear genes encoding proteins implicated in RNA processing in mitochondria (and plastids) occurs just prior to, or coincident with, changes in the target organelle transcript abundance. This close correspondence may in some cases be due to a causative relationship: some PPR proteins have been shown to directly control RNA stability and accumulation (Prikryl et al., 2011).
Profiling mitochondria-targeted PPRs and their known organelle-encoded transcript targets. A, Hierarchical clustering of the relative expression levels of 261 genes encoding PPR proteins targeted to the mitochondria. Of the four clusters identified, cluster 3, which is characterized by transient expression, was significantly (P < 0.001) enriched (62%) when compared with the total genome (15%). B, PPR proteins that have been characterized previously in the literature were normalized alongside qRT-PCR data showing expression of their known mitochondria-encoded RNA targets.
In addition to the genes encoding PPR proteins, it is clear that a significant number of other nuclear genes encoding mitochondrial proteins display a transient expression pattern, peaking in abundance at 48 h S before decreasing in abundance after 6 h SL (cluster 3; Fig. 3; Supplemental Fig. S1B). Notably, this group included the mitochondrial components of the purine biosynthesis pathway, phosphoribosylaminoimidazole synthase (At3g55010) and phosphoribosylformylglycinamidine amidotransferase (At1g74260), as well as a series of mitochondrial carriers that were recently predicted to be nucleotide carriers based on sequence consensus analysis across organisms (Palmieri et al., 2011). This suggests a coordination of mitochondrial nucleotide synthesis and nucleotide transport with this early RNA metabolism phase of mitochondrial biogenesis initiation.
Considering the potential role of this transient phase in the development of functional mitochondria, we sought to gain an insight into the importance of the roles played by the 239 genes highlighted through a survey of the genetic impact of their loss on Arabidopsis growth, development, and morphology (cluster 3; Supplemental Fig. S1B). To do this, a number of databases and large-scale forward and reverse genetic studies were queried, identifying phenotypes for knocked out/silenced genes. The databases included the SeedGenes database, which documents phenotypes for all genes showing embryo-lethal/defective phenotypes (Meinke et al., 2008); the Chloroplast 2010 Project, which reveals phenotypes for genes predicted to be localized to the chloroplast (Ajjawi et al., 2010); and the Agrikola database (Hilson et al., 2004). Furthermore, these were supplemented with individual gene searches in The Arabidopsis Information Resource (TAIR; Swarbreck et al., 2008). Remarkably, of the 239 genes queried, one in three (78 genes) had a previously observed (and published) altered phenotype (Fig. 7; Supplemental Table S5). Furthermore, the finding that nearly 40% of these showed embryo-arrested/lethal phenotypes is a highly significant overrepresentation (P < 0.01) in comparison with the expected number in the genome (Berg et al., 2005; Bryant et al., 2011). The number of developmentally impaired phenotypes (specifically embryo lethal; Fig. 7) suggests that the role of these proteins is highly significant for normal plant development. Taken together with the observed synchronous expression seen for these genes during germination, this suggests that these genes not only carry out crucial mitochondrial functions but that their expression represents a crucial control point during germination and seedling establishment through the linkage of nucleotide synthesis and transport with RNA metabolism.
Phenotypes observed in a transiently expressed set of genes encoding mitochondrial proteins during seed germination. Of the 239 transiently expressed genes (Supplemental Fig. S1B; cluster 3), one in three (78 genes) were observed to have a previously published phenotype when silenced or knocked out (Supplemental Table S5). Of these, 37% presented an embryo-arrested/lethal phenotype, which is three times greater than the expected number of embryo-arrested/lethal phenotypes for the whole genome. Other categories are as follows: leaf shape/development/amino acid content/fatty acid content/polysaccharide (AA/FA/PS) content, whole plant affected, developmental, flower development/timing/siliques, reduced/altered hypocotyl and/or root growth, and seed pigment/yield/amino acid content/fatty acid content (for details, see Supplemental Table S5). WT, Wild type.
