|Home | About | Journals | Submit | Contact Us | Français|
Conceived and designed the experiments: OL SS JLR IBF. Performed the experiments: OL SS JLR IBF. Analyzed the data: OL SS JLR IBF. Contributed reagents/materials/analysis tools: OL SS JLR IBF. Wrote the paper: OL SS JLR IBF. Funding: JLR IBF.
Growth, protein amount, and activity levels of metabolic pathways in Trichodesmium are influenced by environmental changes such as elevated pCO2 and temperature. This study examines changes in the expression of essential metabolic genes in Trichodesmium grown under a matrix of pCO2 (400 and 900 µatm) and temperature (25 and 31°C). Using RT-qPCR, we studied 21 genes related to four metabolic functional groups: CO2 concentrating mechanism (bicA1, bicA2, ccmM, ccmK2, ccmK3, ndhF4, ndhD4, ndhL, chpX), energy metabolism (atpB, sod, prx, glcD), nitrogen metabolism (glnA, hetR, nifH), and inorganic carbon fixation and photosynthesis (rbcL, rca, psaB, psaC, psbA). nifH and most photosynthetic genes exhibited relatively high abundance and their expression was influenced by both environmental parameters. A two to three orders of magnitude increase was observed for glnA and hetR only when both pCO2 and temperature were elevated. CO2 concentrating mechanism genes were not affected by pCO2 and temperature and their expression levels were markedly lower than that of the nitrogen metabolism and photosynthetic genes. Many of the CO2 concentrating mechanism genes were co-expressed throughout the day. Our results demonstrate that in Trichodesmium, CO2 concentrating mechanism genes are constitutively expressed. Co-expression of genes from different functional groups were frequently observed during the first half of the photoperiod when oxygenic photosynthesis and N2 fixation take place, pointing at the tight and complex regulation of gene expression in Trichodesmium. Here we provide new data linking environmental changes of pCO2 and temperature to gene expression in Trichodesmium. Although gene expression indicates an active metabolic pathway, there is often an uncoupling between transcription and enzyme activity, such that transcript level cannot usually be directly extrapolated to metabolic activity.
The marine filamentous N2 fixing (diazotroph) cyanobacteria Trichodesmium spp. form extensive blooms contributing 25 to 50% of the estimated rates of N2 fixation in the oligotrophic subtropical and tropical oceans . Trichodesmium's dominant role in carbon and nitrogen cycling has prompted investigations examining the effects of rising sea surface temperatures and elevated atmospheric pCO2 (leading to ocean acidification) on the growth and abundance of this organism.
Elevated pCO2 supports enhanced N2 fixation and growth rates in Trichodesmium –. These trends are further accentuated when elevated pCO2 and higher temperatures are combined , . The higher N2 fixation and growth rates are enabled via flexible phosphorus stoichiometry, changes in the activity of the CO2 concentrating mechanism (CCM), and modified protein activity –.
In Trichodesmium, as in other cyanobacteria, metabolic pathways (e.g. respiration, photosynthesis, Ci fixation, N2 fixation, and combined nitrogen assimilation) share cellular complexes such as plastoquinone (PQ) pool, succinate dehydrogenase and ferredoxin –. Trichodesmium's unique metabolism allows oxygenic photosynthesis and oxygen-sensitive N2 fixation to occur concurrently during the photoperiod via a complex spatial-temporal separation of these processes –. Photosynthetic activity in Trichodesmium is coupled with CCM activity. PSII driven electron transport is responsible for generating energy needed to pump HCO3− into the cell. This HCO3− is subsequently converted to CO2 to be used by the ribulose-1,5-bisphosphate carboxylase oxygenase (RubisCO) within the carboxysomes . Regulation of the photosynthetic and N2 fixation processes occurs at the transcription, translation, and post-translational (activity) levels , , , , , , . While elevated pCO2 and temperature resulted in higher growth rates, higher N2 fixation rates, and higher CP ratios, photosynthesis, protein pools, and total cellular allocation of carbon and nitrogen were not significantly affected . Importantly, the abundance of nitrogenase and glutamine synthetase (mediating combined nitrogen assimilation) did not increase in parallel to the increased N2 fixation rates, implying that environmental factors can allow higher reaction turnover rates through the same protein amounts , . In addition, our previous study showed that pCO2 changed the mRNA diel expression patterns, but not the abundance, of five genes (nifH, glnA, hetR, psbA, and psaB), resulting in a more synchronized expression pattern under elevated pCO2 . We therefore decided to check the combined effect of pCO2 and temperature on the expression levels of 21 genes of interest (GOI) representing key metabolic aspects in Trichodesmium, as part of Trichodesmium's acclimation response.
