Experimental evidence and isotopomer analysis of mixotrophic glucose metabolism in the marine diatom Phaeodactylum tricornutum

Abstract

Background: Heterotrophic fermentation using simple sugars such as glucose is an established and cost-effective method for synthesizing bioproducts from bacteria, yeast and algae. Organisms incapable of metabolizing glucose have limited applications as cell factories, often despite many other advantageous characteristics. Therefore, there is a clear need to investigate glucose metabolism in potential cell factories. One such organism, with a unique metabolic network and a propensity to synthesize highly reduced compounds as a large fraction of its biomass, is the marine diatom Phaeodactylum tricornutum (Pt). Although Pt has been engineered to metabolize glucose, conflicting lines of evidence leave it unresolved whether Pt can natively consume glucose.

Results: Isotope labeling experiments in which Pt was mixotrophically grown under light on 100% U-(13)C glucose and naturally abundant (~99% (12)C) dissolved inorganic carbon resulted in proteinogenic amino acids with an average 13C-enrichment of 88%, thus providing convincing evidence of glucose uptake and metabolism. The dissolved inorganic carbon was largely incorporated through anaplerotic rather than photosynthetic fixation. Furthermore, an isotope labeling experiment utilizing 1-(13)C glucose and subsequent metabolic pathway analysis indicated that (i) the alternative Entner-Doudoroff and Phosphoketolase glycolytic pathways are active during glucose metabolism, and (ii) during mixotrophic growth, serine and glycine are largely synthesized from glyoxylate through photorespiratory reactions rather than from 3-phosphoglycerate. We validated the latter result for mixotrophic growth on glycerol by performing a 2-(13)C glycerol isotope labeling experiment. Additionally, gene expression assays showed that known, native glucose transporters in Pt are largely insensitive to glucose or light, whereas the gene encoding cytosolic fructose bisphosphate aldolase 3, an important glycolytic enzyme, is overexpressed in light but insensitive to glucose.

Conclusion: We have shown that Pt can use glucose as a primary carbon source when grown in light, but cannot use glucose to sustain growth in the dark. We further analyzed the metabolic mechanisms underlying the mixotrophic metabolism of glucose and found isotopic evidence for unusual pathways active in Pt. These insights expand the envelope of Pt cultivation methods using organic substrates. We anticipate that they will guide further engineering of Pt towards sustainable production of fuels, pharmaceuticals, and platform chemicals.

Figure 1

Principal pathways for the mixotrophic metabolism of glucose and CO₂ to amino acids in Pt. Central carbon metabolic pathways convert glucose and/or CO₂ (fixed photosynthetically or anaplerotically) to the 15 amino acids (metabolites shown as open circles) experimentally detected by GC-MS in hydrolysates of Pt cell pellets. In most organisms, glycolysis proceeds via the EMP pathway. However, two alternate glycolytic pathways of bacterial origin were found in this organism’s annotated genome. Of these, the phosphoketolase (PKP) enzyme converts phosphorylated pentose and/or hexose sugars to glyceraldehyde 3-phosphate/erythrose 4-phosphate and acetylphosphate, which is then converted to either acetate via acetate kinase, or acetyl-CoA via phosphate acetyltransferase. Both phosphorylated pentose and hexose sugars are shown as substrates for the PPK pathway because the enzyme specificity in Pt is unknown. The second alternative pathway (ED) uses two enzymes to convert 6-phospho-D-gluconate to pyruvate and glyceraldehyde 3-phosphate. Differences in the carbon atom rearrangements of the EMP, PPK and ED pathways become evident in the MIDs of glycolytic amino acids.

Figure 2

¹³C-enrichments of amino acid fragments synthesized from 100% and 50% U-¹³C glucose evidence significant glucose uptake. The 41 measured proteinogenic amino acid fragments in cell hydrolysates of Pt are grouped according to their metabolic precursor(s) (Figure 1). In each of the 100% and 50% U-¹³C glucose ILEs, the fragments show a ¹³C-enrichment approximately proportional to the ¹³C enrichment of the supplied glucose. In contrast, purely photoautotrophic cells would only be ¹³C-enriched to the 1.1% natural abundance CO₂ from the flask headspace. A 2% dilution by initially present biomass and a combination of anaplerotic and photosynthetic inorganic carbon fixation explain the slightly lower average enrichments than would be expected for cells consuming glucose as their sole carbon source: 88% ± 3% in the 100% U-¹³C glucose ILE and 45% ± 1% in the 50% U-¹³C glucose ILE.

Figure 3

Pathways used to assemble the metabolic pathway analysis models. This metabolic network includes 10 metabolic pathways distinguished by line color. Metabolic Models I-IX consist of different combinations of these pathways. Open circles represent amino acids detected in cell hydrolysates, whose isotope labeling patterns were used in the MPA. The amino acids are connected to their metabolic precursor(s) by dotted lines. Genes encoding proteins catalyzing all reactions in these pathways were found in the annotated Pt genome.