DISCUSSION
Mitochondrial populations in dormant dry seeds represent the minimum structure (promitochondria) required to produce mitochondria and are essentially in a state of synchronized dormancy with each other. The rapid development of mitochondria following imbibition (the passive uptake of water by the seed) allows the synchronous maturation of mitochondria to be investigated (Logan et al., 2001; Howell et al., 2006). Previous studies on mitochondrial biogenesis during seed germination in maize have detailed the transition from promitochondria to mature mitochondria that contain typical cristae morphology (Logan et al., 2001), but the regulatory and developmental processes that drive this maturation are unknown. In order to uncover the earliest events in mitochondrial activation during germination, we carried out transcript and protein profiling of Arabidopsis seeds from freshly harvested seed (collected on the day from mature siliques) through stratification (at 4°C in the dark on solid Murashige and Skoog [MS] medium) and then imbibition in light for 48 h. In parallel with this, tagging of organelles with fluorescent marker proteins was used to monitor mitochondrial biogenesis. Organelle transcript abundance changes were also monitored to investigate the coordination of gene expression between the organelle and nuclear genomes.
Combining these investigations revealed that mitochondrial RNA and DNA metabolism, in terms of transcription, translation, and replication, represent the earliest mitochondrial functions that are reestablished during the transition from promitochondria to metabolically active mitochondria during germination (Fig. 8). The activation of DNA and RNA functions is supported by the early peak in expression of factors involved in cytosolic nucleotide metabolism and mitochondrial transporters that supply nucleotides to the mitochondrial matrix (Fig. 8). There are two sources of evidence that this peak in transcripts encoding mitochondrial DNA- and RNA-related processes is functionally important. First, this peak precedes the rise in mitochondrial transcript abundance, which in turn precedes the rise in transcripts for nucleus-encoded components of mitochondrial metabolism (Figs. 4 and 5). Second, mutational analysis of the genes that encode these mitochondrial proteins shows that inactivation leads to either complete embryo arrest or slower development (Fig. 7; Meinke et al., 2008). Thus, this peak in transcript abundance for genes encoding mitochondrial proteins involved in DNA replication and transcription is essential for normal development. Following this “biogenesis” state, the mitochondria accumulate the apparatus necessary for energy production and enter what we have termed a “metabolic” state (Fig. 8), which represents the mature mitochondria. Interestingly, the transcript abundance profiles of the components of the electron transport chain particularly displayed a high level of similarity with the corresponding protein abundances (derived from quantitative western-blot data and peptide mass spectroscopy), with an accumulation of these components toward the final time points, peaking in abundance at 24 and 48 h SL (Fig. 8, green asterisk). In contrast, mitochondrial import components were seen to exhibit dissimilar patterns between transcript and protein accumulation (Fig. 8); however, this was expected, given that low correlations between transcript and protein abundance for these components have been previously observed both in Arabidopsis (Lister et al., 2007) and rice (Howell et al., 2006).
A model for mitochondrial biogenesis during germination. Mitochondria in dry seeds lack cristae structures (Stasis). The earliest events that occur in mitochondrial biogenesis occur at the end of stratification and the first hour of transfer to light and are associated with RNA transcription and translation, leading to an increase in the transcript abundance of mitochondria-encoded genes. The next stage, which occurs at 24 h after transfer to light, is the increase in the transcript abundance of genes encoding various metabolic components. For a number of transcripts, one or more proteins were quantitated. The stages at which the lowest/highest corresponding protein abundances are observed are indicated with black/green asterisks, respectively.
The phasic nature of mitochondrial biogenesis outlined in this study offers an intriguing insight into the role of the host over the endosymbiont. Being of prokaryotic ancestry, it would not be surprising to observe similarities in the transition from promitochondrion to mature mitochondrion with that of the germination of a bacterial endospore. Both structures represent internalized bodies of a bacterial nature, and both undergo a period of stasis prior to an “activation event.” However, in contrast to promitochondria, the dormant endospore is in fact already replete with the necessary apparatus required for germination, and activation is a biophysical process without major macromolecular synthesis (Moir, 2006). This deviation in molecular maturation serves to emphasize the crucial role of the eukaryotic host cell.
