Abstract
Diatoms are one of the most successful phytoplankton groups in our oceans, being responsible for over 20% of the Earth’s photosynthetic productivity. Their chimeric genomes have genes derived from red algae, green algae, bacteria and heterotrophs, resulting in multiple isoenzymes targeted to different cellular compartments with the potential for differential regulation under nutrient limitation. The resulting interactions between metabolic pathways are not yet fully understood.
We previously showed how acclimation to Cu limitation enhanced susceptibility to overreduction of the photosynthetic electron transport chain and its reorganization to favor photoprotection over light-harvesting in the oceanic diatom Thalassiosira oceanica (Hippmann et al., 2017). In order to understand the overall metabolic changes that help alleviate the stress of Cu limitation, we generated comprehensive proteomic datasets from the diatom Thalassiosira oceanica grown under Cu-limiting and -replete conditions. The datasets were used to identify differentially expressed proteins involved in carbon, nitrogen and oxidative stress-related metabolic pathways and to predict the proteins cellular location.
Metabolic pathway analysis showed integrated responses to Cu limitation in diatoms. The up-regulation of ferredoxin (Fdx) was correlated with up-regulation of plastidial Fdx-dependent isoenzymes involved in nitrogen assimilation as well as enzymes involved in glutathione synthesis thus integrating nitrogen uptake and metabolism with photosynthesis and oxidative stress resistance. The differential regulation of glycolytic isoenzymes located in the chloroplast and mitochondria enables them to channel both excess electrons and/or ATP between these compartments. Additional evidence for chloroplast-mitochondrial cross-talk is shown by up-regulation of chloroplast and mitochondrial proteins involved in the proposed malate shunt.
Introduction
Diatoms form an integral part of our oceans, influencing nutrient cycling and productivity of many marine foodwebs (Armbrust, 2009). Annually, marine diatoms fix as much carbon dioxide through photosynthesis as all terrestrial rainforests combined (Field et al., 1998; Nelson et al., 1995), thus having a significant impact on atmospheric CO2 levels and global climate. One key to their success may lie in their complex evolutionary history (Moustafa et al., 2009; Oborník and Green, 2005) which resulted in a mosaic genome with genes derived from the original heterotrophic eukaryotic host cell, the engulfed green and red algal endosymbionts, and a variety of associated bacteria (Armbrust et al., 2004; Bowler et al., 2008; Finazzi et al., 2010). As a result, diatoms possess multiple isoenzymes in many metabolic pathways, especially in carbon metabolism (Ewe et al., 2018; Gruber et al., 2009; Gruber and Kroth, 2014; Kroth et al., 2008; Smith et al., 2012).
The presence of multiple isoenzymes with different evolutionary histories also led to novel locations and interactions among metabolic pathways compared to green algal and animal ancestors (Allen et al., 2011; Gruber and Kroth, 2017). For example, in animals the complete set of proteins involved in glycolysis is located in the cytosol, whereas in green algae the first half of glycolysis (glucose to glyceraldehyde-3-phosphate, GAP) is located in the chloroplast and the second half (GAP to pyruvate) in the cytosol. In diatoms, an almost complete set of glycolytic proteins is found in both the cytosol and the chloroplast, with an additional set of proteins from the second half of glycolysis located in the mitochondria (Kroth et al., 2008; Río Bártulos et al., 2018; Smith et al., 2012). Furthermore, proteins involved in the ancient Entner-Dourodoff pathway, which is predominantly restricted to prokaryotes and catabolizes glucose to pyruvate, have also been identified in diatom genomes and are targeted to the mitochondria (Fabris et al., 2012; Río Bártulos et al., 2018).
The genome of Phaeodactilum tricornutum (P. tricornutum) encodes five different fructose-bisphosphate aldolase (FBA) isoenzymes, three targeted to the chloroplast and two to the cytosol (Allen et al., 2012). Each FBA has its own phylogenetic history. The expression pattern of these five isoenzymes changes depending on the nutritional status of the cell (Allen et al., 2012).
One of the most surprising discoveries from diatom genome sequencing was a complete urea cycle (Allen et al., 2011; Armbrust et al., 2004). In contrast to the catabolic nature of the urea cycle in animals, in diatoms it is an integral part of cellular metabolism and a hub of nitrogen and carbon redistribution within the cell. It is involved in amino acid synthesis, cell wall formation, carbon and nitrogen recycling, and it interacts with the citric acid cycle (Allen et al., 2011; Armbrust et al., 2004).
Most molecular studies on acclimation to nutrient limitation have focused on macronutrients, or on the essential micronutrient Fe, which limits phytoplankton in over 30% of the ocean (Moore et al., 2004). Some studies have shown an intricate interaction between Fe and Cu nutrition in phytoplankton (Annett et al., 2008; Guo et al., 2012; Maldonado et al., 2006, 2002; Peers and Price, 2006), but there are only a handful of studies on physiological adaptations to Cu limitation alone (Guo et al., 2015, 2012; Kong and M. Price, 2020; Lelong et al., 2013; Lombardi and Maldonado, 2011; Maldonado et al., 2006; Peers et al., 2005; Peers and Price, 2006).
Our recent comprehensive investigation on the physiological and proteomic changes to the photosynthetic apparatus of two strains of the open ocean diatom Thalassiosira oceanica (T. oceanica) in response to chronic Cu limitation revealed both similar and different strategies compared to those observed in response to low Fe (Hippmann et al., 2017). Acclimation to low Cu caused a bottleneck in the photosynthetic electron transport chain that was accompanied by major increases in the electron acceptors ferredoxin and ferredoxin:NADP+ reductase, which have major roles in counteracting reactive oxygen species. Along with changes in the composition of the light-harvesting apparatus, this resulted in a shift from photochemistry to photoprotection.