Figure 4

MPA of the 100% 1-¹³C glucose ILE data suggests an active ED pathway. (a) Four different Metabolic Models (I-IV) were constructed to explain the ILE data; each column represents a different model. Pathways included in a model are denoted by a “Y”, color-coded according to the color of the pathway in Figure 3. (b) Vertical bars represent the SSR of each model and horizontal lines represent the acceptable SSR corresponding to the number of redundant isotopomer measurements in each model. Models II and IV, both containing the ED pathway show a significantly decreased SSR compared to otherwise identical models lacking the ED pathway. (c) The carbon rearrangements of the EMP and ED pathways are shown, with ¹³C atoms shown as blue squares and ¹²C atoms shown as white squares. The EMP pathway transfers ¹³C from glucose C-1 to pyruvate C-3, whereas the ED pathway transfers ¹³C from glucose C-1 to pyruvate C-1 (red boxes). (d) The isotopomers of pyruvate reflect those of the amino acid alanine. The measured abundances of alanine isotopomers are compared against the simulated enrichments of Models I and II. Model I that lacks the ED pathway over-simulates the abundance alanine{123}, and under-simulates the abundance of alanine{123}. These errors are corrected in Model II, which utilizes the ED pathway. Isotopomer notation is explained in text.

Figure 5

Unique carbon-carbon bond re-arrangements explain abnormal isotope abundances from the 100% 1-¹³C Glc ILE. (a) Five different Metabolic Models (V-IX) were constructed to explain the ILE data; each column represents a different model. Pathways included in a model are denoted by a “Y”, color-coded according to the color of the pathway in Figure 3. (b) Vertical bars represent the SSR of each model and horizontal lines represent the acceptable SSR (65) corresponding to the number of redundant isotopomer measurements (45) in each model. (c) The carbon rearrangements of SGAT and SHMT demonstrate how 3-¹³C triose phosphates (derived from 1-¹³C glucose through glycolysis) result in 2-¹³C serine and glycine. Serine is conventionally known to be synthesized directly from 3-phosphoglycerate without carbon rearrangements via PSAT shown with the dashed arrow. (d) The photorespiratory action of RuBisCO yields glyoxylate from pentose phosphate, whereas the PPK pathway yields acetate that is converted to glyoxylate via the glyoxylate shunt. Each pathway yields 2-¹³C glycine and serine from 1-¹³C pentose phosphate arising from the reductive PPP. (e) Anaplerotic fixation of a mixture of intracellular ¹²CO₂ and ¹³CO₂ results in 1-¹³C pyruvate through reversible reactions in the TCA cycle. Succinate is a symmetric molecule; therefore C-1 and C-4 are equivalent. Oxaloacetate is 6% ¹³C-enriched at the C-4 position and 13% enriched at C-1. In comparison, pyruvate is 18% enriched at C-1, indicating that anaplerotic fixation cannot fully account for the labeling on pyruvate. (f) The oxidative PPP yields ¹³CO₂ and U-¹²C ribulose 5-phosphate from 1-¹³C glucose-6-phosphate. Photosynthetic fixation of CO₂ via RuBisCO then results in 1-¹³C pyruvate. (g) The predominant isotopomers of serine and glycine that were simulated in the poorly fit Model VII and the well fit Model IX are compared against the experimental abundance of each isotopomer.

Figure 6

100% 2-¹³C glycerol ILE shows that glycine and serine are predominantly synthesized from glyoxylate rather than 3-phosphoglycerate. Feeding 2-¹³C glycerol to Pt confirmed the MPA prediction that the MIDs of glycine and serine do not represent the labeling of 3-phosphoglycerate as is usual in many organisms. Were this the case, the majority of the ¹³C label from 2-¹³C glycerol would appear on the C-2 of glycine and serine, contradicting observation. The observed isotope labeling patterns in serine and glycine can be explained as follows. First, 2-¹³C glycerol is metabolized to pyruvate and alanine. Carbon rearrangements in the TCA cycle (gray) and back-mixing through anaplerotic reactions and the pentose phosphate pathway account for the small amount of label on alanine{1}. As 3-phosphoglycerate and pyruvate are closely linked to one-another, their MID’s are assumed to be identical. The high abundance of the glycine{12} and serine{123} result from the conversion of pyruvate to acetyl-CoA and glyoxylate. Aminotransferases convert glyoxylate to glycine, which then combines with MTHF to form serine via SHMT. A linear combination of the fluxes from alanine to serine and (glycine + MTHF) to serine produced a set of isotopomers that exactly matched the measured values when the SHMT reaction contributed 91% of the total flux and phosphoserine transaminase contributed 9% of the flux. Arrow widths correspond to relative fluxes. See text for isotopomer notation.

Figure 7

Expression levels of key glucose-related and glycolytic genes under different environmental conditions. Expression levels of genes under eight conditions (light/L1, dark/L1, light/HCO₃^–, dark/ HCO₃^–, light/Glc, dark/Glc, light/urea, and dark urea, see Materials and methods for details), as compared to the dark/L1 condition. Light exposure induces a significant increase in the expression of the cytosolic fructose bisphosphate aldolase 3 gene (Fba3) encoding the protein that catalyzes the reversible conversion of fructose 1,6-bisphosphate to glyceraldehyde 3-phosphate and dihydroxyacetone phosphate. The overexpression is independent of the presence of glucose or bicarbonate in the growth media. Conversely, the expression levels of the two genes corresponding to known glucose transporters in Pt (GLUT1 and GLUT3) appear to be unaffected by light exposure and the presence of organic carbon sources. Fold changes were calculated with respect to the housekeeping gene 18S and were verified with respect to two other housekeeping genes. Results are presented as mean ± SD of three biological and three technical replicates (a total of 9 replicates per gene and condition). *: 0.01 < p < 0.05 when compared to the dark/L1 condition; **: p ≤ 0.01 when compared to the dark/L1 condition.