Investigation of germination offers an excellent biological system for studying the coordination of a variety of processes, given the rapid, wholesale changes observed in a strict temporal progression. In this study, we have examined correlations between nuclear and mitochondrial transcripts, correlations between putative regulators of organelle gene expression and their target transcripts, and correlations between transcript and protein abundance. Analysis of protein abundance using western-blot analysis and mass spectrometry revealed a number of positive correlations between transcript and protein abundance (Fig. 2). In many cases where negative correlations were observed, it was simply due to a temporal dislocation (i.e. a period of lag between transcript abundance and protein abundance). It has been previously shown in other systems that the degree of correlation between transcript and protein abundance increases if time offsets are used (Irmler et al., 2008). Thus, the negative correlation observed for cruciferin (Fig. 2) is possibly due to the fact that while transcript abundance has peaked at 48 h SL, the encoded protein has not yet accumulated. Similarly, for mitochondrial ASP1 (At2g30970) and AAC2 (At5g13490), it was seen that the transcript level increased from 48 h S to 6 h SL and then decreased/stabilized at 48 h SL, while the corresponding proteins were still absent at 48 h S and only increased in abundance from 6 h SL before peaking at 48 h SL (Fig. 2; Supplemental Table S1). Given the technical limitations of obtaining quantitative protein abundance, these data revealed (in this system at least) that when active cellular synthesis and growth are occurring, transcript abundance is likely to be the primary level of control. A recent study investigating the plastid-specific transcriptional program of germination found that during stratification and germination, transcript and protein accumulation are not concordant and postulated that plastid transcription was more important for germination than translation (Demarsy et al., 2012). It should be noted that only semiquantitative macroarrays were used in that study, time points were much farther apart than in our work (dry seeds, imbibed seeds, and 2-, 4-, and 6-d-old plantlets), and a 16-h-light/8-h-dark cycle was used, which imposes a daily modulation on top of the germination process. Considering these differences, particularly the light/dark cycle, we would expect to see differences between that study and ours. Interestingly, while it was noted that specific inhibition of plastid translation with lincomycin did not prevent germination, it was seen that the inhibition of both plastid and mitochondrial translation with chloramphenicol completely prevents it (Demarsy et al., 2012), further emphasizing the integral role of mitochondrial biogenesis during germination.
The analysis of nuclear and mitochondrial transcripts for proteins located in the same respiratory chain complex revealed that the peak in transcript abundance occurs at different times. In fact, by 48 h S, mitochondria-encoded transcripts had peaked in abundance, whereas nucleus-encoded transcripts had not yet significantly changed in abundance compared with the harvested seeds (Fig. 5). Interestingly, even at the protein level, nucleus-encoded proteins (complex I subunit Ndufs4 [At5g67590] and cytochrome c subunit [At1g22840]) increased almost 10-fold in abundance from dry seed to 48 h SL, in contrast to the mitochondrially encoded complex I protein NAD9 (AtMg00070), which only increased by more than 2-fold in abundance from dry seed to 48 h SL (data not shown). This apparent lack of coordination between mitochondrial and nuclear transcripts has been previously observed in mitochondrial biogenesis in Arabidopsis during sugar starvation and refeeding (Giegé et al., 2005).
This study has also allowed an investigation into the time course of the accumulation of mitochondrial transcripts compared with that of transcripts encoding proteins involved in processing these mitochondrial transcripts (Fig. 6). The expression of genes encoding PPR proteins during germination is strikingly enriched in cluster 3, and it was revealed that the transcript encoding a protein involved in transcript maturation, such as editing, always peaked in abundance prior to or coincident with the target transcript for those with known targets (Fig. 6). It was shown that the first hour of stratification represents a phase of acute activity, where these genes are expressed and likely active, which helps to explain the lethal or severe growth phenotypes observed when these genes are inactivated (Fig. 7; Meinke et al., 2008). This rapid response is particularly intriguing in light of the seed’s capacity to cycle between different depths of dormancy for months or even years at a time in anticipation of environmental conditions suitable for seedling establishment. As yet, the molecular regulation of dormancy cycling remains a mystery, although a number of studies have identified a series of possible mechanisms (Cadman et al., 2006; Holdsworth et al., 2008; Footitt et al., 2011).
The regulation of nuclear genes encoding mitochondrial proteins is often divided into anterograde and retrograde regulatory pathways (Millar et al., 2011). While retrograde pathways and signaling are intensively studied under stimulation via various stresses that induce changes in transcript and protein abundance under stress (Wessel and Flügge, 1984), little is known about the retrograde pathways that exist to regulate the expression of the majority of mitochondrial proteins involved in carbon metabolism and respiration during normal growth and development, or even if such pathways exist. One possibility is that during seed germination, no communication takes place and that as soon as the conditions for germination are satisfied, all events commence in essentially a preprogrammed manner limited only by biophysical constraints and the availability of substrates to achieve germination and seedling establishment. While this model could not account for temporal differences in transcript abundance observed between mitochondria- and nucleus-encoded transcripts, it is harder to explain the temporal difference observed for nucleus-encoded transcripts that encode different functions.