To better understand how carbon and nitrogen metabolism are affected and may interact when Cu is limiting, we now expand our proteomics analysis to include proteins involved in various carbon and nitrogen metabolic pathways (e.g. Calvin-Benson-Bassham cycle, glycolysis, TCA cycle, nitrogen acquisition and assimilation, urea cycle, malate shunt, glutathione metabolism), taking into account their predicted cellular compartments.
Results
Overview of proteomic datasets
We investigated two strains, CCMP 1003 and CCMP 1005, of the centric diatom T. oceanica (here referred to as TO03 and TO05, respectively). Cu limitation had a stronger and more comprehensive effect on proteins of the carbon and nitrogen metabolism in TO03 than T005, in line with observations for photosynthetic electron transport proteins (Hippmann et al., 2017). Although the proteomic dataset of TO05 contains twice as many distinct proteins as that of T003 (1,431 versus 724), TO03 has three times more significantly up-regulated and ten times more significantly down-regulated proteins (Fig. 1, overview Fig. S 5 and S 6). For this reason, if not noted otherwise, we will focus on the TO03 results only (Table 2, Table 3). A short discussion on the different adaptational strategies of the two strains can be found in Notes S 1. The data for all relevant proteins in both strains, and both proteomic datasets (main and extended) are given in the Supplementary Table S 2 – S 9. Expression differences are classed as “highly regulated” (greater than or equal to 2-fold difference) or “regulated” (1.3 to 2-fold difference, see Methods). All differential expression data discussed in the text are significantly up- or down-regulated (p<0.05), unless otherwise noted.
Of the 724 distinctive proteins in TO03, 525 have associated Kegg Orthology (KO) identifiers, and 52% of these were related to metabolism (Fig. 2). Furthermore, 77-78% of these metabolic proteins were particularly affected by Cu limitation, with general trends of down-regulation of proteins involved in energy metabolism, up-regulation of those in carbohydrate metabolism, and a modification of those in amino acid metabolism.
Carbon fixation, Glycolysis and the Citrate (TCA) cycle
Diatom genomics have shown that enzymes of glycolysis are found in all three major compartments: chloroplast stroma, cytosol and mitochondria (Gruber and Kroth, 2017; Kroth et al., 2008; Río Bártulos et al., 2018; Smith et al., 2012). Four (or seven, if the 3 triose isomerase isoenzymes are counted) of the 15 proteins involved in the carbon fixing Calvin-Benson-Bassham (CBB) cycle are part of the chloroplast glycolytic pathway (Table 2, Fig. 3 (CBB and TCA cycle), Fig. 4 (Glycolysis), Table S 4). In the initial step of CO2 fixation, the large and small subunits of Rubisco were not affected by Cu limitation but the essential Rubisco activator protein cbbX (To24360) was down-regulated by 2.3-fold. Six proteins were up-regulated: phosphoglycerate kinase (PGK, To07617) by 6.8-fold, the two fructose-bisphosphate aldolase, class II proteins by 1.4 and 2-fold (FBA II, To00388 and To12069), and the three triose phosphate isomerase isoenzymes (TPI, To02438, To35826, To32006) by 3.3-, 1.9- and 1.5-fold, respectively. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH, To13085) and phosphoglycerate mutase (PGAM, To21902, 2.3-fold) were the only proteins down-regulated, by 4.2- and 2.3-fold, respectively. Of the nine expressed proteins targeted to the chloroplast, only fructose-bisphosphate aldolase, class I (FBA I, To02112) was not affected by Cu limitation.
The up-regulation of TPI (Fig. 3, Fig. 4) combined with the down-regulation of GAPDH and PGAM could lead to an increase in triose-phosphates and their subsequent export from the chloroplast. Probing the genome for gene models containing the triose-phosphate transporter Pfam domain identified seven candidate genes (Table S 3) of which only two were expressed. Neither of them was differentially expressed.
Nine expressed proteins involved in the citrate cycle in the mitochondria were identified (Fig. 3, Table 2, Table S 5). Malate dehydrogenase (MDH1, To03405) was the only one up-regulated (1.6-fold). Aconitase hydratase (ACO, To20545) and two isocitrate dehydrogenases (IDH, To37807, To34595) were all down-regulated by 4.7-, 3.0-, and 1.6-fold, respectively. Of the proteins considered to be part of mitochondrial glycolysis (Fig. 4), glyceraldehyde-3-phosphate dehydrogenase (GAPDH, To33331) and enolase (ENO, To34936) were both up-regulated by 3.8 and 1.5-fold respectively, while pyruvate kinase (PK, To07097) was down-regulated by 1.3-fold.
Of the eight expressed cytosolic proteins detected, three were down-regulated: phosphoglucomutase (PGM, To06412) by 3.2-fold, phosphofructokinase (PFK, To16559) by 1.8-fold, and fructose-bisphosphate aldolase, class I (FBA I, To24978) by 1.6-fold (or 2.7-fold considering expression of a contig associated with the same gene). The only cytosolic protein that was up-regulated was pyruvate kinase (PK, To34937, by 1.6-fold).
Nitrogen metabolism
Twenty-two proteins involved in the urea cycle, nitrogen acquisition and assimilation, as well as four membrane transporters were identified (Table 3, Fig. 5, Table S 6). At the plasma membrane, the urea (URT, To31656) and nitrate/nitrite (NRT, To04919) transporters were both significantly up-regulated (6.9 and 11-fold, respectively). However, the expression of the two transporters putatively located in the chloroplast envelope, the formate/nitrate (NiRT, To00240) and ammonium (AMT, To07247) transporters were not affected by Cu limitation.