Genomic analyses demonstrate that Trichodesmium (IMS101) has a partial suite of CCM components (, ; http://genome.jgi-psf.org/finished_microbes/trier/trier.home.html). Accordingly, Trichodesmium possesses β-carboxysomes, a cellular compartment containing RubisCO, and a low-affinity, high-flux HCO3− uptake system called BicA –. Trichodesmium also has a specialized NADPH dehydrogenase, NDH-I4, which acts as a low-affinity CO2 uptake system, converting CO2 to HCO3− using the ChpX protein . The presence of a true internal carbonic anhydrase (CA) was not found in the genome and direct measurements by means of 18O2 exchange method  revealed only a low activity, close to the detection limit of the method . Yet, there is a distinct possibility that the N-terminal domain of the essential β-carboxysomal ccmM gene found in Trichodesmium can act as a γ–CA in an oxidized β-carboxysome interior as was observed in Thermosynechococcus elongatus , , .
Currently, there is no genetic system for Trichodesmium transformations, limiting the physiological study of CCM activity to the examinations of fluxes of inorganic carbon (Ci) and O2 . Here we present the expression and abundance of genes related to CCM (bicA1, bicA2, ccmM, ccmK2, ccmK3, ndhF4, ndhD4, ndhL and chpX), energy metabolism (atpB, sod, prx, glcD), nitrogen metabolism (glnA, hetR, nifH), and photosynthesis and Ci fixation (rbcL, rca, psaB, psaC, psbA) in Trichodesmium acclimated to a matrix of pCO2 (400 and 900 µatm) and temperature (25 and 31°C). Since diurnal regulation is essential for metabolic functions in Trichodesmium, we performed our measurements over the day and sampled 1, 5, 9 and 13 h after the onset of light. The sampling times were chosen for time periods that represent different metabolic preferences in Trichodesmium : time of maximal photosynthesis (1 h), maximal N2 fixation rates (5 h), late afternoon (9 h) and 1 h after dark induction (13 h). We compare the expression levels and patterns of these genes and look at the correlation of their coordinated expression.
Trichodesmium IMS101 stock cultures were grown in YBCII medium  at 25°C, 12 12 light/dark cycle at ~80 µmol photons m−2 s−1 white light and 400 µatm pCO2. Diluted batch cultures were grown in sterile square 1 L Nalgene bottles as single filaments with gentle bubbling, sufficient for preventing aggregates formation without harming the integrity of the filaments. Stock cultures were unialgal and under exponential growth the bacterial biomass was negligible and was not observed under light microscopy or by DAPI staining. Experimental cultures were enriched with CO2 and air mixes of 400 µatm (current) pCO2 and 900 µatm (expected 2100,) and were gradually acclimated to 31°C (1°C increase per week). Cultures were acclimated for at least 1.5–2 months before sampling. Biomass was kept under 0.2 µg chl a ml−1, thereby maintaining a low enough biomass that did not additionally influence the carbonate chemistry of the experimental setup. For more information about carbonate chemistry in similar experimental setups, see Kranz et al. .
Samples of Trichodesmium IMS101 were collected at 4 time points during the diurnal cycle, 1, 5, 9 and 13 h after the onset of light (the last point is 1 h after dark induction). Acclimated cultures were filtered on polycarbonate filters of 1 µm pore size; 25 mm diameter filters (Osmonics). Filters were placed in sterile DNase and RNase free centrifuge tubes and put directly into liquid nitrogen until transfer to -80°C for storage.
mRNA was extracted with the RNeasy Plant Mini Kit (Qiagen Cat.74904) according to the producers instructions. Additionally a DNase treatment was accomplished with RNase-Free DNase Set (Qiagen Cat.79254) on column during the extraction as well as with TURBO DNA free (Ambion Cat.AM1907) after the extraction, following the manufacturer's specifications for rigorous DNase treatment to remove any gDNA contamination. RNA concentration was measured with a NanoDrop ND-1000 Spectrophotometer (peqLab Biotechnologie) and quality was tested with 1% agarose gels. Reverse transcription was conducted with the QuantiTect Reverse Transcription Kit (Qiagen Cat.205311) according to the kit's manual. Each cDNA reaction contained 100 ng template RNA and was stored at −20°C until further utilization. RT-qPCR was carried out with Platinum SYBR Green qPCR SuperMix-UDG with ROX (Invitrogen Cat.11744-500) on an ABI PRISM 7000 Sequence Detection System. All three biological replicates of each sample (acclimation and time point) were measured in duplicate 25 µl reactions. The reaction mixture contained 5 µl diluted cDNA (equivalent to approx. 2 ng RNA), 12.5 µl SYBRgreen, 0.5 µl per primer (10 pmol µl−1) and 6.5 µl PCR water. Non-Template-Controls (NTC's) and samples with the cleaned RNA as template were run to exclude contaminations with gDNA. All NTC's and all RNA samples were below the detection limit. Cycling conditions were: 50°C for 2 min, 95°C for 2 min, 40 cycles of 95°C for 15 sec, and 60°C for 30 sec, followed by a dissociation stage of 95°C for 15 sec, 60°C for 20 sec, and 95°C for 15 sec. Primers for target genes were designed using Primer Express Software v2.0 (Applied Biosystems) and are presented, by name and function, in Table 1.