Another possibility is that retrograde signals are sent from mitochondria to the nucleus to coordinate gene expression that ensures that the germination process only proceeds when different metabolic preconditions have been met in a series of checkpoints. While the existence and nature of such signals remain to be defined, it has been reported that perturbation of both mitochondrial and plastid translation by inactivation of the dual targeted, nucleus-encoded prolyl-tRNA synthetase 1 results in a retrograde response, with leaky mutants of this gene shown to result in the differential expression of genes encoding proteins involved in photosynthetic functions (Pesaresi et al., 2006). Furthermore, plastidial gene expression represents an important group of retrograde signaling from plastids to the nucleus, indicating that organelle gene expression can act as a retrograde signal (Pfannschmidt, 2010). Thus, it is postulated in this study that the early burst in the transcript abundance of mitochondrial genes uncovered here may be acting as a signal to prompt the expression of nuclear genes encoding proteins involved in various metabolic functions required for germination and seedling establishment. However, intensive follow-up studies are necessary to validate this hypothesis. Overall, it appears that while mitochondrial functions appear to be under developmental control during the earliest stages of germination (i.e. the activation of mitochondrial transcription and translation), the next phase of gene expression may require signals from this preceding event before these responses can occur and germination can progress to seedling establishment. Thus, it is proposed that nucleotide synthesis and transport, and the concomitant mitochondrial transcription and translation, during early germination are essential steps to mediate the transition from promitochondria into mature mitochondria (Logan et al., 2001).
CONCLUSION
The transition of promitochondria to mature mitochondria displaying high metabolic activity is characterized by a burst in the expression of genes encoding functions in nucleotide biosynthesis and transport as well as in organelle RNA- and DNA-related functions. This switch occurs prior to the increase in mitochondrial numbers, prior to the assumption of typical oval mitochondrial morphology, and is probably essential for germination to occur. While this switch has not been identified in other plant species studies to date, it provides a developmental target that may be used to ensure that premature germination or sprouting does not occur in cereals, and it may also represent the earliest time points that putative signals potentially originating from the mitochondria could signal the nucleus to proceed with mitochondrial biogenesis.
MATERIALS AND METHODS
Transformation of Arabidopsis Ecotype Columbia with Organelle-Targeted Fluorescent Proteins
A stably transformed line of Arabidopsis (Arabidopsis thaliana ecotype Columbia [Col-0]) seed, expressing three distinct organelle-targeted fluorescent proteins, was generated via Agrobacterium tumefaciens-mediated transformation. The first construct consisted of the 42-amino-acid mitochondrial targeting signal of alternative oxidase fused to GFP (Carrie et al., 2007) and cloned into pCambia 1301 in place of GUS. The second construct consisted of the full-length cDNA of the plastid-targeted small subunit of Rubisco (Rubisco SSU) fused to RFP (Carrie et al., 2007) and cloned into the pGreenII expression cassette (Hellens et al., 2000). The final construct consisted of the peroxisomal targeting signal 1 fused to the C terminus of CFP in the binary plasmid pFGC (Nelson et al., 2007). These constructs were stably transformed by three consecutive rounds of Agrobacterium-mediated transformation utilizing the floral dipping method outlined by Clough and Bent (1998). Following the floral dipping, seeds were harvested and screened using antibiotic selection, with hygromycin for the GFP construct, kanamycin for the RFP construct, and glufosinate for the CFP construct.
Confocal Microscopy to Observe Organelle Biogenesis during a Germination Time Course
The transformed seeds were surface sterilized by treatment with 0.05% (v/v) Triton X-100 in 70% (v/v) ethanol followed by successive washes with 70% (v/v) ethanol and 100% (v/v) ethanol. The seeds were then plated on Arabidopsis solid medium, containing 4.3 g of Gamborg’s B5 basal medium (Austratec), 0.5 g of MES, and 30 g of Suc, made up to 1 L with double distilled water. The seeds were then stratified for 48 h at 4°C in the dark, followed by an additional 48 h in continuous light (100 μE m−2 s−1) at 23°C. Seeds were collected throughout this time course, and the embryos were excised from the surrounding endosperm and testa by submerging in water and applying pressure with a coverslip. GFP, RFP, and CFP localization was monitored using a Leica TCS SP2 multiphoton confocal microscope, with excitation wavelengths of 410/440 nm (CFP), 460/480 nm (GFP), and 535/555 nm (RFP) and emission wavelengths of 460/480 nm (CFP), 500 to 540 nm (GFP), and 570 to 625 nm (RFP). Images were analyzed using ImageJ image-processing software (Abramoff et al., 2004).
