The absence of recognizable promoter sequences on dinoflagellate genes and the presence of 5' SL on dinoflagellate gene transcripts [32, 33] suggest, by analogy with trypanosomes, the post-transcriptional control of gene expression. Consistent with this hypothesis, many physiological processes have been found to be regulated post-transcriptionally [26–31]. Similarly, microarray analysis of acute stress responses in K. brevis did not reveal the activation of stress genes  under conditions (1-4 hours) where they were shown to be induced at the protein level . Microarray analyses found only ~3% of transcripts changing over a circadian day in Pyrocystis lunula  and K. brevis . Yet in K. brevis ~10% of transcripts were differentially expressed over a light-dark cycle , a percentage that does not differ substantially from other photosynthetic eukaryotes that utilize typical transcriptional controls [51, 52]. Since microarrays report only changes in transcript abundance, it remains unresolved by what mechanism(s) these changes are achieved.
In this study, we explored in two different scenarios the response of the K. brevis transcriptome to nutrients. We first investigated the differences in the transcriptome between nutrient depleted cells and cells grown in nutrient replete media. This may be considered a chronic response in which the transcriptome has had an extended period of time to remodel, and therefore, changes in the transcriptome could result solely from differences in message stability. We found that the transcript profiles of N-depleted cells did indeed reflect their status by the increased expression levels of transcripts for N- acquisition. In contrast, evidence for P-limitation was not apparent from the transcriptome. We next investigated the acute response of the transcriptome of nutrient depleted cells to the addition of N or P over a short time course of 1-48 h following addition. The acute remodeling of the transcriptome within hours of nutrient addition was surprising, given the presence of the spliced leader on even the earliest responding transcripts. The resulting changes in transcript profile reflect primarily the rapid reactivation of the chloroplast metabolic machinery by nuclear encoded regulators. Changes in the expression of hallmark N- or P-acquisition genes were not observed upon addition of nitrogen or phosphorus to nutrient depleted cultures. To our knowledge there are currently no global transcriptional studies in phytoplankton following nutrient addition.
Transcriptome Responses to Nitrogen Starvation and Addition
The increased expression of putative ammonium transporter, nitrate transporter and glutamine synthetase transcripts observed in N-starved K. brevis is consistent with their roles in N-uptake and assimilation. Transcript abundance of all three of these genes was strongly up-regulated in the N-depleted stationary phase cultures relative to nutrient replete cultures in log phase of growth. All of the up-regulated glutamine synthetases had homology to glutamine synthetase III (glnN), while other glutamine synthetase probes present on the array with homology to glutamine synthetase I or II did not respond. In the haptophyte Isochrysis galbana ammonium and nitrate transporters are strongly up-regulated under N-starvation, along with glutamine synthetase II as measured by qRT-PCR . In contrast, glutamine synthetase II mRNA expression in the diatom Skeletonema costatum was low under N-starvation  and increased with NO3 addition following approximately a five day delay. Digital expression profiling of EST libraries of the diatoms Phaeodactylum and Thallassiosira revealed altered expression of numerous genes involved in nitrogen metabolism and regulatory elements in nitrogen starved cells . Among these, an ammonium transporter was the most highly up-regulated gene in an N-starved Phaeodactylum library. Ammonium transporters have similarly been shown to increase in expression in N-starved diatom Cylindrotheca . The nitrate reductase transcript in the diatom Cylindrotheca is expressed under N-starvation or in the presence of nitrate, but is inhibited by the presence of ammonium . Thus it appears that the regulation of these pathways may differ somewhat across different phytoplankton phyla.