Within the chloroplast, three nitrite reductases were identified. Of these, the NAD(P)H-dependent isoenzyme was not differentially expressed (NAD(P)H-NiR, To35252), whereas two ferredoxin-containing nitrite reductases (Fe-NiR, To00016 and To02363) were up-regulated by 1.3- and 2.3-fold, respectively, in concert with the increase of reduced ferredoxin in the chloroplast (Hippmann et al., 2017). Glutamine (GSII, To31900) and glutamate synthases (GOGAT, To13288) were both up-regulated (1.7- and 1.6-fold, respectively). In contrast, aspartate aminotransferase (AAT, To16827) was the only chloroplast protein involved in core nitrogen metabolism that was down-regulated (2.3-fold).
In the mitochondria, glutamine synthase (GSIII, To06032) was 5.3-fold down-regulated while glutamate synthase (GOGAT, TO04828) expression did not change. The mitochondrial aspartate aminotransferase (AAT, To15049) was up-regulated by 3.6-fold. Glycine decarboxylase t- and p-proteins (GDCT/P, To17688 and To36273), involved in photorespiration, were not affected. In the urea cycle, six proteins were identified but only ornithine carbamoyltransferase (OTC, To05385) was up-regulated by 1.7-fold.
In the cytosol, glutamate dehydrogenase (GDH, To06254) was up-regulated by 2.09-fold (p = 0.05) and nitrate reductase (NR, To34460) was up-regulated by 1.5-fold. The only other cytosolic protein upregulated was spermidine synthase (SRM, To22108; by 2.3-fold), which is essential for silica deposition.
Malate shunt
In plants, malate transfers excess NAD(P)H reducing equivalents from one compartment (i.e. the chloroplast) to another (i.e. the mitochondria) (reviewed by (Scheibe, 2004)). Metabolite antiporters and two isoenzymes for malate dehydrogenase (MDH) and amino aspartate transaminase (AAT) are involved (Table 2, Fig. 6, Table S 8). In diatoms, the malate shunt has been proposed to connect the chloroplast with the mitochondria (Bailleul et al., 2015; Prihoda et al., 2012). In P. tricornutum, both MDH1 and MDH2 are targeted to the mitochondrion (Ewe et al., 2018) but in T. pseudonana MDH2 is predicted to be targeted to the chloroplast (Smith et al., 2012). Aligning the T. oceanica model with Smith’s extended TpMDH2 model (Fig. S 2), supports the conclusion that ToMDH2 is also targeted to the chloroplast.
In our proteomic datasets, we found evidence for regulation of both isoenzyme sets in the putative malate shunt (MDH1, MDH2, AAT, AAT2), as well as two isoenzymes of pyruvate carboxylase (PC) that could be feeding into this metabolic pathway (Fig. 6). Of these 6 enzymes, 4 were up-regulated: both, plastidial pyruvate carboxylase (PC, To31413) and malate dehydrogenase (MDH2, To30817) by 2.6- fold and mitochondrial MDH1 (To03405) and AAT2 (To15049) by 1.6- and 3.6-fold, respectively. Only the plastidial AAT (To16827) was down-regulated (by 2.3-fold).
Glutathione and antioxidant metabolism strongly upregulated
Glutathione is a small tripeptide (Glu-Cys-Gly) that is involved in redox sensing and counteracting ROS. Twenty-one expressed proteins involved in glutathione metabolism and other antioxidant agents (eg. three thioredoxins, three glutaredoxins, and three superoxide dismutases) were identified (Table 3, Fig. 7, Table S 7). Nine proteins are predicted to be targeted to the chloroplast. Six of these were up-regulated: two isoenzymes for cysteine synthase (CYS, To27524 by 2.5-fold, To10442 by 1.5-fold), glutamate synthase (GOGAT, To13288 by 1.6-fold), glutathione reductase (GR, To07268, by 2.5-fold), thioredoxin (TXN, To31425, by 1.5-fold), and the Mn-containing SOD (MnSOD, To02860, by 1.8-fold). Two glutaredoxin isoenzymes (GRX; To07269, To18234) were only mildly down-regulated proteins (both by 1.3-fold).
Of the nine cytosolic proteins, glutathione-S-transferase (GST, To09062) and TXN (To05213) were highly up-regulated (by 7.3- and 4.2-fold, respectively), while one of the two Ni-dependent SOD isoenzymes was moderately up-regulated (NiSOD, To10112, by 1.4-fold). Glutamate-cysteine ligase (GCL/GCS, To23355) was only identified in one of the biological triplicates, but with a 4.5-fold increase in expression. In contrast to the chloroplast CYS isoenzymes, the expression of cytosolic CYS (To05931) did not change. The TO03 GST (To09062) has its closest homologs in the polyp Hydravulgaris, the anemone Nematostella, and the brachiopod Lingula, and not in other diatoms (Table S 7).
Only 2 of the expressed proteins involved in glutathione metabolism are predicted to be targeted to the mitochondria; TXN (To13864) was up-regulated by 1.8-fold, whereas glutaredoxin (GRX, To02323) was not differentially expressed in Cu-limited TO03.
