Calvin cycle regulation in diatoms compared to the green lineage

It has been proposed that the regulation of the Calvin cycle in diatoms should be different to higher plants [19] due to significant differences in the genetic structure and the enzyme properties of the Calvin cycle and associated proteins. This assumption was based on the annotation of two complete diatom genomes (P. tricornutum and Thalassiosira pseudonana) [26, 30] and on several studies on Calvin cycle enzymes in diatoms. It has been shown that a classical redox regulated RuBisCO activase, like as known from plants, is missing [19].

Interestingly, a red algae AAA⁺ (ATPases associated with various cellular activities) type RuBisCO activase called CbbX has been identified recently, which can activate RuBisCO [88]. The mechanism proposed differs from higher plants and the CbbX protein is redox insensitive as well. A look into the P. tricornutum genome reveals a CbbX homologue (JGI protein ID 42728), which likely acts as redox insensitive RuBisCO activase. The study of Michels et al. [32] demonstrated that the diatom enzyme phosphoribulokinase (PRK) is in principal redox sensitive, but its redox midpoint potential was shifted in such extent that in vivo no redox regulation is expected to occur. The regulatory protein CP12, known to form a co-regulating inhibiting complex with plastidic glyceraldehyde-3-phosphate dehydrogenase (GAP-DH) and PRK in higher plants/green algae [38, 77, 230], is absent in the genomes of P. tricornutum and T. pseudonana [19] - only CP12-like proteins lacking the functionally important cysteine domains, have been found (Supplemental Figure 9-SI) and no indications for CP12 complexes were detectable in several diatoms [71]. There are some indications for a CP12 protein in the diatom Asterionella formosa [95] which apparently associates with the plastidic GAP-DH and not with PRK [96]. Only certain fructose-1,6-bisphosphatase (FBP) isoenzymes seem to be targets of redox regulation in diatoms [19, 32]. The preliminary results covered in chapter 4 support the assumption of redox sensitivity of P. tricornutum FBPs and the gel filtration and immunoblot data indicate that FBPs are part of multimeric protein complexes. Such complexes could enable the redox sensitive FBPs to transfer their active state allosterically to other enzymes, thus making them indirectly controlled by redox

regulation as well. Yet, taken together the redox regulation in diatoms apparently is limited in comparison to higher plants.

Thus the general mode of Calvin cycle regulation in diatoms has to be different. A possible reason for this apparently reduced regulation at the enzyme level via redox regulation in diatoms may be the different organisation of the metabolic pathways: in diatoms there is no short term risk of futile cycling of carbon compounds in the plastid as reductive and oxidative pentose phosphate pathways are spatially separated [19, 27, 32], making a very quick and immediate regulation of the Calvin cycle enzymes less important. Diatoms experience regularly quick changes of light and darkness in their natural marine environment by floating in the turbulent waters of the water column [231, 232]. This may be another aspect for a reduced redox regulation: a fast acting energy consuming regulation system like a redox control may under such conditions not be energy efficient for the diatom and thus was abolished during diatom evolution.

To elucidate alternative regulation systems of the Calvin cycle, the expression levels of the key enzymes that are controlled in higher plants via thioredoxin mediated redox regulation were investigated in this thesis. To this end a set of genes usable as endogenous controls for qPCR was established as described in chapter 2. Endogenous reference genes for qPCR in diatoms have been described before [60, 61], but the experimental setup applied in this thesis provide specific advantages. This is the first time that a set of endogenous reference genes for diatoms is described, which is verified to be stably expressed over time in a simulated day night cycle and in extended darkness of up to 33 hours. These properties turn the genes for hypoxanthine-guanine phosphoribosyltransferase (HPRT), TATA-box binding protein (TBP) and ribosomal protein S1 (RPS) into very valuable candidate genes for any light and time dependent transcript analysis in P. tricornutum and possibly other diatoms as well.

These reference genes were used for extensive transcript analyses of Calvin cycle genes as described in chapter 3. The transcript levels of most of the seven investigated Calvin cycle genes were changing rhythmically during the day and their expression was clearly light induced even if an underlying light independent diurnal rhythm was discernible in each case as well. The relative changes were mostly unspectacular being similar to what has been shown for plant relative transcript levels [44, 48], but the PRK and GAP C1 genes featured extremely strong relative changes of transcript levels during the day not described for these genes before. These huge relative changes are even more intriguing as our newly developed analysis strategy of relative qPCR data, PAR-qPCR, verified that these huge relative changes indeed resulted in huge absolute changes. These changes were several times larger than the observed maximum transcript levels of all other investigated Calvin cycle genes. Such a strong regulation virtually imposes the notion of an important transcriptional control, which exceeds simple daily rhythm adaptive expression. Interestingly GAP C1 and PRK are proteins which are highly co-regulated in the green lineage by formation of a complex with CP12, which inhibits their activity and only TRX mediated reduction of CP12 releases PRK and plastidic GAP-DH again [38, 77, 84, 230, 233]. As a functional CP12 is missing in P. tricornutum this may indicate that the observed enormous transcript regulations are in part substituting for this missing CP12 mediated co-regulation of PRK and GAP C1.

A recent microarray based study in P. tricornutum described the diurnal transcript levels of several Calvin cycle genes in a simulated day night cycle as well [79]. Regardless of some differences in experimental setup the results of this study principally are in accord to the

results of this thesis, describing light dependent diurnal rhythms peaking at similar times during the day to the results of our study. A distinct difference between the results of Chauton et al. and this thesis is that the extreme regulation of PRK and GAP C1 was not found by Chauton et al. This is likely a consequence of the limitations of microarray analysis and an unfortunate choice of the calibrating sample. Microarrays suffer from a more restricted dynamic range compared to qPCR. The hybridisation spots in a microarray only have a limited binding capacity, which can lead to saturation effects if their maximum binding capacity is reached [80]. The PAR-qPCR analysis presented in this thesis demonstrated that there are huge absolute transcript numbers of GAP C1 and PRK during their peak expression.

