One of the goals for developing cell factories is finding flexible organisms that can be rapidly tailored to produce any of a large range of products using the most cost effective substrate available. Unicellular diatoms have the potential to meet this role due to their unique metabolic capabilities and the ease in which they can be genetically manipulated. While Pt has demonstrated a host of advantageous characteristics for cell factories, its utility has been limited by the perception that it cannot consume simple sugars such as glucose.
The ILEs reported in this work have convincingly shown that Pt mixotrophically metabolizes glucose and uses the resulting carbon to synthesize each of the 15 amino acids we measured. As these amino acids are synthesized from multiple nodes encompassing all of primary metabolism across at least three separate intracellular compartments (cytosol, plastid, mitochondrion), it is reasonable to generalize that Pt metabolizes glucose and uses its carbon for the full range of biosynthetic activities. Given this information, it is natural to question why Pt mixotrophically consumes glucose only under light. Our gene expression analysis revealed that transcription level changes of membrane glucose transporters in Pt poorly correlate with exposure to light or glucose. This suggests that the inability of Pt to grow on glucose in the dark is not due to insufficient expression of glucose transporters. However, we cannot discount the possibility that the products of either or both GLUT1 and GLUT3 does not transport glucose into the cell, but instead shuttles glucose from the vacuole to the cytosol.
In Pt, light availability has a significant effect over a 24 h period on the expression levels of many genes encoding enzymes in central carbon metabolism . Of the genes encoding glycolysis and glucan biosynthesis, cytosolic Fba3 is most strongly regulated by light, suggesting that its product may be a rate-limiting enzyme in this pathway. Our gene expression analysis assays confirmed that Fba3 expression is upregulated by light and further showed that glucose has a negligible regulatory effect under light or dark. Allen et al.  showed that of the five fructose bisphosphate aldolase-3 genes in Pt, cytosolic Fba3 is the only one actively involved in glycolysis and gluconeogenesis, facilitating synthesis of photosynthetically fixed triose phosphates into chrysolaminaran (β:1–3 and β:1–6 glucose polymers) . Our results taken together with previous work on the role of Fba3, suggest that in the dark, glucose metabolism is impeded either by the lack of sufficient Fba3 expression or insufficient transport of glucose into the cell. Our ongoing work is focused on testing the hypothesis that glucose is not transported in the dark by elucidating a light-dependent mechanism for glucose transport and metabolism.
We tested several models of carbohydrate metabolism in Pt. Based upon the SSR criteria, it is clear that an accurate metabolic model includes multiple glycolytic pathways. However, it is certainly possible that other reactions or different combinations of the reactions we chose could produce equally valid results. For example, Model VIII nearly approached the acceptable SSR threshold despite not utilizing the ED and PPK pathways. In this instance, photosynthesis and photorespiration nearly accounted for the unusual observed labeling patterns. If some variation of this model were accurate, it would indicate that Pt uses photosynthesis to re-fix a significant fraction of the CO2 that it generates from intracellular reactions, thus conserving organic carbon.
These conflicting possibilities are ordinarily resolved through the use of parallel labeling experiments using large sets of measured metabolites, which greatly increase the confidence intervals of key fluxes and narrow the number of possible pathway models to one or only a small handful of options . However, the high degree of uncertainty regarding the metabolic significance and role of these pathways poses a significant challenge for large-scale 13C metabolic flux analysis in Pt. In addition, the compartmentalization of various catabolic and biosynthetic pathways between the cytosol, plastid and mitochondrion is incompletely defined, and the possible effects of metabolic channeling have not been investigated . As a result, we limited our MPA to a single compartment model only including measurements of known cytosolic and mitochondrial amino acids . Due to these simplifications, this article does not report a full quantitative flux map for glucose metabolism using the measurements from multiple ILEs, as was done for multiple bacterial species with active ED pathways . Nevertheless, MPA unraveled much information on the usage of pathways, providing a basis for further developing a quantitative flux map of glucose metabolism in Pt.