Discussion
In response to low Cu, T. oceanica (CCMP1003) restructures the photosynthetic electron transport proteins, resulting in a decrease in carbon assimilation, and increased susceptibility to overreduction of the photosynthetic electron transport chain (Hippmann et al., 2017). Over-reduction of the photosynthetic electron transport chain at super-saturating light intensities can lead to an increase in reactive oxygen species (ROS). Consequently, there is an increased need to safely dissipate excess energy, for example through additional electron sinks (Niyogi, 2000). Our findings of a ~40-fold increase in ferredoxin (Fd, petF) and a 2.5-fold increase in ferredoxin : NAD(P)H oxidoreductase (FNR) under Cu limitation (Hippmann et al. 2017) imply that there is indeed a surplus of reduced ferredoxin (Fdred) and NAD(P)H in the chloroplast. Here, we describe how the interaction between various metabolic pathways (e.g. nitrogen assimilation, glycolysis, citrate and the urea cycle) and the sophisticated coordination between the chloroplast and the mitochondria facilitate the re-oxidation of Fdred and NAD(P)H in the chloroplast.
Carbon metabolism – the Calvin-Benson-Bassham cycle is down-regulated via its activase, and glycolysis is used to redistribute ATP and NAD(P)H within the cell
The three most thoroughly annotated diatom genomes [T. pseudonana, Armbrust et al, 2004; P. tricornutum, Bowler et al, 2008; F. cylindricus, Mock et al, 2017] revealed many isoenzymes, particularly those involved in C metabolism. Indeed, homologous C metabolism isoenzymes exist among and between these diatoms (Gruber and Kroth, 2017; Kroth et al., 2008; Smith et al., 2012), and their differential expression is thought to manage cellular carbon flow. Furthermore, given that within the chloroplast more than 50% of the proteins involved in glycolysis are also part of the Calvin-Benson-Bassham cycle (Smith et al., 2012), to regulate C flow, some isoenzymes might be preferentially involved in glycolysis over carbon fixation. For example, in P. tricornutum, the three plastidial fructose-bisphosphate aldolases (FBAs) are differently targeted and regulated under low vs. high Fe conditions (Allen et al., 2012), (Table 4). Here, we hypothesize that to overcome Cu limitation, T. oceanica down-regulates the Calvin-Benson-Bassham cycle, while modulating glycolysis to promote the redistribution of ATP and NAD(P)H reducing equivalents among cellular compartments.
Similarly to P. tricornutum, Cu-limited T. oceanica also regulates the expression of FBA homologs, albeit differently than Fe-limited (Table 4): while the chloroplast FBA (FbaC2 homolog, To12069) is up-regulated, one of the pyrenoid-associated FBAs is only mildly up-regulated (FbaC1 homolog, To00388). We propose that FbaC2 is preferentially involved in glycolysis over C assimilation. This is supported by: (1) C assimilation decreased by 66% in Cu-limited cultures compared to the control (Hippmann et al, 2017), suggesting it is less likely for the C fixation proteins to be up-regulated; (2) the three significantly up-regulated proteins involved in the Calvin-Benson-Bassham cycle can also be part of glycolysis (i.e. PGK, TPI, and FBA, Table 2); (3) none of the distinct Calvin-Benson-Bassham cycle proteins (i.e. Rubisco, RPI, and RPE) were differentially expressed; (4) the red algal-type Rubisco activase (cbbX) was down-regulated by 2.25-fold. The down-regulation of cbbX results in slower carbon fixation and activity of Rubisco proteins―though Rubisco levels remain unchanged (Mueller-Cajar et al., 2011). Since RPI and RPE abundance remain constant, ribulose-bisphosphate would be bound to Rubisco. Consequently, once nutrient conditions are favorable, only the cbbX would need up-regulation for C fixation to proceed. We suggest that this strategy might be advantageous in nutrient limited environments with short-lived nutrient-rich conditions.
In general, most reactions facilitated by proteins in glycolysis can proceed in either direction, i.e. glycolysis or gluconeogenesis. Smith et al. (2012) suggest that gluconeogenesis prevails in the mitochondria. However, assuming that the required metabolite transporters are present in the mitochondria (e.g. aspartate/glutamate shuttle, malate/2–oxoglutarate shuttle, citrate/malate shuttle, and fumarate/succinate shuttle), modelling flux balances in P. tricornutum predict that glycolysis would indeed be more favorable than gluconeogenesis in the mitochondria (Kim et al., 2016).
In T. oceanica, in each cellular compartment, different subsets of glycolytic proteins were up- or down-regulated under Cu limitation (Fig. 4, Table S 4, overview Fig. S 5). Focusing on the up-regulated proteins (Fig. S 3), a pattern emerges suggesting ATP formation in the chloroplast and cytosol, as well as NAD(P)H consumption in the chloroplast and its coupled formation in the mitochondria. By reducing chloroplast GAPDH (To13085) and increasing mitochondrial GAPDH (To33331), NAD(P)H reducing equivalents would be generated in the mitochondria, whereas increasing PGK (To07617) in the chloroplast would increase ATP in this compartment. Therefore, a decreased ATP/NAD(P)H ratio in the plastid would be predicted under Cu limitation.
Interestingly, Hockin et al (2012) postulated that T. pseudonana increases glycolytic activity when nitrogen starved. However, when we mapped the involved proteins in T. pseudonana to their cellular target compartments, a regulation of isoenzymes similar to the response of Cu-limited T. oceanica emerges (i.e. PK down-regulated in mitochondria and up-regulated in the cytosol, Fig. S 3, Table S 4). Thus, the coordinated regulation of particular glycolytic isoenzymes to distribute NAD(P)H reducing equivalents and/or ATP production might be a general trait in diatoms.