The RT-qPCR results were checked for inaccurate reactions. Single measurements with deficient primer characteristics and with bad primer efficiencies according to the LinRegPCR software were removed prior to calculations . Results were reported using the comparative CT method (2−ΔΔCt method) which calculates the relative changes in gene expression determined from RT-qPCR experiments, according to . We chose this relative quantification method as we compare not only different conditions but also results from a time course. To check if the efficiencies of the different primer pairs allow the usage of this method, we compared the mean efficiencies of all primer pairs. According to Schmittgen and Livak  a rough guide is that the efficiencies should be within 10% of each other. This provides us with a frame of values between 1.8 and 2.2. Our primers were within this range, with exceptions for atpB, hetR and glnA that yielded values of 1.6. We decided to include them into our calculations as the trends are still valid. The comparative CT method examines the threshold cycle (Ct) and indicates the fractional cycle number at which the amount of amplified target reaches a fixed threshold . The Ct values of the gene of interest (GOI) are normalized first to the 16S rRNA gene which is used here as the endogenous reference gene for each time point. This results in ΔCt values which are equal to the differences in thresholds for the GOI and the endogenous reference gene . In time course experiments, the gene expression is often compared internally by normalization to a calibrator, which can be the time zero point. Here we have chosen the average ΔCt values of the nifH from the 400 µatm/25°C treatments (our control treatment, already normalized to 16S rRNA) as a calibrator, since we wanted to compare the relative abundance of the different genes, as well as their time dependence. The expression of this gene was relatively constant over the day.
The rnpB gene, encoding RNase P, was examined as a potential endogenous reference gene and revealed unexpected large variations in its expression. Therefore we decided to use 16S rRNA for further calculations, as its expression was stable under all the different conditions. Following Bustin et al. , standard deviations were chosen to present statistical differences between independent replicates of mRNA transcript enrichments.
mRNA abundances of all 21 GOIs are presented in Figure 1 as the average mRNA abundance from all acclimations at each time point, n=12–13.
To examine the influence of sampling time and the applied environmental factors, pCO2 and temperature, we performed a 3-Way ANOVA (time, pCO2 and temperature, p<0.05) for the enrichment values of each GOI over the time course measured. Number of independent replicates was n=24–25 for each temperature, n=23–25 for each pCO2 concentrations, and n=12–13 for each measuring time point (Figure 2 and Table 2).
Pearson correlations for the enrichment values of our 21 GOIs were done for each sampling point, n=11–13. Correlations in which the Pearson correlation coefficient- r>0.75 are presented by color coding in Figure 3 (CCM-pink, energy metabolism-yellow, nitrogen metabolism-blue, Ci fixation and photosynthesis-green). All correlation coefficients and significances according to the Pearson correlation are supplied in the supplemental data (Table S1 i-iv).
We examined the expression levels of 21 GOIs over the day and under different pCO2 concentrations (400 and 900 µatm) and temperatures (25 and 31°C). The samples for the RT-qPCR analysis were taken from the exact experimental set-up described in Levitan et al. . The physiological characteristics of Trichodesmium IMS101 cultures used in these experiments are summarized in Table 3: growth rates (chl d−1), elemental stoichiometry [CN, CP, NP (molmol)], nitrogen fixation rates (nmol N2 chl−1 h−1), and NifH amounts (pmol µg protein−1). These data were presented and discussed in Levitan et al. .
The GOIs can be divided to 4 functional groups: CCM (bicA1, bicA2, ccmM, ccmk2, ccmK3, ndhF4, ndhD4, ndhL, chpX), energy metabolism (atpB, sod, prx, glcD), nitrogen metabolism (glnA, hetR, nifH), and photosynthesis and Ci fixation (rbcL, rca, psaB, psaC, psbA). The gene description and the primers sequences are presented in Table 1.
We examined the mean enrichment levels of the different GOIs over the day, arranged by their metabolic function, using the average enrichment values of all acclimations (Figure 1). The expression levels of each of our GOIs were on the same order of magnitude for all acclimations (excluding the high pCO2/high temperature acclimation for hetR and glnA; Figure 1).