Arabidopsis Tissue Collection for Transcript and Protein Analysis
Approximately 80 and 150 mg of Arabidopsis (Col-0) seeds was surface sterilized as outlined above and sown on Arabidopsis solid MS medium (3% Suc) for transcript and protein analysis, respectively. Sampling was carried out at 10 time points: freshly harvested seed (H) on the day of collection from plants, seeds following 15 d of dark, dry, ripening (0 h), seeds after 1 h of stratification (imbibition of seeds on Arabidopsis MS medium at 4°C in the dark; 1 h S), 12 h of stratification (12 h S), 48 h of stratification (48 h S), and on plates transferred to continuous light and collected 1 h into the light (1 h SL), 6 h into the light (6 h SL), 12 h into the light (12 h SL), 24 h into the light (24 h SL), and 48 h into the light (48 h SL; sampling as described by Narsai et al. [2011]). Collections were repeated three times for three biological replicates.
RNA Isolation, qRT-PCR, and Microarray Analysis
The Ambion Plant RNA isolation aid and RNAqeous RNA isolation kit were used for effective isolation of RNA from all samples collected. qRT-PCR was carried out using the LightCycler 480 system (Roche). SYBR Green I was the selected assay detection format, as carried out previously (Delannoy et al., 2009). The primer sequences used in this study have been outlined (Delannoy et al., 2009; Kühn et al., 2009). For microarray experiments, 400 ng of total RNA was used as the starting amount of RNA for the ATH1 Arabidopsis genome expression array; for details, see Narsai et al. (2011). In order to view the profiles of expression changes over the time course, all GC-robust multiarray average was made relative to the maximum intensity over the time course. For Supplemental Figure S1, these data were then hierarchically clustered using average linkage clustering (based on Euclidean distance) in Partek Genomics Suite version 6.5.
MapMan Analysis
To visualize the shifting patterns of the abundance of transcripts encoding proteins targeted to the mitochondria, transcripts were mapped into a custom mitochondrial MapMan pathway (Usadel et al., 2005), with 1,059 of the total present set (15,789) being classified into nontrivial MapMan BINS. The transcript abundance of each gene was scaled to range between −1 and 1, and values were averaged across each BIN.
Compiling Lists of Organellar Proteins
To compare the changing patterns of transcript abundance between the mitochondria, the chloroplast, and the peroxisome, it was necessary to generate lists of transcripts encoding proteins targeted to these organelles. To make these lists as comparable as possible, a consistent set of criteria was employed, using SUBA (Heazlewood et al., 2007). Based on the output from SUBA, next to each gene in each organelle list a source number is indicated, where 1 refers to an experimentally determined localization by either mass spectrometry or GFP analysis, 2 refers to a computationally predicted localization based on protein sequence (however, for a gene to be denoted as 2, at least five predictors needed to have predicted to the same localization), and 3 refers to a protein localization determined by both experimental analysis and prediction. Given the limited experimental annotation for peroxisomal localization and the limited number of peroxisomal predictors (PeroxP [Emanuelsson et al., 2007] and WoLF PSORT [Horton et al., 2007]), the peroxisomal list was augmented with genes from the AraPerox database (Reumann et al., 2004), and these were denoted as 4. The predictor programs utilized were Predotar (Small et al., 2004), TargetP (Emanuelsson et al., 2007), WoLF PSORT (Horton et al., 2007), MitoProt2 (Claros and Vincens, 1996), SubLoc (Hua and Sun, 2001), iPSORT (Bannai et al., 2002), MITOPRED (Guda et al., 2004) PeroxP (Emanuelsson et al., 2003), MultiLoc (Höglund et al., 2006), and LOCtree (Nair and Rost, 2005).