Following the addition of NO3 to N-depleted cells, we did not observe changes in the expression of any N-uptake or assimilation genes within the first 48 h. The responses in K. brevis were instead enriched in several processes including plastid functions, ribosomes, nitrogen compound metabolism, and amino acid biosynthesis. Members of these gene ontologies were also responsive in a microarray study of the Arabidopsis transcriptomic response at 2 and 24 h following N-addition . In both studies, transcript levels for ribosome and plastid functions increased by 24 h. Confoundingly, transcripts belonging to the GO functions of nitrogen compound metabolism and amino acid biosynthesis differed in their direction of change, where their abundances increased in Arabidopsis by 24 h, but decreased in K. brevis over the 48 h time course. Specifically, in K. brevis genes involved in the synthesis of arginine, methionine, and threonine, such as homoserine dehydrogenase and ornithine carbamoyltransferase are down-regulated. These genes were not up-regulated under N-depletion so the observed down-regulation does not appear to be a return to levels expressed in N-replete log phase cultures. In contrast, genes involved in cysteine biosynthesis, including catechol-o-methyltransferase and cystathionine beta-synthase were increasingly up-regulated 1.76-1.90 fold starting at 12 h following N-addition. A possible explanation is that genes involved in the synthesis of amino acids exhibit an overall down-regulation during N-limitation. Upon addition of a N source biosynthesis of amino acids initiates and, as the sulfate needed to produce methionine is derived from cysteine , it is possible that the expression of genes involved in methionine biosynthesis will increase only after a sufficient pool of cysteine is produced as a result of increased nitrogen levels. The regulation of nitrogen and sulfur assimilation pathways has been shown to be tightly coupled in many other plants and algae including the green alga Chlamydomonas reinhardtii  with nitrogen limitation causing a down-regulation in genes involved in sulfate assimilation as observed in this study.
Transcriptome Responses to Phosphorus Starvation and Addition
P-depleted cells showed little consistent indication of P-starvation through the transcript levels of genes putatively involved in P- uptake or utilization, despite the strong evidence based on the growth response to P-addition. Under P-stress, ATP pools are significantly reduced, affecting nearly all metabolic processes, including DNA, RNA, and phospholipid biosynthesis, as well as regulatory phosphorylation of proteins and generation of phosphorylated intermediates for photosynthetic carbon fixation. Plastid inorganic pyrophosphatases and plastid phosphate translocators are important mechanisms for recycling PPi needed for regenerating ATP used for CO2 fixation. We also queried acid phosphatases and vacuolar type H+-translocating inorganic pyrophosphatases, which in higher plants and Chlamydomonas increase in both expression and activity under P-starvation, thereby providing alternative energy sources to the limited ATP pools available under P-starved conditions . These probes showed mixed responses to P-starvation in K. brevis. Lastly, alkaline phosphatase, whose activity is often used as an indicator of phosphate stress in phytoplankton, showed no response at the transcript level. Alkaline phosphatase enzyme activity has been shown to be induced in K. brevis under similar low phosphate (1 μM) conditions . By comparison, in the coccolithophore Emiliania huxleyi, alkaline phosphatase transcripts are dramatically induced by phosphate starvation and rapidly repressed after phosphate addition . The absence of any changes in transcript levels in the current study suggests this activity may be regulated at a translational or post-translational level, which is consistent with the presence of the SL mechanism.
Following P-addition, the transcriptome response was enriched in GO categories that include ribosome constituents, RNA binding, plastid, and electron transfer functions. As in the response to N-addition, the earliest changes were dominated by the increase in transcripts for PPR proteins that in the P study were measurable as early as 1 h following P-addition. However, in marked contrast with the response to N-addition, the ribosomal and chloroplast functions were strongly down-regulated by 24-48 h following P-addition. The reason for the disparity in response of these transcripts to N- and P-addition is unknown. It has been shown in yeast that the initiation of ribosome biogenesis is tied to a critical cell size that is controlled by nutrient signals . While cell size was not measured in this study, N-limitation has been reported to decrease cell size while P-limitation increases cell size in other dinoflagellate and algal species [64, 65]. Thus, the opposing responses of these genes may reflect complex differences in the physiological status of N- and P-starved cells that will require further investigation.