The choice of a late night sample taken shortly before light exposure as calibrator in the study by Chauton et al. additionally reduced sensitivity. At that time already huge numbers of GAP C1 transcripts exist, likely saturating the microarray spot. The increasing transcript levels of GAP C1 during the morning to afternoon described in chapter 3 of this thesis would result in the same signal strength due to the saturation, which would explain the observed stable transcript levels of GAP C1 during the day by Chauton et al. PRK is more dependent on light, thus shortly before the light the transcript levels were likely not yet saturating.

Accordingly, Chauton et al. were able to monitor a peak expression profile, even if not as strong as in this thesis. Thus the transcriptional data of this thesis complements and expands the Calvin cycle data of the microarray study of Chauton et al.

The protein levels of the three investigated Calvin cycle enzymes described in chapter 3 are principally in accord with the transcriptional levels. The protein levels of RBC L and PRK both follow the transcript levels, featuring low expression at complete darkness and a light induced daily rhythm in the simulated day night cycle. GAP C1 protein levels, however, are difficult to interpret and not obviously following the transcript levels. The relative changes are small and not homogenous within the replicates, showing no clear rhythm in any of the conditions (LD or DD). Immunoblotting is a relative quantification strategy usually missing endogenous reference proteins, due to limitation of available antibodies. Thus a protein which features a more or less constant protein level over all samples becomes difficult to analyse, as immunoblot quantification features a relatively high variability as was shown in supplemental figure 3-SI. This variability may result in small erratic relative changes like observed for GAP C1. The immunoblots of GAP C1 readily produced signals after short exposure times, which is supportive for a high abundance of the protein. Thus it is likely that the high transcript levels of GAP C1 in LD as well as DD conditions result in equally strong more or less constant protein levels that are not reduced during the day between transcript expression phases regardless of the light condition. This hypothesis will have to be proven in the future for example by establishing an endogenous reference protein for immunoblotting or by establishing a quantitative standard curve with purified GAP C1 protein. The putatively constant levels of GAP C1 would raise another important question: why do increasing transcript levels not lead to increased amounts of protein. A possible explanation would be that the GAP C1 protein pool is refreshed once per day by a concerted strong expression and degradation of the protein. Investigations of protein stability by pulse chase experiments [234, 235] may be promising to answer this question.

Another interesting aspect of GAP C1 expression is the fact that it is diurnally strongly expressed in total darkness as well. As the Calvin cycle is a strictly light dependent metabolic

the dark, as well. Multifunctional roles have been described for GAP-DH before even in uncommon functionalities which also gave rise to its classification as “moonlighting enzyme”

[81-84, 236]. It may be speculated that in the dark partial gluconeogenesis converts residual C3 carbon compounds and glycerol released from storage lipids to C6 carbohydrates, which in turn may be important for glycosylations or become substrate for the oxidative pentose phosphate pathway (OPP) in the dark potentially providing C5 carbohydrates and NADH/H⁺ for NADH/H⁺ dependent reactions and anabolic reactions like nucleotide synthesis or osmolyte generation. Due to the cytosolic localisation of the OPP in diatoms [27] such reactions would depend on an export of the C6 carbon compounds. Kroth et al. described a gene for a UDP-glucosyl pyrophosphorylase (UGP) in diatoms with a predicted plastidic localisation [29]. This enzyme can form UDP-glucose with phosphorylated glucose, which could be provided by gluconeogenic reactions in the plastid. This UDP-glucose could be transported by sugar nucleotide transporters like described for ancient plastids [237-239] and similar phosphate transporters of yet unknown specificity were described for chromalveolate secondary plastids [240]. Protein alignments for such sugar nucleotide transporters in P. tricornutum yield gene homologs with predicted plastidic localisation (Supplemental Figure 9-SII), making such transports in P. tricornutum conceivable.

In summary, this thesis demonstrates that the role of expressional regulation of the Calvin cycle in the model diatom P. tricornutum is apparently greater than in higher plants emphasising the differences in regulation between diatoms and the green lineage.

There are four different thioredoxins (TRX) located in the plastid of P. tricornutum, the transcriptional analysis of them in chapter 3 revealed very similar transcript patterns for TRX F and M, which both are diurnally expressed with strong light induced peak levels during midday and early afternoon. TRX of class F and M are known actors in plastid redox regulation of higher plants. Both types are very similar to each other but distinguish themselves by clearly distinct target specificities, while TRX F type features a strong specificity for Calvin cycle enzymes, TRX M type features generally reduced specificities for Calvin cycle enzymes [10, 34, 226, 227], but has also been to be able to activate PRK efficiently [10, 241] . The expressional data suggests an involvement of both TRX F and M in light dependent regulatory functions. One TRX exhibited a very interesting transcriptional pattern: TRX Y2 levels were apparently diurnally regulated but were noticeably increased in extended darkness enhancing the pattern observed in the light. An involvement in processes needed in extended darkness like electron donor reactions in the dark [89] seems likely.

From the expressional data and current knowledge of diatom Calvin cycle regulation a preliminary model describing the Calvin cycle enzyme specific importance of transcriptional and posttranscriptional regulation in P. tricornutum was conceived (Figure 3.4).