The transcript abundance of the CCM and energy metabolism GOIs were low compared to the nitrogen metabolism and Ci fixation and photosynthetic genes. The CCM and energy metabolism genes spanned over 2 orders of magnitude, with the exception of ccmK2 and prx (Figure 1). The prx gene, encoding for the cyanobacterial peroxiredoxin 1-Cys revealed a unique diurnal trend and had higher transcript abundance relative to the other genes examined for energy metabolism (Figure 1). The ccmK2 gene is encoding for the main β-Carboxysome shell protein. The pores in the CcmK2 protein hexamers may enable diffusion of small essential metabolites into the carboxysome lumen .The ccmK2 gene revealed a trend similar to the other CCM genes, yet its expression was slightly higher (Figure 1). Most GOIs related to nitrogen metabolism (glnA, hetR and nifH) and photosynthesis and Ci fixation (rbcL, rca, psaB, psaC, psbA), exhibited 1–4 orders of magnitude higher average enrichment levels than the CCM and energy metabolism genes with varying expression patterns (Figure 1). Our findings correspond with results published from a community gene expression of a Trichodesmium spp. bloom in the Southwest Pacific Ocean showing that during the day, the highest abundance of Trichodesmium related genes was that of the photosynthetic and nitrogen metabolism pathways .
We previously showed the pCO2 influences gene expression patterns over the diurnal cycle for five of the above GOIs, nifH, hetR, glnA, psaB and psbA, in Trichodesmium IMS101 (data from another set of experiments, ). Three of these genes, nifH, psaB and psbA, had similar enrichment levels in both studies (Figure 1, ). The combination of elevated temperature and high pCO2, (not examined in ) significantly increased the transcript abundance of glnA and hetR. This acclimation resulted in ~2 orders of magnitude higher transcript levels than previously reported for glnA (Figure 2, ). The increase in glnA transcript abundance was not reflected in the GlnA protein pool size, and there was no significant difference for the GlnA amount between treatments and over the diurnal cycle (as measured by a quantitative western blot, one way ANOVA, p<0.05; Table 3; ). The combined influence of elevated pCO2 and high temperature increased the average hetR enrichment levels to levels similar to those previously reported . In Trichodesmium, the hetR gene was suggested to be constitutively expressed (under a 1212 Light/Dark cycle), yet it's diurnal abundance ranged 3–10 fold and was regulated by combined nitrogen concentration levels . Based on our results, we believe that the sensitivity of hetR to changes in pCO2, temperature and time, further points to an environmental sensitivity of this gene.
To determine whether the diurnal cycle interacted with the applied environmental factors to influence the transcript abundance of the GOIs, we applied a 3-Way ANOVA (pCO2, temperature and time of day) for all the GOIs tested. The results are summarized in Table 2 and the enrichment levels of the influenced genes for all acclimations are presented in Figure 2. Out of 21 GOIs, six genes (bicA1, ccmM, ccmK3, ndhL, atpB and psaC; Table 2) were influenced by the diel cycle alone and six other genes (bicA2, ndhF4, ndhD4, chpX, rca and sod) were influenced by neither time nor the applied environmental factors. Transcripts abundances of GOIs that were not affected by any of the three main factors, or were influenced by time only, are presented in the supplemental data (Figure S1). Only nine GOIs were significantly affected by pCO2 and/or temperature (nifH, glnA, hetR, rbcL, psbA, psbA prx, glcD, and ccmK2). Seven of the nine GOIs that appeared sensitive to pCO2 and/or temperature (nifH, glnA, hetR, rbcL, psbA, psbA, and prx; Figure 2) correspond with the nine genes expressed at the highest abundance (Figure 1). These genes are representative of the photosynthetic and Ci fixation, nitrogen metabolism and energy generation pathways.
While photosynthesis in Trichodesmium is relatively insensitive to changes in pCO2 , , N2 fixation rates vary significantly with changes in ambient pCO2 –, . N2 fixation, and possibly the sequential assimilation of ammonium, were affected by pCO2 at the mRNA and activity level, while protein pools remained relatively constant (Table 3; ). Abundance of the nitrogenase Fe-protein gene, nifH, was affected by pCO2, time of day, and the combined influence of pCO2 and temperature (Figure 2; ). This corroborates findings showing that nifH expression is pCO2 sensitive  and controlled by a circadian rhythm . Relative stability of nifH expression to temperature changes was also reported for Trichodesmium IMS101 grown at 24, 28.5 and 31°C . Similarly, temperature did not appreciably affect the abundance of the NifH protein and the nitrogenase N2 fixation rates in the temperature range applied here (Table 3; ). Trichodesmium cultures tested under a broader temperature range revealed changes in growth and N2 fixation rates . As natural populations of Trichodesmium spp. range from 20 to 34°C (reviewed in ), it would be advisable to further examine the acclimation responses and levels of regulation under a wider temperature range.