Nitrogen metabolism is essential for Fdred oxidation and glutamate synthesis to fight ROS
Another striking feature in the response to Cu limitation in T. oceanica was the up-regulation of nitrogen acquisition and assimilation (Fig. 5, Table S 6, overview Fig. S 5). In plants, nitrogen assimilation is an important sink for excess NAD(P)H (Hoefnagel et al., 1998). In T. oceanica, the up-regulation of nitrate assimilation may alleviate the stress incurred by low Cu, namely by re-oxidizing Fdred in the chloroplast. This is achieved via up-regulation of only those NiR isoenzymes that use Fdred as their cofactor (To00016, To02363). Glutamine synthase (GSII, To31900) and the Fdred-dependent glutamate synthase (GOGAT, To13288) were also up-regulated, thereby easing the chloroplast electron pressure. The importance of this strategy for Cu-limited cells is strengthened by the fact that both the membrane-bound urea (To31656) and nitrate (To04919) transporters are among the 15 highest up-regulated proteins in our dataset, while its carbon:nitrogen content ratio remains unaffected (Kim and Price, 2017).
Counteracting reactive oxygen species – glutathione, thioredoxin, and superoxide dismutases
An enhanced nitrogen assimilation increases glutamate, which can be incorporated into (or be a precursor of) glutathione (GSH, γ-L-glutamyl-L-cysteine-glycine) to detoxify ROS via either direct scavenging or the ascorbate-glutathione cycle (Foyer and Noctor, 2011). Glutathione biosynthesis involves: (1) the cytosolic glutamate cysteine ligase (GCL, also known as γ-glutamylcysteine synthase, GCS) which combines glutamate and cysteine to γ-glutamyl-L-cysteine and (2) the plastid glutathione synthase (GSS) which adds glycine. Strikingly, both proteins were up-regulated in Cu-limited T. oceanica. However, in plants, the rate-limiting step in glutathione production is cysteine biosynthesis (Zechmann, 2014). Under Cu limitation, two chloroplast cysteine synthase isoenzymes were up-regulated (CS, To27524 and To10442; Fig. 7, Table 3, Table S 7) indicating an increase in glutathione production. Furthermore, glutathione-S-transferase was one of the most highly up-regulated proteins (GST, To09062), which is able to add glutathione to nucleophilic groups to detoxify oxidative stress (Gallogly and Mieyal, 2007). The up-regulation of glutathione reductase (GR, To07268), which oxidizes the over-abundant NAD(P)H in the chloroplast further supports that in T. oceanica glutathione counteracts ROS.
Thioredoxins (TXN) are important redox regulators in plants, especially in the chloroplast (Balmer et al., 2003), although their role in diatoms is unclear (Weber et al., 2009). In T. oceanica, three thioredoxins were up-regulated, and each one was targeted to either the chloroplast (TXN, To31425), the cytosol (To05213), or the mitochondria (To31425).
Another defence mechanism against ROS is the production of superoxide dismutases (SOD), which catalyze the conversion of superoxide radicals into hydrogen peroxide and oxygen. Of the three SODs identified in Cu-limited cultures, two were up-regulated: chloroplast Mn/Fe-SOD (To02860) and cytosolic Ni-SOD (To10112). Thus, under Cu limitation, cells are able to control ROS levels by increasing the expression of both glutathione and SODs. The increase of thioredoxin isoenzymes in all three major cellular compartments (i.e. cytosol, chloroplast, mitochondria) points to their involvement in sensing the cellular redox state and regulating excess NAD(P)H.
The malate shunt drains NAD(P)H reducing equivalents from the chloroplast to the mitochondria, thus integrating the nitrogen and carbon metabolisms
The efficiency of photosynthesis (both electron transport and carbon fixation) depends on an adequate supply of ATP/ADP and NAD(P)H/NAD(P)+ (Allen, 2002). In plants, the malate shunt can channel excess NAD(P)H reducing equivalents from the chloroplast to other cellular compartments, via the differential regulation of malate dehydrogenase (MDH) isoenzymes (Heineke et al., 1991; Scheibe, 2004). In this process, NAD(P)H in the chloroplast reduces oxaloacetate to malate, a compound that can be transported across membranes and re-oxidized, resulting in the production of NAD(P)H in the target compartment. NAD(P)H can then be used in reactions such as nitrate reduction in the cytosol or ATP production in the mitochondria.
In diatoms, the interaction between the chloroplast and mitochondria is expected to be multifaceted, possibly with direct exchange of ATP/ADP (Bailleul et al., 2015) and indirect exchange of NAD(P)H via the ornithine/glutamate shunt (Broddrick et al., 2019; Levering et al., 2016) and the malate/aspartate shunt (Bailleul et al., 2015; Prihoda et al., 2012). Some support for the spatial interconnectedness between chloroplast and mitochondria in diatoms has been found recently (Flori et al., 2017). However, the location of the potential transporters needed (e.g. malate/2-oxoglutarate antiporter, glutamate/aspartate antiporter) have yet to be proven (Bailleul et al., 2015; Kim et al., 2016; Kroth et al., 2008).
