Software LS-MIDA for efficient mass isotopomer distribution analysis in metabolic modelling
- Zeeshan Ahmed†1, 7,
- Saman Zeeshan†1, 8,
- Claudia Huber5,
- Michael Hensel2,
- Dietmar Schomburg3,
- Richard Münch4,
- Wolfgang Eisenreich5 and
- Thomas Dandekar1, 6Email author
© Ahmed et al.; licensee BioMed Central Ltd. 2013
Received: 28 May 2012
Accepted: 23 June 2013
Published: 9 July 2013
The knowledge of metabolic pathways and fluxes is important to understand the adaptation of organisms to their biotic and abiotic environment. The specific distribution of stable isotope labelled precursors into metabolic products can be taken as fingerprints of the metabolic events and dynamics through the metabolic networks. An open-source software is required that easily and rapidly calculates from mass spectra of labelled metabolites, derivatives and their fragments global isotope excess and isotopomer distribution.
The open-source software “Least Square Mass Isotopomer Analyzer” (LS-MIDA) is presented that processes experimental mass spectrometry (MS) data on the basis of metabolite information such as the number of atoms in the compound, mass to charge ratio (m/e or m/z) values of the compounds and fragments under study, and the experimental relative MS intensities reflecting the enrichments of isotopomers in 13C- or 15 N-labelled compounds, in comparison to the natural abundances in the unlabelled molecules. The software uses Brauman’s least square method of linear regression. As a result, global isotope enrichments of the metabolite or fragment under study and the molar abundances of each isotopomer are obtained and displayed.
The new software provides an open-source platform that easily and rapidly converts experimental MS patterns of labelled metabolites into isotopomer enrichments that are the basis for subsequent observation-driven analysis of pathways and fluxes, as well as for model-driven metabolic flux calculations.
Metabolism is central for all cellular processes including adaptation of organisms to their respective life style and conditions. Triggered by the presence and activity of metabolic enzymes and the metabolite fluxes through pathways, cellular reactions constitute a highly dynamic network that can be rapidly and efficiently modulated in response to environmental changes. A number of theoretical techniques has been established to predict metabolic fluxes [1-4]. Implementing different mathematical parallel and sequential algorithms, several desktop and web based batch and interactive software applications  have been also developed towards quantitative metabolic flux analysis and modeling .
In contrast, only few methods allow direct determination of metabolic fluxes, one of which is based on in vivo experiments using stable isotope labelled precursors, such as 13C-glucose or 13CO2. The transfer of label to the metabolic network and the specific isotope distribution in metabolic products can then be taken as evidence of metabolic pathways and fluxes during the experimental period. However, robust technology is required to quantitatively determine the isotopomer abundances in multiple metabolites. Specifically, experimental intensities of mass signals (typically of silylated derivatives, metabolites and fragments thereof in GC/MS experiments) have to be converted into relative and molar isotopomer abundances.
Isotopologues are species of a compound that differ only in their isotopic composition . The term isotopomer is a contraction of ‘isotopic isomer‘, grouping isotopologues into those molecules which contain the same number of a specific isotope (e.g. 13C) at different positions. As an example, 64 stable carbon isotopologues exist for glucose. Out of these, six species constitute isotopomers with one 13C-atom at position 1, 2, 3, 4, 5, or 6. In natural compounds, i.e. obtained from the natural environment, the natural abundance is the consequence of the natural isotope abundance (i.e. ca. 1.1% for 13C at a given carbon position) that is diverted through the complete population of isotopologues due to statistical reasons. In contrast, increased isotopologue abundances are observed in labelling experiments where isotope-enriched precursors (e.g. 13C-labelled) are supplied to the organism under study. This results in the enrichment of specific isotopologues, i.e. on top of the natural abundances, in the metabolic products. The deconvolution of mass intensities yielding isotopomer enrichment is the key task of the software described in this manuscript. Notably, mass intensities provide information on the abundances of isotopologues harbouring a specific number of the isotope, i.e. one, two, three, etc. 13C-atoms, and therefore, the enrichment of isotopomeric groups (isotopomer distribution) is obtained. Since metabolic pathways lead to specific isotopomer enrichments and, as a consequence, to specific isotopomer distributions, the latter values can be used to identify and to quantify the relative contributions of metabolic routes from the labelled precursor to the products observed by MS.
So far, three different methods are available for positional isotopomer determination, nuclear magnetic resonance (NMR), mass spectrometric analysis of a sufficient number of useful metabolite fragments [8, 9], and multiple reaction monitoring (MRM). These methods can provide orthogonal information and can be combined using our software to improve positional isotopomer determination .
In this manuscript, a new freely available software is described that is capable of providing a user friendly graphical interface for the efficient and independent (no third party application is needed) data storage, management and processing towards mass isotopomer distribution analysis . The implemented software enables the user to load data from previously created data files or add data manually into the software application at run time and to process it. Furthermore, it directly parameterizes input experimental data to Brauman´s algorithm for accurate estimation of natural and relative abundances. No such application exists, implementing similar mathematics into a user friendly software package.