The early responses of the transcriptome to both N and P were dominated by increases in transcripts for PPR proteins. Of the array features that responded significantly to nutrient addition, 29 and 25 annotated features of the N- and P- addition trend sets, respectively, were PPRs. This represents over 13% of annotated features in the trend sets, and approximately a quarter of the features annotated as PPRs on the array. PPR (pentatricopeptide repeat) proteins are a novel family of proteins first discovered when the Arabidopsis genome was sequenced, defined by a 35 amino acid (pentatricopeptide) motif that is repeated in tandem up to 30 times [66, 67]. Most PPR transcripts in Arabidopsis possess chloroplast (~25%) or mitochondrial (~75%) targeting sequences . PPR proteins are sequence-specific RNA binding proteins that, in plants, bind in a sequence-specific manner to unique organellar mRNAs, resulting in recruitment of enzyme complexes that modulate their expression through post-transcriptional processes, including editing, splicing, translation, and stability [67, 69, 70]. PPR proteins are absent from bacteria, but are present in all eukaryotes examined, generally at low copy number (5 and 6 in yeast and human, respectively ). Trypanosoma brucei is an exception among non-photosynthetic eukaryotes, encoding 28 unique PPR proteins [71, 72] that are essential for mitochondrial rRNA biogenesis and stability . Among the 11,000 unique genes in the K. brevis EST database approximately 100 are annotated as PPR proteins. Preliminary analysis of these PPRs contigs with ChloroP  has identified the presence of a chloroplast transit peptide on 40% of contigs from the EST database. Further, 100% of the contigs that have the 5' end (as defined by the presence of the splice leader) were found to have a chloroplast transit peptide in K. brevis.
The large representation of PPR proteins among the K. brevis transcripts responding to N- or P- addition was a driving force behind many of the enrichment categories involving the chloroplast and ribosomal proteins. Their peak expression (12-24 h following nutrient addition) generally preceded the changes in expression of the chloroplast encoded photosystem and electron transport genes, and ribosome and RNA binding proteins, which peaked at 48 h following N-addition. This temporal relationship, in addition to their known roles in organellar RNA processing, suggests a link between PPR transcript abundance and subsequent expression of chloroplast encoded genes.
The Photosystem and Photosynthetic Electron Transport Chain
A limited number of plastid encoded genes involved in the photosynthetic electron transport chain are present on the array and were among the strongest responding transcripts to both N- and P-addition. The interpretation of chloroplast-encoded gene expression using the microarray requires special consideration because the oligo(dT) primed RNA labeling methods employed depend on the presence of a polyA tail, but in the chloroplast, polyadenylation serves as a signal for RNA degradation . Therefore, an increase in transcript abundance of a chloroplast-encoded gene on the array may reflect increased polyadenylation, or destabilization of the message pool, rather than increased expression. Conversely, a decrease in chloroplast encoded transcripts on the array may reflect their decreased polyadenylation, or increased mRNA stability. At 48 h following N-addition, Photosystem I (p700 chlorophyll a apoproteins A1 and A2), photosystem II (44 kDa, 47 kDa, D1, and D2 proteins) and ATP synthase (CF0 α and F0 subunit C) transcripts were strongly up-regulated on the array, likely indicating increased turnover of these chloroplast encoded genes. Down-regulation (e.g., stabilization) of the same genes was observed as K. brevis moves from log phase into stationary phase (Johnson JG, Morey JS, Neely MG, Ryan JC, Van Dolah FM: Transcriptome remodeling associated with chronological aging in the dinoflagellate Karenia brevis, submitted), so the increase following nutrient addition may reflect the transition back to active cellular division.
