Limits...
High-throughput metabolic screening of microalgae genetic variation in response to nutrient limitation.

Bajhaiya AK, Dean AP, Driver T, Trivedi DK, Rattray NJ, Allwood JW, Goodacre R, Pittman JK - Metabolomics (2015)

Bottom Line: Limitation of nutrients including nitrogen and phosphorus can induce metabolic changes in microalgae, including the accumulation of glycerolipids and starch.These results demonstrate that the PSR1 gene is an important determinant of lipid and starch accumulation in response to phosphorus starvation but not nitrogen starvation.However, the SNRK2.1 and SNRK2.2 genes are not as important for determining the metabolic response to either nutrient stress.

View Article: PubMed Central - PubMed

Affiliation: Faculty of Life Sciences, The University of Manchester, Michael Smith Building, Oxford Road, Manchester, M13 9PT UK.

ABSTRACT

Microalgae produce metabolites that could be useful for applications in food, biofuel or fine chemical production. The identification and development of suitable strains require analytical methods that are accurate and allow rapid screening of strains or cultivation conditions. We demonstrate the use of Fourier transform infrared (FT-IR) spectroscopy to screen mutant strains of Chlamydomonas reinhardtii. These mutants have knockdowns for one or more nutrient starvation response genes, namely PSR1, SNRK2.1 and SNRK2.2. Limitation of nutrients including nitrogen and phosphorus can induce metabolic changes in microalgae, including the accumulation of glycerolipids and starch. By performing multivariate statistical analysis of FT-IR spectra, metabolic variation between different nutrient limitation and non-stressed conditions could be differentiated. A number of mutant strains with similar genetic backgrounds could be distinguished from wild type when grown under specific nutrient limited and replete conditions, demonstrating the sensitivity of FT-IR spectroscopy to detect specific genetic traits. Changes in lipid and carbohydrate between strains and specific nutrient stress treatments were validated by other analytical methods, including liquid chromatography-mass spectrometry for lipidomics. These results demonstrate that the PSR1 gene is an important determinant of lipid and starch accumulation in response to phosphorus starvation but not nitrogen starvation. However, the SNRK2.1 and SNRK2.2 genes are not as important for determining the metabolic response to either nutrient stress. We conclude that FT-IR spectroscopy and chemometric approaches provide a robust method for microalgae screening.

No MeSH data available.


Related in: MedlinePlus

Partial least squares discriminant analysis (PLS-DA) prediction results of non-stressed (high P and high N) grown wild type in comparison to cells cultured in response to P limitation (low P) and N limitation (low N) at day-7 of growth. Ability of PLS-DA linear regression models trained using replicate spectra to predict the identification of high P/N spectra (a), low P spectra (b) and low N spectra (c). The predicted Y values represent a yes (1) or no (0) classification decision for each replicate sample (an average of three technical replicates) in the validation set. Error bars indicate 95 % confidence interval around each predicted Y value. Training and validation data consisted of independent biological replicates, and each dataset contained 9 control TAP replicates, 9 low P replicates and 5 low N replicates
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4644200&req=5

Fig2: Partial least squares discriminant analysis (PLS-DA) prediction results of non-stressed (high P and high N) grown wild type in comparison to cells cultured in response to P limitation (low P) and N limitation (low N) at day-7 of growth. Ability of PLS-DA linear regression models trained using replicate spectra to predict the identification of high P/N spectra (a), low P spectra (b) and low N spectra (c). The predicted Y values represent a yes (1) or no (0) classification decision for each replicate sample (an average of three technical replicates) in the validation set. Error bars indicate 95 % confidence interval around each predicted Y value. Training and validation data consisted of independent biological replicates, and each dataset contained 9 control TAP replicates, 9 low P replicates and 5 low N replicates