glnA and hetR transcripts were statistically influenced by all three variables: pCO2, temperature and time of day (Table 2), in line with the reported influence of pCO2 and time on both genes , , . To our knowledge, scant data on the effect of environmental conditions on glnA and hetR expression in Trichodesmium shows that fixed-nitrogen sources and diurnal rhythmicity regulated glnA expression in natural populations of marine Synechococcus spp  and hetR expression in Trichodesmium . The hetR gene, previously suggested to be involved only in heterocysts differentiation, was also found in the non-heterocystous filamentous diazotroph Symploca PCC8002  and Lyngbya PCC8106 . The existence of hetR in Trichodesmium, Symploxa PCC8002a and Lyngbya PCC8106, its regulation by combined nitrogen status and time , , and the apparent sensitivity of Trichodesmium's hetR to pCO2 (also affecting N2 fixation; Figure 2, Table 2), further suggest that hetR must play a critical role in diazotroph nitrogen metabolism and is not limited to heterocyst differentiation .
The regulation of photosynthetic genes is essential in Trichodesmium where photosynthetic O2 evolution is separated from N2 fixation by a complex spatial-temporal strategy. To enable N2 fixation, down regulation of photosystem II (PSII), possibly controlled by redox state of the plastoquinone (PQ) pool, occurs at midday . Changing expression levels and patterns of photosynthetic genes can influence Trichodesmium's metabolism. In Trichodesmium, the expression pattern of psaB and psbA were affected by pCO2 . Our data reveal that while photosystem I (PSI) core protein gene psaC was influenced only by time, psaB (PSI) was influenced by both pCO2 and time and psbA (PSII) was influenced by time and the interaction between pCO2 and temperature, with no apparent sensitivity to temperature as the predominant factor (Table 2, Figure 2). psaA and psbA are controlled by a circadian rhythm and their expression pattern did not change between 24 and 28.5°C . psaA and psaB are closely located in Trichodesmium's genome and their interaction is highly conserved for many cyanobacteria . Therefore, we deduce that the temperature insensitivity of psaA  corroborates our finding (Table 2, Figure 2).
Three additional genes, ccmK2, rbcL and glcD, were influenced by the 2-Way interaction of pCO2 and temperature (Table 2). Presently, there is no report on the expression and regulation of these genes in Trichodesmium. In Synechococcus PCC7942, but not in all cyanobacteria, the ccm operon is located in the 5′-flanking region on the rbcL-rbcS operon , . However, our genomic analysis reveals that rbcL and ccmk2 are not closely oriented in the Trichodesmium genome. In Synechocystis PCC6803, rbcL expression was insensitive to changes in pCO2 (0 to 3% CO2 in air; ). Our findings show that rbcL expression was modified only when combining high pCO2 with high temperature, yet its abundance was still at the same order of magnitude for all acclimations. In Synechocystis PCC6803 the Ci derived transcriptional changes in rbcL transcript amount were uncoupled from changes in RbcL protein level, possibly resulting from low protein turnover rate due to the protective effect of the carboxysome, slowing down the protein degradation . This was also indicated for Trichodesmium in our study. While pCO2 alone and the combination of pCO2 and temperature influenced the RbcL protein amount, the highest protein level was at 900 µatm/25°C while the highest transcript level appeared at 900 µatm/31°C (unpublished data; Figure 2).
The oxygenase activity of RubisCO forms 2-phosphoglycolate (2PG), considered toxic for Ci fixation in the Calvin cycle. The GlcD protein helps protect the Calvin cycle by converting two molecules of 2PG into one 3-phosphoglycarate (3PGA) molecule (the product of RubisCO's carboxylase activity), and thus enables Ci fixation to proceed. glcD metabolism was found essential for the viability of the cells and oxygenic photosynthesis in the cyanobacterium Synechocystis PCC6803 at ambient CO2 conditions . Statistical analysis (3-Way ANOVA, p<0.05) revealed that the glcD mRNA abundance was sensitive to the combined influence of pCO2 and temperature, yet its abundance was the same for all our acclimations (Figure 2, Table 2). Although glcD is also found in Arabidopsis and Anabaena , there is generally scarce information regarding the expression and regulation of this gene.