Although pigment analyses were not performed, greening of the cells was clearly visible within 12 hours of N-addition as pigment production was re-established in chlorotic cells; this process preceded the increase in polyadenylated transcript abundance for photosystem and ATPase genes. Thus the observed increase in polyadenylated plastid messages late in the recovery from N-starvation may reflect a subsequent turnover of messages. Consistent with this interpretation, during greening of Chlamydomonas reinhardtii following N-addition an increase in photosystem transcripts occurred within 2-3 hours, then returned to basal levels within 12 hours . In the current study, every one of the photosystem and ATP synthase transcripts that was up-regulated on the array following N-addition was down-regulated at 48 h following P-addition, possibly indicating increased message stability (i.e., decrease in polyadenylated messages). Although P-amended K. brevis cultures returned to growth with similar kinetics as those that received N-additions, the processes underlying this response clearly differ. No obvious greening of the cells occurred following P-addition as chlorosis did not occur in the P-limited cultures of K. brevis. In plants, both nitrogen and phosphorus limitations decrease photosynthesis but do so through different mechanisms. N-stress reduces photosynthesis directly by decreased light absorption capacity through diminished expression of photosystem protein complexes, whereas P-stress decreases rates of CO2 fixation through changes in the activity of Calvin cycle enzymes, both mechanisms resulting in further feedback inhibition of photosynthesis . In Chlamydomonas, removal of either sulfate or phosphate reduces photosynthesis, but whereas S-depletion decreases chloroplast transcript levels, P-depletion results in an increase in chloroplast RNA stability and abundance . The opposing responses of photosystem gene transcripts following the addition of N or P in the current study suggest differences also exist in the recovery from N- vs P-depletion at the level of chloroplast mRNA transcription and/or stability. A better understanding of this process will require targeted studies on K. brevis chloroplast regulation.
Post-transcriptional Control in K. brevis
Given the prevalence of PPR proteins in these data sets and their rapid increase in response to nutrient addition, we were particularly interested in determining if these genes are among those under SL control, or if they represent a class of dinoflagellate genes regulated by transcription. Using PCR with a SL specific primer and a gene specific primer, we identified the presence of the 5' SL cap on more than 40 unique PPR gene transcripts. This suggests that the up-regulation of PPRs in response to nutrient addition does not reflect typical eukaryotic transcriptional control. Changes observed in a global transcriptional profile could be the result of several different mechanisms, including differential rates of trans-splicing, differential mRNA stability, and changes in chromatin structure that modulate transcription.
In trypanosomes, transcriptional silencing of the spliced leader gene (the only gene under transcriptional control) occurs in response to stress, leading to depletion of SL available for trans-splicing and the accumulation of polycistronic messages [78, 79]. If this mechanism occurs in dinoflagellates, the recovery from nutrient starvation may be mediated by a rapid resumption in trans-splicing of polycistronic messages, which would generate an increased number of mature monocistronic mRNAs available for detection by the microarray without changes in transcription. However, gene expression in trypanosomes is further regulated by changes in RNA stability mediated by 3'UTR elements including U-rich instability elements (UREs), short interspersed degenerated retroposons (SIDERs), and RNA binding proteins . A clock-controlled RNA-binding protein has been identified in Chlamydomonas reinhardtii (Chlamy 1) and is involved in regulating nitrogen metabolism; however it is believed that it does so through inhibition of translation rather than impacting mRNA stability . Another RNA binding protein, circadian controlled translational regulator (CCTR), has been identified in the dinoflagellate Lingulodinium polyedrum and likewise regulates translation of luciferin binding protein . Increasingly, RNA binding proteins are recognized to play a role in mRNA stability in many systems, which has significant impacts on measurements of gene expression . A recent study in the toxic dinoflagellate Alexandrium catenella identified several RNA binding proteins proposed to be involved in RNA silencing and other post-transcriptional regulation mechanisms . The up-regulation of RNA binding proteins at 1 and 4 hours post-P-addition may similarly reflect a role in differential mRNA stability in K. brevis. It is also apparent that local changes in chromatin condensation can rapidly alter the availability of genes  and thereby modulate levels of transcription  even in the apparent absence of regulation by basal transcription factors in dinoflagellates.