We present here compelling evidence for an activated malate/aspartate shunt in T. oceanica in response to low Cu. We observe the up-regulation of chloroplast and mitochondrial MDH (MDH2, To30817; MDH1, To03405), as well as mitochondrial aspartate aminotransferase (AAT2, To15049, Fig. 6). Chloroplast oxaloacetate (OAA) is reduced to malate via MDH2. Malate is then transported into the mitochondria via a putative malate/2-oxoglutarate antiporter (Kim et al., 2016). NAD(P)H reducing equivalents are released in the mitochondria via the re-oxidation of malate to OAA by mitochondrial MDH1. In turn, mitochondrial AAT2 transfers an amine group from glutamate to OAA, thereby releasing aspartate and 2-oxoglutarate into the mitochondria. To close the cycle, aspartate is transported back, via a glutamate/aspartate antiporter, into the chloroplast where the plastidial AAT isoenzyme would resupply OAA. However, chloroplast AAT was significantly down-regulated. We suggest that chloroplast OAA, the substrate for MDH2, is resupplied in the chloroplast via the ATP-dependent carboxylation of pyruvate due to the significant up-regulation of pyruvate carboxylase (PC). This leads to a net decrease of NAD(P)H in the chloroplast and a net increase of NAD(P)H in the mitochondria. Furthermore, the channeling of NAD(P)H reducing equivalents towards respiration, instead of the Calvin-Benson-Bassham cycle, is supported by a 66% decreased in C fixation, while respiration rates remained constant (Hippmann et al, 2017).
The increase in 2-oxoglutarate and aspartate in the mitochondria, due to an up-regulation of mitochondrial AAT2, can be helpful for the cell. If the putative malate/2-oxoglutarate antiporter is indeed involved in the malate shunt, 2-oxoglutarate will be transported back into the chloroplast. As chloroplast AAT is down-regulated, 2-oxoglutarate can be used as a substrate for the up-regulated Fd-dependent glutamate synthase (GOGAT) in nitrogen assimilation. Any surplus 2-oxoglutarate in the mitochondria can feed into the citrate cycle. Fittingly, aconitase (To20545) and isocitrate dehydrogenase (To34595), the two proteins involved in the citrate cycle immediately before 2-oxoglutarate, were both significantly down-regulated (Fig. 3).
Mitochondrial aspartate can be channelled into the urea cycle, where it will produce argininosuccinate, which can then be diverted back into the mitochondrial citrate cycle as fumarate via the aspartate/fumarate shunt (Allen et al., 2011). Thus, even though two of the first three steps in the citrate cycle were down-regulated, the malate shunt in combination with the urea cycle would ensure the continuation of this vital metabolic pathway by supplying it with essential carbon skeletons, i.e. 2-oxoglutarate and fumarate.
In addition to the malate shunt, other pathways have been proposed to alleviate electron pressure in diatoms. In P. tricornutum, modelling experiments suggest the prevalence of the glutamine-ornithine shunt over the malate shunt (Broddrick et al., 2019). However, none of the homologs involved in this shunt were identified in Cu-limited T. oceanica (e.g. n-acetyl-γ-glutamyl-phosphate reductase; n-acetylornithine aminotransferase). Furthermore, the activation of alternative oxidase (AOX) in Fe-limited P. tricornutum to alleviate electron stress in the impaired mitochondrial respiration (Allen et al., 2008) was not observed in Cu-limited T. oceanica (Hippmann et al., 2017). Future research is needed to elucidate the regulation of shuttle system/compartmental cross talks in diatoms.
Conclusion
The success of diatoms in the modern ocean is thought to be due to their complex genomic makeup, and their successful integration and versatility of metabolic pathways. This was exemplified in the present study, where we show how interaction among metabolic pathways act to maximize growth in T. oceanica (CCMP 1003) acclimated to severe Cu-limiting conditions. Our data show increased metabolic cross-talk between (1) Photosynthesis – N metabolism (0 Nitrogen metabolism, Fig. 5): the overreduced state of the photosynthetic apparatus results in an increase in Fdred, which was then oxidized by an N-assimilatory, Fdred-dependent isoenzyme located in the chloroplast; (2) Nitrogen metabolism and ROS fighting (Glutathione and ROS, Fig. 7): the increase in glutamate and cysteine synthase, as well as other key proteins in the glutathione metabolism, ensures the ability to counteract ROS; (3) Photosynthesis and Malate Shunt (Malate Shunt, Fig. 6): excess NAD(P)H reducing equivalents generated by the photosynthetic apparatus are channeled from the chloroplast to the mitochondria via the malate shunt. (4) Photosynthesis – glycolysis – carbon fixation (Carbon Metabolism, Fig. 3): to counteract the imbalance between ATP/NAD(P)H in the chloroplast, specific reactions in glycolysis occur in different compartments, where NAD(P)H reducing equivalents or ATP are needed (i.e. NAD(P)H generation in mitochondria, ATP production in cytosol and chloroplast). (5) Malate shunt – urea cycle– citrate cycle – glycolysis (Malate Shunt, Nitrogen Metabolism, Carbon Metabolism): the products of the malate shunt can feed into the TCA cycle and the urea cycle―the master cellular C and N redistribution hub. Furthermore, pyruvate can be carboxylated in the chloroplast to feed into the malate shunt, again transferring both NAD(P)H reducing equivalents and carbon skeletons from the chloroplast to the mitochondria. The up-regulation of N assimilation in response to chronic low Cu in TO03 contrasts the response of TO03 to acute Fe limitation, as well as the response of P. tricornutum to N limitation. Whether this is due to differences in species, nutrients or level of stress remains an intriguing question.
Methods
Cell Culturing
Strains CCMP 1003 and CCMP 1005 of the centric diatom T. oceanica (here referred to as TO03 and TO05, respectively) were obtained from the Provasoli-Guillard Center for Culture of Marine Phytoplankton, now National Centre for Marine Algae and Biota (NCMA) at Bigelow Laboratory for Ocean Sciences. Their identity as the same species was confirmed by ITS sequencing (Hippmann et al., 2017). For both strains, triplicate 10 L cultures of Cu-replete and Cu-limited strains were grown and harvested as detailed in Hippmann et al. (2017), with samples taken for a variety of physiological parameters and for differential proteomic analysis, as described also therein.