Temperature, time of day and their interaction affected the 1-cys peroxiredoxin gene, prx (Table 2), increasing its abundance by 3 orders of magnitude from 1 to 9 h after the onset of light (Figure 2). 1-cys prx mRNA increased in response to different metabolic imbalances in Synechocystis PCC6803, including irradiation, salinity, and iron deficiency . No data are currently available on changes of prx at different pCO2 and/or temperatures in Trichodesmium and other cyanobacteria. O2 generated in PSII is reduced to H2O2 by PSI related components . In cyanobacteria peroxiredoxin reduces H2O2 to H2O using electrons donated from a variety of substrates . Increased expression of iron and oxidative stress genes at the end of the high N2 fixation period was detected for cultures of the unicellular diazotroph Crocosphaera watsonii . Biological fixation of one N2 molecule requires at least 16 ATP molecules that can be generated via cyclic electron flow around PSI . Thus, in Trichodesmium, the higher expression of prx in the second half of the photoperiod may be required to recover from the high energetic demand for N2 fixation, leaving the cell susceptible to oxidative stress. In addition, peroxidases function as regulators of redox-mediated signal transduction in some eukaryotes , , and are therefore important components for the cellular antioxidant defense system and redox homeostasis . Redox state of shared components between photosynthesis and respiration regulates gene expression in Trichodesmium , . Hence, changes in prx expression reported here (Figure 2, Table 2) indicate that oxidative defense, photosynthesis and/or respiratory redox state in Trichodesmium are temperature and time dependent.
Our results indicate that these nine genes (nifH, glnA, hetR, rbcL, psbA, psaB prx, glcD and ccmK2) are non-constitutively expressed and are regulated both by a diurnal cycle and by environmental factors such as pCO2 and temperatures.
The nine CCM-related GOIs that were tested are representative of all known CCM complexes in Trichodesmium: the carboxysome (ccmM, ccmK2, ccmK3) that contains the cellular Ci fixation enzyme RubisCO (rbcL; ), HCO3− transporter named BicA (bicA1, bicA2), and the specialized NADPH dehydrogenase NDH-I4 (ndhF4, ndhD4, ndhL, and chpX). Trichodesmium lacks any genes of inducible-high affinity uptake system for both CO2 and HCO3−, such as NDH-I3 (CO2), BCT1 or SbtA (HCO3−; , ), and has no recognizable carbonic anhydrase (CA) genes .
CCM operation in algae is regulated by environmental factors with elevated CO2 levels expected to reduce the cellular requirements for concentrating Ci, and enabling enhanced growth . All of our nine examined CCM-related GOIs (Table 1) exhibited similar expression patterns and low expression levels when compared to Ci fixation, photosynthesis and nitrogen metabolism genes (Figure 1). Only one gene, the ccmK2, was affected by changes in environmental conditions (Figure 2, Table 2). For all time points measured, the expression of the CCM related genes had the highest correlations of all the GOIs metabolic groups. This applies within the CCM group and also with the other functional groups (Figure 3).
CCM genes of high Ci affinity are known to be regulated at the transcript level . Our experimental setup is different in two aspects from “classical” cyanobacterial CCM induction experiments: 1. we report on a steady state expression of CCM genes under long term constant CO2 conditions, whereas in most publications cells are rapidly transferred (usually less then 1 day) from one CO2 concentration to another; 2. Trying to work on ecologically relevant pCO2 concentrations, we applied 400 and 900 µatm pCO2. This change is very small in comparison to concentrations >1% CO2 that are usually referred to as high CO2 in CCM-literature. As publications reporting steady state CCM gene expression and acclimations to ecologically relevant pCO2 levels are scarce we will use the available literature to discuss our data.
In Synechocystis PCC6803, genes of low affinity CCM components such as ndhD4, ndhF4. chpX,and ccmK-N ,  and bicA ,  were Ci insensitive, constitutively expressed, and revealed relatively low transcript abundance as we found for Trichodesmium (Figures 1 and and3,3, Table 2). In Synechococcus PCC7002 bicA is regulated by a ccmR gene , which is absent in the Trichodesmium genome .
In Trichodesmium's genome, most of the CCM-related genes are not arranged in operons or clusters (http://genome.jgi-psf.org/finished_microbes/trier/trier.home.html), as was previously shown for Synechocystis PCC6803 . This also applies for the three CCM-related gene-pairs that were co-expressed over the day, ccmM-ccmK3, ccmK3-ccmK2 (caboxysome shell) and ndhF4-ndhD4 (NDH-I4; Table 4). We conclude that in Trichodesmium, CCM genes are constitutively expressed and are mostly unaffected by the applied changes in pCO2 and temperature.
Genomic analyses indicate that Trichodesmium lacks inducible inorganic carbon (Ci) uptake systems , . Yet, physiological measurements of Ci uptake showed that Trichodesmium changes its Ci uptake characteristics when acclimated to high CO2 (900 µatm; , ). While the cell's affinity to total DIC decreased with elevated pCO2 , the cell's CO2 uptake increased . Under a range of pCO2 (150-900 µatm pCO2), Trichodesmium uses HCO3− for over 90% of its Ci source (; Kranz and Levitan, unpublished data). Based on the genetic analysis (Figure 1; ), the Km of Ci uptake  and BicA being a low-affinity but high flux HCO3− uptake system , it is likely that a major part of Trichodesmium's Ci uptake is via the Na+ dependent HCO3− transporter BicA.