MPA suggested that the ED pathway plays an important role in glucose catabolism. Our models consistently showed cycling from both the ED and PPK pathways through the reverse EMP pathway. Such a cycle generates significant excess NADPH that is available for numerous functions including increased lipid production, quenching of reactive oxidative species, and nitrogen assimilation. The traditional oxidative PPP pathway maximally yields 2 moles each of NADPH and NADH per mole of glucose with a 56% carbon yield for lipid synthesis. In comparison, the maximal ED/PPK pathway yields are 1.67/3.33 moles of NADPH and 2.33/0.67 moles of NADH with a matching 67% carbon yield. This hypothesis that Pt can use a combination of the ED and PPK pathways to enhance carbon and cofactor yields for lipid synthesis agrees with the previously reported 17% increase in Pt EPA content when glucose was used in a mixotrophic fed-batch system . Such a metabolic cycle need not be limited to mixotrophic growth on glucose, as the upper reactions in the EMP shuttle carbon in this manner to replenish the supply of 5-C compounds during photosynthesis. Therefore, it is also plausible this metabolic cycling enhances Pt’s ability to synthesize such a high proportion of its biomass as lipids [2–4].
Our gene expression analysis also hints at metabolic cycling, as Fba3 expression levels were largely unchanged in response to differing carbon sources (dissolved CO2, HCO3
-, and glucose), yet changing the nitrogen source from nitrates to urea dramatically inhibited expression in both light and dark conditions. Given that Fba3 is a key regulator of a major carbon assimilation pathway during photosynthesis, we expected that different carbon sources would have a larger effect on expression levels than changing the nitrogen source. Urea is produced as a consequence of amino acid breakdown, which is then converted to ammonia for nitrogen assimilation. Cells normally synthesize ammonia from nitrates, which requires a substantial supply of reductant. Therefore, urea inhibition of Fba3 expression may serve to regulate reductant generation.
It is also important to note that the majority of the EMP pathway genes have been found in the cytosol, mitochondria, and plastid; however the gene encoding enolase in the lower half of the pathway has only been identified in the plastid and mitochondria, e.g. from the Diatomcyc metabolic pathway database . Therefore, Pt may lack a complete cytosolic EMP and instead utilize a combination of the ED and PPK pathways to feed the lower half of the mitochondrial EMP.
Another key result we predicted by using MPA and subsequently confirmed through a 2-13C glycerol ILE was that serine and glycine are largely derived from glyoxylate instead of 3-phosphoglycerate under some mixotrophic growth conditions. The 2-13C glycerol label was used instead of a 2-13C glucose label in an attempt to minimize scrambling of the label through the PPP, ED, and upper EMP. Identification of the precursor of serine/glycine could be complicated by PPP carbon atom rearrangements (mostly) and phototrophic 13CO2 re-fixation. Both these effects can produce 1-13C glyceraldehyde 3-phosphate 1-13C 3-phosphoglycerate and 1-13C serine. However, since pyruvate and alanine are synthesized downstream of 3-phosphoglycerate without net carbon rearrangement, the aforementioned effects should also manifest in the isotopomer distribution of alanine and should generate 1-13C alanine with an abundance greater than the abundance of 1-13C serine (26%). However, our measurements indicate that the abundance of 1-13C alanine is only 6%, substantially less than 26%. Therefore, the caveats of PPP carbon rearrangement and phototrophic 13CO2 fixation are insufficient to explain the observed 1-13C serine, leaving glyoxylate-based synthesis as the only remaining explanation.
These findings are important for future metabolic flux analysis on Pt, as prior work on bacteria and plants grown on glucose has long held that 3-phosphoglycerate [24, 35, 36] is the sole source of serine and glycine. Further explorations into the mixotrophic synthesis of serine and glycine using multiple organic carbon substrates are warranted, as both AGAT and SHMT are key intermediate reactions in photorespiration and our analysis of the serine and glycine isotopomers indicated that the reactions are active when grown on glycerol. However, the calculated labeling on the transferred methyl group of THF was significantly lower than the labeling on C-2 of glycine, indicating that GDCH is minimally active. Formate is the other major methyl group donor to THF, which is formed as a byproduct of reactions in sterol and cofactor synthesis pathways. Therefore, it is possible that photorespiratory synthesis of glycine and serine is partially due to the need for Pt to recycle THF and clear formate from the cell . Formate is also enzymatically oxidized to CO2; however, the reincorporation of formate into serine minimizes carbon loss and should thus be the metabolically favored reaction.