Note that in our study, the cells were acclimated to low Cu concentrations for many generations. Hence, the acclimation strategies in their physiology and proteome are not sudden, short-lived stress responses, but rather another ‘normal’ state for the cell to sustain growth under low Cu conditions.
Protein purification and preparation for mass spectrometry
Cells from the triplicate cultures were harvested by filtration, concentrated by centrifugation, flash-frozen with liquid N2 and stored at −80°C until final processing. Protein extraction and preparation for differential proteomic analysis by mass spectrometry have been described in Hippmann et al. (2017).
Liquid chromatography-tandem mass spectrometry – LC-MS/MS
For TO03, purified peptides were analyzed on the linear-trapping quadrupole-Orbitrap mass spectrometer (LTQ-Orbitrap Velos; ThermoFisher Scientific) on-line coupled to an Agilent 1290 Series HPLC using a nanospray ionization source (ThermoFisher Scientific) including a 2 cm long, 100 μm-inner diameter fused silica trap column, 50 μm-inner diameter fused silica fritted analytical column and a 20 μm-inner diameter fused silica gold coated spray tip (6 μm-diameter opening, pulled on a P-2000 laser puller from Sutter Instruments, coated on Leica EM SCD005 Super Cool Sputtering Device). The trap column was packed with 5 μm-diameter Aqua C-18 beads (Phenomenex, www.phenomenex.com), while the analytical column was packed with 3 μm-diameter Reprosil-Pur C-18-AQ beads (Dr Maisch, www.Dr-Maisch.com). Buffer A consisted of 0.5% aqueous acetic acid, and buffer B consisted of 0.5% acetic acid and 80% acetonitrile in water. The sample was loaded onto the trap column at 5 μL min−1 and the analysis was performed at 0.1 μL min−1. Samples were eluted with a gradient method where buffer B was from 10% to 25% over 120 min, from 25% to 60% over 20 min, from 60% to 100% B over 7 min, kept at 100% for 2.5 min and then the column was reconditioned for 20 min with buffer A. The HPLC system included Agilent 1290 series Pump and Autosampler with Thermostat set at 6°C. The LTQ-Orbitrap was set to acquire a full-range scan at 60,000 resolution from 350 to 1600 Th in the Orbitrap to simultaneously fragment the top fifteen peptide ions by CID in each cycle in the LTQ (minimum intensity 200 counts). Parent ions were then excluded from MS/MS for the next 30 sec. Singly charged ions were excluded since in ESI mode peptides usually carry multiple charges. The Orbitrap was continuously recalibrated using the lock-mass function. The mass error measurement was typically within 5 ppm and was not allowed to exceed 10 ppm.
TO05’s purified peptides were analyzed using a quadrupole – time of flight mass spectrometer (Impact II; Bruker Daltonics) on-line coupled to an Easy nano LC 1000 HPLC (ThermoFisher Scientific) using a Captive nanospray ionization source (Bruker Daltonics) including a column setup identical to that for TO03. Buffer A consisted of 0.1% aqueous formic acid, and buffer B consisted of 0.1% formic acid and 80% acetonitrile in water. Samples were run on a gradient method where buffer B was from 5% to 20% over 45 min, from 20% to 40% over 45 min then to 100% over 2 min, held at 100% for 15 min. Re-equilibration back to 5% buffer B was done separately by the LC automatically. The LC thermostat temperature was set at 7°C. The sample was loaded onto the trap column at 800 Bar and the analysis was performed at 0.25 μL min−1. The Impact II was set to acquire in a data-dependent auto-MS/MS mode fragmenting the 17 most abundant ions (one at the time) after each full-range scan from m/z 200 Th to m/z 2000 Th. The isolation window for MS/MS was 2 to 3 Th depending on parent ion mass to charge ratio. Parent ions were then excluded from MS/MS for the next 0.4 min. Singly charged ions were excluded since in ESI mode peptides usually carry multiple charges. The error of mass measurement was typically within 5 ppm and was not allowed to exceed 10 ppm.
Analysis of mass spectrometry data
Analysis of mass spectrometry data was performed using MaxQuant 1.5.1.0. The first search (herein “main dataset”) was performed against a database composed of the protein sequences from the sequenced genome of strain CCMP 1005 (publicly available) plus common contaminants using the default MaxQuant parameters with match between run and re-quantification options turned on. Only those peptides exceeding the individually calculated 99% confidence limit (as opposed to the average limit for the whole experiment) were considered as accurately identified. A second extended search was performed (herein “extended dataset”) with identical search parameters but against a larger database that combined protein sequences from both the TO05 genome and predicted proteins from the assembled EST contigs of the TO03 transcriptome (Hippmann et al., 2017). For reasons of clarity, if not noted otherwise, only differential expression data from the main dataset is discussed, as both datasets were in good agreement. Differential expression data from both the main and the extended dataset are presented in the supplementary Table S 2-S 9. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Vizcaíno et al., 2016) partner repository with the dataset identifier PXD006237.
Statistical analysis of differential protein expression
As described above, peptides were labelled with different isotopologues of formaldehyde depending on their nutrient regime (i.e. control, lowCu, lowFeCu), then mixed in a 1:1:1 ratio and analyzed by LC-MS/MS. Differential expression is then derived from the ratio of the intensities (area under the curve) of the light and heavy peaks for each peptide (see schematic in Fig. S 1). All three nutrient regimes (control, lowCu, lowFeCu) were processed together to be as consistent as possible and to decrease the number of false positives. However, in the present study we discuss the lowCu data only.