The operation of a Ci uptake system that maintains constant transcription levels while its affinity is modified indicates that CCM operation in Trichodesmium is controlled at the translational or post-translational levels. Changes in CCM operation without altering gene expression or the cell's capacity to transport Ci was proposed by Beardall and Giordano , i.e. via fluctuations in the redox state of the PQ pool. A rapid increase in HCO3− transport activity appears to involve phosphorylation events, possibly by activating two or three component regulatory systems, is of considerable importance when looking at CCM regulation , . The thioredoxin regulatory system and internal Ci pools can also act in controlling CCM operation, away from the transcript level , . Moreover, it is possible that large transcript changes were not detected in Trichodesmium's CCM genes due to the long acclimations (>2 months) of the cultures, whereas a rapid transfer of Trichodesmium from low to high CO2 may result in changes in transcript abundance. Finally, although there are physiological changes in cells grown at different pCO2s –, it could be that the acclimation to 900 µatm pCO2 does not simulate a large enough increase to detect significant differences in CCM gene abundance.
Our analyses show low transcript abundance of co-expressed CCM genes in Trichodesmium (Figures 1 and and3)3) that are insensitive to changes in pCO2 and temperature (Figure 2; Table 2). This, together with genomic analysis  and physiological data, suggest that CCM genes in Trichodesmium are constitutively expressed under our applied conditions.
We explored co-expression of GOIs by examining the correlations between their enrichment at all measured time points. Figure 3 presents GOIs with Pearson correlation coefficients higher than 0.75 (r>0.75, p<0.01, all correlations are given in Table S1 i-iv in the supplemental data). The highest number of significant correlations between GOIs of different metabolic functions appeared 5 h after the onset of light, when high N2 fixation rates and assimilation are detected , , . A large number of correlations were also observed between genes 1 h after the onset of light. Later in the day, at 9 and 13 h, a significantly lower number of correlations were detected, especially between GOIs related to different metabolic groups, indicating only limited co-expression.
Expression levels of nitrogen metabolism GOIs (glnA, hetR, nifH) were correlated with GOIs of other metabolic functions predominantly at 5 and 9 hours after the onset of light, yet the highest number of correlations was found at 5 h (Figure 3). The limited literature on gene expression patterns in Trichodesmium demonstrated diurnal regulation of genes correlated with photosynthesis and nitrogen metabolism , , , , . At 9 h, only nifH was co-expressed with GOIs of other metabolic functional groups, especially with the carboxysomal, NDH-I4 and photosynthesis related genes. Chen et al.  demonstrated a time-dependent cycling and coupling between nifH and photosynthetic transcripts in Trichodesmium. Our results showed that nifH was co-expressed with the photosynthesis-related genes (psaC, psbA and rbcL), with all CCM components, and with two of the energy metabolism genes (atpB and sod).
Out of the three nitrogen metabolism genes tested, only the glnA-hetR gene pair was co-expressed for all time points measured (Figure 3, Table 4), although they are not closely localized in the Trichodesmium genome (http://genome.jgi-psf.org/finished_microbes/trier/trier.home.html). hetR was also highly correlated to the NDH-I4 genes, to atpB, and to sod. The co-expression of glnA and hetR did not correspond with nifH expression (Figure 3), in agreement with other studies showing that hetR expression was inversely correlated with nifH expression in Trichodesmium , . In Trichodesmium, both glnA and hetR are likely under ntcA regulation , yet we couldn't verify this in our experiment.
Energy metabolism related GOIs (atpB, sod, prx, and glcD) were co-expressed and positively correlated with all CCM related GOIs and photosynthetic genes at 1 and 5 h after the onset of light. This could be related to the energetically demands of the CCM, and to the connection between photosynthetic electron transfer and the use of these electrons in sequential processes (Figure 3; , ). prx correlated with many other GOIs only 1 h after the onset of light. This correlation disappeared from 5 h onwards, when prx mRNA abundance rapidly increased (Figures 1 and and22).
Positive correlations between Ci fixation and photosynthesis to CCM GOIs were observed at all measured time points. A high number of correlations were especially noted 1 and 5 h after the onset of light for rbcL and psbA together with the CCM GOIs (i.e. rbcL-ccmK2 at 1, 5 and 9 h). Fewer correlations, predominantly occurring at 5 h, were detected between photosynthetic and nitrogen metabolism GOIs (Figure 3). Co-expression of CCM and photosynthetic genes at the first half of the photoperiod (when photosynthesis and carbon fixation take place) could account for the tight interaction observed between the two mechanisms , .