To produce a comprehensive list of genes encoding proteins targeted to the mitochondria, a custom list of all currently defined genes encoding proteins located in the mitochondria in Arabidopsis was manually compiled and annotated based on a number of functional classifications: carrier proteins and import components of the mitochondrial outer and inner membrane, constituents of the electron transport chain, prohibitins, generation enzymes, RNA editing/splicing, RNA polymerase and organelle-encoded genes, transcription factors, ribosomal proteins, translation factors, tRNAs, assembly, ascorbate/glutathione, sulfur metabolism, heme, folate, and photorespiration. For defining lists of mitochondrial proteins involved in amino acid metabolism, cytosolic glycolysis, and the TCA cycle, annotations were based on MapMan ATH1 TAIR9 annotation BIN files with localization, as these functional categories have been extensively defined previously. The annotated gene list for 945 mitochondria-located proteins was categorized into a BIN hierarchy for display on the custom mitochondria MapMan image according to the BIN structure shown in Supplemental Table S2. A full outline of the manual curation of this list, including all references, can be found in Supplemental Materials and Methods S1.
Phenotype Analysis of Transiently Expressed Genes
A subset of 239 transiently expressed genes encoding proteins targeted to the mitochondria was cross-referenced against a number of large-scale forward and reverse genetic studies, identifying phenotypes for knocked out/silenced genes. Databases used were the SeedGenes database (Meinke et al., 2008), the Chloroplast 2010 Project (Ajjawi et al., 2010), and the Agrikola database (Hilson et al., 2004). Furthermore, these databases were supplemented with manual database checking of TAIR (Swarbreck et al., 2008). Sources and full references for each phenotype are shown in Supplemental Table S5.
Total Protein Extraction from Seed, and Protein Quantification for SDS-PAGE and Western Blotting
Following collection, seeds were snap frozen with liquid nitrogen and ground with a mortar and pestle. The extraction was carried out according to the methodology outlined at http://www.seed-proteome.com. Total protein was extracted in triplicate, at 2°C, in 2.2 mL of lysis buffer containing 7 m urea (Bio-Rad), 2 m thiourea (AnalaR), 4% (w/v) CHAPS (Sigma-Aldrich), 18 mm Tris-HCl (Amresco), and 0.2% (v/v) Triton X-100 (Sigma-Aldrich). Following a 10-min incubation at 4°C, 14 mm dithiothreitol was added. The protein extracts were then stirred for 20 min at 4°C, followed by rounds of centrifugation at 35,000g for 10 min at 4°C. Protein concentrations were then measured in accordance with a methodology adapted from Schaffner and Weissmann (1973). For each sample, 4 μL of protein extract was diluted with 25 μL of double distilled water, followed by the addition of 300 μL of Amido black (taken from a stock solution containing 26 mg of Amido black in 100 mL of acetic acid:MetOH [1:10]). This mixture was then mixed by vortexing and incubated for 5 min at room temperature, followed by 4 min of centrifugation at 13,000 rpm at room temperature. Once the supernatant had been removed, the pellet was washed with 500 μL of acetic acid:MetOH (1:10), followed by a second round of centrifugation. Then, the dyed protein pellet was resuspended in 1 mL of 0.1 m NaOH and the absorbance was measured at 615 nm on a UV-VIS spectrophotometer (Shimadzu). A standard curve derived from eight dilutions of a bovine serum albumin standard quantitated in the same manner was then used to quantitate the samples.
SDS-PAGE and Western Blotting
SDS-PAGE was utilized to separate 30 μg of total protein extract from each time point, alongside a purified mitochondrial sample (Supplemental Fig. S6), which was then transferred to a supported polyvinylidene difluoride membrane (Millipore) in Towbin buffer (Towbin et al., 1979), at 30 V for 16 h, using a Trans-Blot Plus Electrophoretic Transfer Cell (Bio-Rad). Probing of proteins of interest was undertaken using an array of different antibodies that have been described previously or purchased from Agrisera. The VDAC protein was obtained from Dr. Tom Elthon (University of Nebraska). The antibody for Kat2 has been described previously (Footitt et al., 2007). The antibodies for ATP synthase, RbcS, RbcL, V-ATPase, and UGPase were purchased from Agrisera. Blots were visualized by chemiluminescence (Roche Applied Science) using an ImageQuant RT ECL Imager (GE Healthcare). The western-blot band intensities were subsequently quantitated using ImageQuant TL software (GE Healthcare).