Mentions: While the lowest concentrations of P and N induced the most marked metabolic changes in the cells, with respect to high lipid and carbohydrate accumulation, the fresh weight biomass yields from these treatments were extremely low; 0.91 and 0.54 mg mL−1 for the P and N starved cultures, respectively. In contrast, significant increases in lipid and carbohydrate were still observed in the 10 µM P and 0.7 mM N treated cells, but with sufficient biomass. These two nutrient stress conditions (referred to as low P and low N) were then used to examine whether cells exposed to specific stress conditions could be predicted and identified from control (non-stressed) strains using the FT-IR spectral data. To classify stressed and non-stressed strains, a PLS-DA statistical analysis was used to develop predictive models of variation between the strains. The model was generated using a training set of replicate spectra derived from nine control and low P samples and five low N samples (Supplementary Fig. 2). These models used three factors for prediction which accounted for 97 % of the total explained variance. An equal number of test spectra were then evaluated using the PLS-DA model. The Y values of 1 or 0 were set as a yes or no decision as to whether or not a sample belongs to the assigned class with a value of 0.5 as a decision borderline. The model was able to predict accurately strains grown in nutrient replete medium and distinguish them from low P and low N strains (Fig. 2a). Furthermore, despite both nutrient stresses being equivalent in ability to induce lipid and carbohydrate induction, the model was able to accurately distinguish low P strains (Fig. 2b) and low N strains (Fig. 2c) from the rest of the strains. There appears to be differences in the amount of lipid and starch synthesized in response to 0.7 mM N compared to 10 µM P (Fig. 1), which may partly explain the ability of the model to differentiate the N and P stressed cells. However, this might also suggest that other metabolic responses differ between the two nutrient treatments in addition to just lipid and carbohydrate. This has been indicated in previous C. reinhardtii metabolomic studies. For example, a GC–MS analysis demonstrated that N limited cells are metabolically distinct from P limited cells and each treatment gives rise to specific changes in distinct amino acids, organic acids and sugars (Bolling and Fiehn 2005). Notably, the lysine biosynthesis metabolite 2-amino-adipic acid increased 9-fold in N limited cells, while tryptophan increased nearly 5-fold, but many metabolites decreased significantly following N limitation including a number of amino acids, fumarate, glucose and malate. In contrast, P limitation induced a 25-fold increase in cysteine concentration, and 4-fold increases in citrate and glycerate, while there were relatively few decreases of amino acids. Likewise, there are clear transcriptional distinctions between P and N limitation, with just a ~5 % similarity in transcriptional response observed between the two stress treatments (Schmollinger et al. 2014). For example, there were no transcript changes related to photosynthetic function that were common between the P and N limitation conditions.Fig. 2


High-throughput metabolic screening of microalgae genetic variation in response to nutrient limitation.

Bajhaiya AK, Dean AP, Driver T, Trivedi DK, Rattray NJ, Allwood JW, Goodacre R, Pittman JK - Metabolomics (2015)

Partial least squares discriminant analysis (PLS-DA) prediction results of non-stressed (high P and high N) grown wild type in comparison to cells cultured in response to P limitation (low P) and N limitation (low N) at day-7 of growth. Ability of PLS-DA linear regression models trained using replicate spectra to predict the identification of high P/N spectra (a), low P spectra (b) and low N spectra (c). The predicted Y values represent a yes (1) or no (0) classification decision for each replicate sample (an average of three technical replicates) in the validation set. Error bars indicate 95 % confidence interval around each predicted Y value. Training and validation data consisted of independent biological replicates, and each dataset contained 9 control TAP replicates, 9 low P replicates and 5 low N replicates
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4644200&req=5