The co-expression of GOIs we observed fundamentally reflects the diurnal patterns of the predominant metabolic pathways in Trichodesmium (CCM, photosynthtesis and carbon fixation, nitrogen metabolism, and energy generation). Transcriptional regulation is the first level of regulation, followed by translational and post-translational regulation. Different levels of metabolic regulations were found in Trichodesmium, for example for nitrogenase , , ,  and PSII , . The differing patterns of co-expression between the metabolic gene families during the day (Figure 3) indicate a strategy of a complex and tightly regulated gene expression. In Trichodesmium, such a strategy is required due to the unique spatial-temporal segregation of oxygenic photosynthesis and N2 fixation –.
Our motivation in this study was to examine changes in expression of essential metabolic genes in Trichodesmium grown under a matrix of pCO2 and temperature. In Trichodesmium IMS101, nitrogen metabolism, Ci fixation, and photosynthesis related GOIs exhibited the highest abundance of all measured genes (Figure 1). These genes were also mostly affected by changes in pCO2, temperature and the time within the diurnal period (Figure 2, Table 2), suggesting that these metabolic functions are also controlled at the mRNA transcript level. To our knowledge this is the first report of CCM gene expression in Trichodesmium. We suggest that CCM genes in Trichodesmium are constitutively expressed under our applied conditions, yet, their corresponding protein activity may be altered by changes in pCO2 , , probably due to translational and/or post-translational regulations .
Protein and activity levels of the CCM and fixation pathways in Trichodesmium are influenced by environmental changes –. Thus, we hypothesized that modifications in the CCM genes expression due to elevated pCO2 may facilitate the reported physiological changes. Our results negate this hypothesis as the expression of CCM-genes under long term acclimation (steady state conditions) was insensitive to changes in experimental conditions. The comprehensive analysis of abundance and expression patterns of the GOIs presented here, demonstrates that gene expression may be uncoupled from translational and protein activity levels. Thus, although gene expression reflects active metabolic pathways, there is often an uncoupling between transcription and enzyme activity. Therefore we conclude that to examine the effects of environmental parameters on Trichodesmium and its biogeochemical impact, studies of gene transcript levels should by be done in parallel with physiological and activity measurements.
Daily mRNA transcript enrichment of 12 GOIs, not significantly influenced by changing environmental factors. Significant influence of pCO2 (400 and 900 µatm) temperature (25 and 31 °C) and their interaction, on the GOI expression was done according to a 3‐Way ANOVA (p<0.05, Table 3). The left panel represent GOIs that no influencing factor (bicA2, ndhD4, ndhF4, chpX, rca, sod) and the right panel represent genes for which time was the only influencing factor. Circles and triangles represent Trichodesmium acclimated to 25 °C and 31 °C, respectively. Black and open symbols represent Trichodesmium acclimated to 400 and 900 µatm pCO2, respectively. Relative abundance estimated according to the 2‐ΔΔCt method, with 16S rRNA as the endogenous reference gene, and average ΔCt values of the nifH from the 400 µatm / 25 °C acclimation (control) as a calibrator. White and black bars on top of the graphs represent light and dark hours, respectively. n = 3 for all. Errors are ±1 standard deviation, following Bustin et al. (2009). Note: 1. the different y‐axes scales; 2. the results and standard deviations are presented using logarithmic scale y axes.
Pearson correlations of the enrichment of the 21 genes of interest over the day. i‐ 1 h after the onset of light; ii‐ 5h after the onset of light; iii‐ 9 h after the onset of light; iv‐ 13 h after the onset of light. Relative abundance estimated according to the 2‐ΔΔCt method, with 16S rRNA as the endogenous reference gene, and average ΔCt values of the nifH from the 400 µatm / 25°C treatments as a calibrator. n = 11‐13 for each gene at a given sampling time. Upper value represents the correlation coefficient (r) and lower values represent the significance (p).
This work was conducted as a partial fulfillment of the requirements for a PhD thesis for O. Levitan at Bar Ilan University. We wish to thank Ms. Diana Hümmer (IFM-GEOMAR). Bioinformatic work was done using the Trichodesmium genome published on the Joint Genome Institute (JGI) website. The work conducted by the U.S. Department of Energy Joint Genome.
Competing Interests: The authors have declared that no competing interests exist.
Funding: This work was conducted as part of the BMBF-MOST grant No. 1950 to IBF and JLR. O. Levitan was supported by a Reiger Fellowship for Environmental Studies and by the Eshkol scholarship from the Israeli Ministry of Science. Trichodesmium genome published on the Joint Genome Institute (JGI) website (http://genome.jgi-psf.org/finished_microbes/trier/trier.home.html), conducted by the U.S. Department of Energy Joint Genome Institute is supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.