Mass Spectrometry Identification of Proteins Expressed during Germination
Following collection, seeds were snap frozen with liquid nitrogen and ground to a powder in a Retsch ball mill. The extraction was carried out based on Wessel and Flügge (1984) and carried out in triplicate. Briefly, the powder was resuspended in SDS-Tris grinding buffer (500 mm Tris, pH 7.0, 7% [w/v] SDS, and 10% [v/v] β-mercaptoethanol), and the solution was spun at 10,000g for 5 min. To the supernatant were added 4 volumes of methanol, 1 volume of chloroform, and 1 volume water; it was then vortexed and spun at 10,000g for 5 min. The protein interface was collected and washed in 500 μL of methanol, and the pellet was air dried. The pellet was then resuspended in 20% acetonitrile and 10 mm NH4HCO3 and mixed for 1 min at 30 Hz in a Retsch ball mill at room temperature. The resulting solution was spun at 2,000g for 5 min, and the resulting supernatant was collected. The protein content was determined by a modified Bradford assay using bovine serum albumin as a standard (Peterson, 1977). A total of 250 μg was then digested and run on the mass spectrometer as outlined previously (Taylor et al., 2010).
Output files from quantitative-time of flight data were conducted using the Mascot search engine version 2.3.02 (Matrix Science) and the TAIR9 database (June 19, 2009; 3,3621 sequences and 13,487,170 residues) utilizing error tolerances of ±100 ppm for mass spectrometry and ± 0.5 D for tandem mass spectrometry, Max Missed Cleavages set to 1, Oxidation (M) and Carboxymethyl (C) variable modifications, the instrument set to ESI-Q-TOF, and Peptide Charge set at 2+ and 3+. Output files for each triplicate treatment were then aligned, including the protein MOWSE score, peptide matches, and the Mascot-calculated exponentially modified protein abundance index values based on Ishihama et al. (2005). Protein content was calculated using the method of Ishihama et al. (2005) and shown as an average of triplicates with se (Supplemental Table S6.) Only proteins identified in two out of three biological replicates in at least one time point were used for further analysis (Fig. 2). In this way, 178 unique proteins were identified.
Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Hierarchical clustering of the transcript abundance of transcripts targeted to the three energy organelles.
Supplemental Figure S2. MapMan visualization of transcripts involved in mitochondrial processes during a germination time course.
Supplemental Figure S3. Protein abundance profiling of Arabidopsis seed during a germination time course.
Supplemental Figure S4. Coordination of nucleus- and organelle-encoded components of the chloroplast photosynthetic electron transport chain.
Supplemental Figure S5. Profiling of chloroplast-targeted PPRs and their known organelle-encoded transcript targets.
Supplemental Figure S6. Confirmation quantitation of total proteins extracted from Arabidopsis seeds during the germination time course.
Supplemental Table S1. Details of the 178 transcripts/proteins analyzed in parallel, shown in accordance with the order presented in Figure 2.
Supplemental Table S2. Mitochondrial lists of genes generated using SUBA through a combination of experimental evidence (mass spectroscopy and GFP) and predictive tools.
Supplemental Table S3. Average transcript abundance of mitochondria-encoded genes of the electron transport chain during germination, as shown in Figure 5.
Supplemental Table S4. Lists of PPR proteins targeted to either the mitochondria or the chloroplast.
Supplemental Table S5. Data for each gene that was found to have a phenotype, details showing the number of genes showing known phenotypes, and full references for these genes.
Supplemental Table S6. Abundance of 178 proteins calculated using the exponentially modified protein abundance index during germination.
Supplemental Materials and Methods S1. A custom list of all currently defined genes encoding proteins located in the mitochondria in Arabidopsis that was manually compiled and annotated based on a number of studies.
Footnotes
The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: James Whelan (jim.whelan{at}uwa.edu.au).
↵1 This work was supported by the Australian Research Council Centre of Excellence (grant no. CEO561495) and the Western Australian State Government Centres of Excellence scheme.
↵2 These authors contributed equally to the article.
↵3 Present address: Functional Genomics of Arabidopsis, Unité de Recherche en Génomique Végétale, INRA, Evry 91057, France.
↵[OA] Open Access articles can be viewed online without a subscription.
↵[W] The online version of this article contains Web-only data.
- Received December 18, 2011.
- Accepted February 13, 2012.
- Published February 16, 2012.