We defined two levels of statistically significant difference in expression: 1) greater than or equal to 2-fold (highly regulated), or 2) between 1.3- and 2-fold (regulated). In addition, the result must be found in at least two of the three biological replicates and result in a p-value of <0.05 for the z test, determining significant difference of the average ratios between treatments, taking the variance into account. Additionally, any protein that had a differential expression ratio of >10 (up-regulated) or <0.1 (down-regulated) in at least one biological replicate was considered to be an all-or-nothing response and was included in the ‘significantly changed’ set.
Protein annotation and targeting prediction
Predicted proteins from both the publicly available genome of TO05 (CCMP1005) and our transcriptome of TO03 (CCMP 1003) were searched against a comprehensive protein database, phyloDB for functional annotation using BlastP. PhyloDB version 1.076 consists of 24,509,327 peptides from 19,962 viral, 230 archaeal, 4,910 bacterial, and 894 eukaryotic taxa. It includes peptides from the 410 taxa of the Marine Microbial Eukaryotic Transcriptome Sequencing Project (http://marinemicroeukaryotes.org/), as well as peptides from KEGG, GenBank, JGI, ENSEMBL, CAMERA, and various other repositories. To predict gene localization for proteins involved in carbon and nitrogen metabolism, four in silico strategies were followed: 1) sequences of candidate genes were compared to the publicly available chloroplast genomes of T. oceanica (CCMP 1005) (Lommer et al., 2010) and T. pseudonana (Armbrust et al., 2004; Oudot-Le Secq et al., 2007), 2) the diatom-specific chloroplast targeting sequence software ASAFind (Gruber et al., 2015) was used in conjunction with SignalP 3.0 (Petersen et al., 2011) to find nuclear encoded, chloroplast targeted proteins, 3) SignalP and TargetP (Emanuelsson et al., 2007) were used for mitochondrial targeting, and 4) comparison with curated subcellular locations of the closest homologs in T. pseudonana, P. tricornutum, and Fragilariopsis cylindricus genomes. We acknowledge that deducing cellular targeting via comparison to predicted or experimentally verified proteins in other diatoms can be challenging, as homologs can be found in different compartments depending on species (Gruber et al., 2015; Gruber and Kroth, 2017; Schober et al., 2019). The corresponding names of all protein abbreviations used throughout the present study (text and figures) are given in Table 1.
Supplemental Tables and Figures
Table S 1: Comparison of cellular localization of various carbon metabolic pathways. (modified from (Gruber and Kroth, 2017).
Table S 2: Differential expression and predicted cellular location of proteins involved in the Calvin-Benson-Bassham cycle in TO03 and TO05 cultured in low Cu conditions vs. control.
Table S 3: Proteins with triose-phosphate transporter PFAM and their expression in TO03 and TO05 in response to Cu limitation vs. control
Table S 4: Differential expression and predicted cellular location of proteins involved in glycolysis in TO03 and TO05 cultured in low Cu conditions vs. control
Table S 5: Differential expression and predicted cellular location of proteins involved in the tricarboxylic acid (TCA) / citrate cycle in TO03 and TO05 cultured in low Cu conditions vs. control. (for diagram, see Fig. 3)
Table S 6: Differential expression and predicted cellular location of proteins involved in nitrogen metabolism including the urea cycle in TO03 and TO05 cultured in low Cu conditions vs. control.
Table S 7: Differential expression and predicted cellular location of proteins involved in glutathione metabolism in TO03 and TO05 cultured in low Cu conditions vs. control.
Table S 8: Differential expression and predicted cellular location of proteins involved in the putative malate shunt in TO03 and TO05 cultured in low Cu conditions vs. control
Table S 9: Differential expression and predicted cellular location of proteins involved in respiration in TO03 and TO05 cultured in low Cu conditions vs. control.
Fig. S 1: Overview of Proteomic Method. A) Workflow, B) Table of preparation and mixing of samples analyzed by LC-MS/MS.
Fig. S 2: Clustal alignment of predicted amino acid sequences of TpMDH2 [Tp25953, original (TpMDH2_old) and EST extended (TpMDH2_new, Smith et al., 2012)] and ToMDH2 (To30817). The new predicted cleavage sequence in TpMDH2_new is underlined.
Fig. S 3: Comparison of differential expression of proteins involved in glycolysis under chronic Cu limitation and acute N limitation.: A) T. oceanica (CCMP1003) under chronic Cu limitation (present study); B) T. oceanica (CCMP 1005) under chronic Cu limitation (present study, supplementary tables); C) T. pseudonana under acute N limitation (Hockin et al., 2012).
Fig. S 4: Differential expression of proteins involved in pyruvate metabolism.
Fig. S 5: An overview of the proteomic response in the nitrogen and carbon metabolisms in T. oceanica (CCMP 1003) grown under Cu-limiting conditions.
Fig. S 6: An overview of the proteomic response in the nitrogen and carbon metabolisms in T. oceanica (CCMP 1005) grown under Cu-limiting conditions.
Notes S 1: Discussion on the contrasting adaptations to Cu limitation in the two strains of T. oceanica (CCMP1003 and CCMP1005).
Acknowledgements
We wish to thank Angele Arrieta and Michael Murphy (Department of Microbiology and Immunology, UBC) for critical reading of the manuscript and NSERC for funding this research (Maria T Maldonado and Leonard J Foster).