Fig2: Partial least squares discriminant analysis (PLS-DA) prediction results of non-stressed (high P and high N) grown wild type in comparison to cells cultured in response to P limitation (low P) and N limitation (low N) at day-7 of growth. Ability of PLS-DA linear regression models trained using replicate spectra to predict the identification of high P/N spectra (a), low P spectra (b) and low N spectra (c). The predicted Y values represent a yes (1) or no (0) classification decision for each replicate sample (an average of three technical replicates) in the validation set. Error bars indicate 95 % confidence interval around each predicted Y value. Training and validation data consisted of independent biological replicates, and each dataset contained 9 control TAP replicates, 9 low P replicates and 5 low N replicates
Mentions: While the lowest concentrations of P and N induced the most marked metabolic changes in the cells, with respect to high lipid and carbohydrate accumulation, the fresh weight biomass yields from these treatments were extremely low; 0.91 and 0.54 mg mL−1 for the P and N starved cultures, respectively. In contrast, significant increases in lipid and carbohydrate were still observed in the 10 µM P and 0.7 mM N treated cells, but with sufficient biomass. These two nutrient stress conditions (referred to as low P and low N) were then used to examine whether cells exposed to specific stress conditions could be predicted and identified from control (non-stressed) strains using the FT-IR spectral data. To classify stressed and non-stressed strains, a PLS-DA statistical analysis was used to develop predictive models of variation between the strains. The model was generated using a training set of replicate spectra derived from nine control and low P samples and five low N samples (Supplementary Fig. 2). These models used three factors for prediction which accounted for 97 % of the total explained variance. An equal number of test spectra were then evaluated using the PLS-DA model. The Y values of 1 or 0 were set as a yes or no decision as to whether or not a sample belongs to the assigned class with a value of 0.5 as a decision borderline. The model was able to predict accurately strains grown in nutrient replete medium and distinguish them from low P and low N strains (Fig. 2a). Furthermore, despite both nutrient stresses being equivalent in ability to induce lipid and carbohydrate induction, the model was able to accurately distinguish low P strains (Fig. 2b) and low N strains (Fig. 2c) from the rest of the strains. There appears to be differences in the amount of lipid and starch synthesized in response to 0.7 mM N compared to 10 µM P (Fig. 1), which may partly explain the ability of the model to differentiate the N and P stressed cells. However, this might also suggest that other metabolic responses differ between the two nutrient treatments in addition to just lipid and carbohydrate. This has been indicated in previous C. reinhardtii metabolomic studies. For example, a GC–MS analysis demonstrated that N limited cells are metabolically distinct from P limited cells and each treatment gives rise to specific changes in distinct amino acids, organic acids and sugars (Bolling and Fiehn 2005). Notably, the lysine biosynthesis metabolite 2-amino-adipic acid increased 9-fold in N limited cells, while tryptophan increased nearly 5-fold, but many metabolites decreased significantly following N limitation including a number of amino acids, fumarate, glucose and malate. In contrast, P limitation induced a 25-fold increase in cysteine concentration, and 4-fold increases in citrate and glycerate, while there were relatively few decreases of amino acids. Likewise, there are clear transcriptional distinctions between P and N limitation, with just a ~5 % similarity in transcriptional response observed between the two stress treatments (Schmollinger et al. 2014). For example, there were no transcript changes related to photosynthetic function that were common between the P and N limitation conditions.Fig. 2

Bottom Line: Limitation of nutrients including nitrogen and phosphorus can induce metabolic changes in microalgae, including the accumulation of glycerolipids and starch.These results demonstrate that the PSR1 gene is an important determinant of lipid and starch accumulation in response to phosphorus starvation but not nitrogen starvation.However, the SNRK2.1 and SNRK2.2 genes are not as important for determining the metabolic response to either nutrient stress.

View Article: PubMed Central - PubMed

Affiliation: Faculty of Life Sciences, The University of Manchester, Michael Smith Building, Oxford Road, Manchester, M13 9PT UK.

ABSTRACT

Microalgae produce metabolites that could be useful for applications in food, biofuel or fine chemical production. The identification and development of suitable strains require analytical methods that are accurate and allow rapid screening of strains or cultivation conditions. We demonstrate the use of Fourier transform infrared (FT-IR) spectroscopy to screen mutant strains of Chlamydomonas reinhardtii. These mutants have knockdowns for one or more nutrient starvation response genes, namely PSR1, SNRK2.1 and SNRK2.2. Limitation of nutrients including nitrogen and phosphorus can induce metabolic changes in microalgae, including the accumulation of glycerolipids and starch. By performing multivariate statistical analysis of FT-IR spectra, metabolic variation between different nutrient limitation and non-stressed conditions could be differentiated. A number of mutant strains with similar genetic backgrounds could be distinguished from wild type when grown under specific nutrient limited and replete conditions, demonstrating the sensitivity of FT-IR spectroscopy to detect specific genetic traits. Changes in lipid and carbohydrate between strains and specific nutrient stress treatments were validated by other analytical methods, including liquid chromatography-mass spectrometry for lipidomics. These results demonstrate that the PSR1 gene is an important determinant of lipid and starch accumulation in response to phosphorus starvation but not nitrogen starvation. However, the SNRK2.1 and SNRK2.2 genes are not as important for determining the metabolic response to either nutrient stress. We conclude that FT-IR spectroscopy and chemometric approaches provide a robust method for microalgae screening.

No MeSH data available.


Related in: MedlinePlus