Novel Aspect Data fusion of positive and negative ion mode DESI-HRMS mass spectra represents an innovative strategy to improve sample characterization/classification. Introduction Several data fusion strategies have been proposed to improve sample characterization/classification by combining the information provided by different analytical platforms and to overcome high-dimensionality issues. Microscopic biological samples represent an analytical challenge due to the low amount and complexity of their analytes. Such samples can be well represented by individual bovine oocytes and preimplantation embryos. By using high mass resolution desorption electrospray ionization (DESI-HRMS) with a compatible solvent system, single preimplantation embryos can be directly analyzed in both ion mode conditions to increase the range of lipid species detected. Therefore, the aim of this work is to provide a more comprehensible metabolic description of embryonic developmental changes by applying data fusion strategy on positive and negative ion mode mass spectra. Methods Bovine oocytes and blastocysts were analyzed by DESI-HRMS in both positive and negative ion modes to profile free fatty acids (FFA), phospholipids (PL), cholesterol-related molecules, and triacylglycerols (TAG). DESI mass spectra were acquired (ca. 1 min) over the m/z range 600-1200 in the positive ion mode, using acetonitrile doped with silver nitrate as solvent. Subsequently, the same embryos were analyzed in the negative ion mode over the m/z range 150-1000, using 1:1 (v/v) acetonitrile/DMF. Principal Component Analysis (PCA) was applied as a data fusion strategy and exploratory tool functional to chemically characterize embryos. Linear discriminant analysis (LDA) was then performed on the fused data for classification purposes. Data processing was performed by means of in-house Matlab routines. Preliminary Data In the negative ion mode DESI high resolution mass spectra, free fatty acids (mainly palmitic, oleic, stearic and arachidonic acids), cholesterol sulphate, several phosphatidylethanolamine (PE), phosphoatidylserine (PS), phosphatidylglycerol (PG) and phosphatidylinositol (PI) lipid species were detected. In the positive ion mode, several cholesteryl esters (CE) and triacylglycerols (TAG) besides squalene (identified for the first time in mammalian preimplantation embryos and oocytes) and ubiquinone were detected. Normalization to respect of the total ion current (TIC) allowed mass spectra to be used as embryo fingerprints and comprehensively compared by means of multivariate analysis, in order to interpret the changes in the lipid profiles according to developmental stages and growing conditions. PCA as a data fusion strategy proved to be extremely efficient in characterizing the embryos on the basis of the DESI-HRMS chemical signatures. When LDA was applied to the positive and negative ion mode data set separately, 79.6% and 89.8% classification accuracy was achieved, respectively. Nevertheless, the LDA strategy on the fused data led to a 95.9% accurate classification for immature oocytes, matured oocytes, blastocysts cultured in-vivo and blastocysts cultured in–vitro. Lipid changes detected by DESI-HRMS were in agreement with up- or down-regulation of selected genes critically involved in lipid synthesis and intracellular cholesterol homeostasis, which confirmed the reliability and feasibility on this analytical protocol. Remarkably, ambient ionization methods, which allow intact samples to be directly analyzed, appear to be suitable for a data fusion strategy either on homogeneous MS-data (same technique and different operating conditions, as presented in this work) or on heterogeneous MS-data (different MS platforms). Multivariate data analysis using data fusion allowed to merge different sources of mass spectral information, thus to explore and interpret the relations among samples and original MS variables, leading to a more exhaustive chemical characterization of complex and microscopic biological samples.
Embryonic metabolic status evaluated by combining desi-hrms positive and negative ion mode mass spectral data by data fusion strategy
PIRRO, VALENTINA;
2013-01-01
Abstract
Novel Aspect Data fusion of positive and negative ion mode DESI-HRMS mass spectra represents an innovative strategy to improve sample characterization/classification. Introduction Several data fusion strategies have been proposed to improve sample characterization/classification by combining the information provided by different analytical platforms and to overcome high-dimensionality issues. Microscopic biological samples represent an analytical challenge due to the low amount and complexity of their analytes. Such samples can be well represented by individual bovine oocytes and preimplantation embryos. By using high mass resolution desorption electrospray ionization (DESI-HRMS) with a compatible solvent system, single preimplantation embryos can be directly analyzed in both ion mode conditions to increase the range of lipid species detected. Therefore, the aim of this work is to provide a more comprehensible metabolic description of embryonic developmental changes by applying data fusion strategy on positive and negative ion mode mass spectra. Methods Bovine oocytes and blastocysts were analyzed by DESI-HRMS in both positive and negative ion modes to profile free fatty acids (FFA), phospholipids (PL), cholesterol-related molecules, and triacylglycerols (TAG). DESI mass spectra were acquired (ca. 1 min) over the m/z range 600-1200 in the positive ion mode, using acetonitrile doped with silver nitrate as solvent. Subsequently, the same embryos were analyzed in the negative ion mode over the m/z range 150-1000, using 1:1 (v/v) acetonitrile/DMF. Principal Component Analysis (PCA) was applied as a data fusion strategy and exploratory tool functional to chemically characterize embryos. Linear discriminant analysis (LDA) was then performed on the fused data for classification purposes. Data processing was performed by means of in-house Matlab routines. Preliminary Data In the negative ion mode DESI high resolution mass spectra, free fatty acids (mainly palmitic, oleic, stearic and arachidonic acids), cholesterol sulphate, several phosphatidylethanolamine (PE), phosphoatidylserine (PS), phosphatidylglycerol (PG) and phosphatidylinositol (PI) lipid species were detected. In the positive ion mode, several cholesteryl esters (CE) and triacylglycerols (TAG) besides squalene (identified for the first time in mammalian preimplantation embryos and oocytes) and ubiquinone were detected. Normalization to respect of the total ion current (TIC) allowed mass spectra to be used as embryo fingerprints and comprehensively compared by means of multivariate analysis, in order to interpret the changes in the lipid profiles according to developmental stages and growing conditions. PCA as a data fusion strategy proved to be extremely efficient in characterizing the embryos on the basis of the DESI-HRMS chemical signatures. When LDA was applied to the positive and negative ion mode data set separately, 79.6% and 89.8% classification accuracy was achieved, respectively. Nevertheless, the LDA strategy on the fused data led to a 95.9% accurate classification for immature oocytes, matured oocytes, blastocysts cultured in-vivo and blastocysts cultured in–vitro. Lipid changes detected by DESI-HRMS were in agreement with up- or down-regulation of selected genes critically involved in lipid synthesis and intracellular cholesterol homeostasis, which confirmed the reliability and feasibility on this analytical protocol. Remarkably, ambient ionization methods, which allow intact samples to be directly analyzed, appear to be suitable for a data fusion strategy either on homogeneous MS-data (same technique and different operating conditions, as presented in this work) or on heterogeneous MS-data (different MS platforms). Multivariate data analysis using data fusion allowed to merge different sources of mass spectral information, thus to explore and interpret the relations among samples and original MS variables, leading to a more exhaustive chemical characterization of complex and microscopic biological samples.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