Currently, only commercial software or user-specific approaches are available for the conversion of mass intensities (provided by the specific software implemented to the mass spectrometer) to the relative and molar isotopomer enrichments, such as tandem mass spectrometric data computing for positional isotopomer distributions , measurements of mass distributions by mass spectrometry , isotopomer analysis using GC-MS [10, 14], and GC-MS analysis for isotopomer balancing .
However, for a broad range of users, a open-source software compatible to data exchange with the standard mass software packages is highly desirable. Here, we present a new open-source software using Brauman’s least square method for the calculation of isotopomer enrichment that can be used in GC/MS and LC/MS experiments (including tandem-MS/MS) by calculating relative and absolute isotopomer abundances from the mass ratios of signals in experimental MS spectra.
where A = relative natural abundance, n = number of carbon atoms, i = index variable to count n iterations. P0 and P1 stand for the abundance of 12C and 13C, respectively.
“The solution of these equations gives the abundance of each organic moiety. Because of the way in which the problem was formulated, the total abundance of the organic moieties must remain constant.” This statement by Brauman  has now to be put into practical calculations. However, the results of this technique depend upon a number of factors: (i) the analysis is based on the assumption that the fragmentation patterns for all heteroatom isotopes are identical (i.e., no differential isotope effect), (ii) the experimental relative abundance of 12C and 13C isotopes induced through derivatization is known and (iii) the relative natural abundances of the isotopes are either known or measured. The LS-MIDA software package itself is not designed to perform integration of the original MS signals. Thus, another software package must be used first. For our examples, it relied on pre-processing by the software LabSolutions by Shimadzu which is standard software directly supplied with the instrumentation for GC-MS. However, any type of pre-processing software can be used in combination with LS-MIDA.
Here, A1, A2, A3, A4 … An are the estimated relative natural abundance values of fragments using eq. 2. These values are used in the abundance matrix A [17, 18], based for linear regression analysis. In general, obtaining the coefficients of matrix A (eq. 3) is the non-trivial part of the method. More specifically, the atoms and their isotope distributions from the derivatisation agent must be taken into account, if present. Overlaps of mass traces by impurities have obvious effects and the mode of ionization (positive or negative) has effects on the m/z values. Our implementation tackles the processing of the pre-processed data to overcome some of these problems. For example, contributions due to the derivatisation agent are filtered out.
Here, the linear regression analysis (initially used by Brauman ) is performed for spectral data analysis, where U is the mass isotopomer distribution, and S and T are the isotope abundances for 12C and 13C, respectively.
In eq. 6, Ri1 Ri2 Ri3, Ri4 ,… Rin are the observed relative intensity values with respect to the m/e values. Now, eq. 1 is applied to estimate the string of calculated relative abundance values (Ra(1-n)) for the fragments, where A = Abundance Matrix (Ri (1-n) ).
The absolute 13C-enrichment is equal to the sum of all labeled isotopomers multiplied with the respective number of labels (0 to n), divided by the number of carbon atoms in the fragment under study. A indicates the labeled isotopomer, the index 0 till n indicates the number of labeled atoms in the fragment, and a indicates the number of carbon atoms in the fragment. The matrix calculations are mathematical simple, however, the optimized combination of pre-filtering software and abundance calculations by LS-MIDA takes into account all required steps as well as experimental complications (e.g. filtering out derivatization agent) in a single user-friendly and open-source software package.
The above described mathematics and calculations were implemented into the software “LS-MIDA” (executable available as Additional file 1; test data in Additional file 2; pre-filtering software considerations see above). We show that the implementation of Brauman’s least square method and the inclusion of binomial expression allow accurate calculations of isotopomer enrichments using experimental GC/MS data of 13C-labelled silylated amino acids. LS-MIDA is a UML designed  and successfully evaluated third party tool independent reusable desktop application (batch) with user friendly graphical interface, capable of sequentially processing standard input and producing visual output presentation (text and spectrum).
Graphical user interface
LS-MIDA provides an intuitive graphical user interface for file based experimental raw data manipulation and management. It is capable of not only managing user input experimental raw data but also provides options to manage resultant data (output of LS-MIDA). It allows the user to create new data files, manage created data files, merge new or already made data files into one or more new or already created data files and manipulate entries of data files. It is an independent file based data management system that does not require any external or third party database to install and use. LS-MIDA is implemented and tested using Microsoft Windows (version XP and 7) operating system as it is developed using Microsoft Dot net framework and C Sharp (object oriented) programming language. The licensed software is freely available for academic use on request.
LS-MIDA advantageously provides a file-based data management system for experimental metabolic mass isotopomers based raw data. The data manager is a supporting utility, developed as a user-friendly file-based experimental data management system. It allows the user to create new experimental data files that later can be used for the analysis using data analyzer. The experimental data is organised following a new data format especially proposed (with extension “*.ls”) for LS-MIDA data files. Data manager allows the user to read, add, edit, update, delete and merge data (from other source files of the same extension) into a file.
Application in metabolite measurements
The software was now tested with different data sets. 13C-Labelled amino acid samples (analyzed as TBDMS-derivatives) were obtained from hydrolysates of Salmonella enterica grown in medium containing [U-13C6]glucose . We have shown earlier that [U-13C6]glucose is efficiently incorporated into most amino acids of Salmonella enterica via intermediates of glycolysis. Under these conditions, alanine is mainly composed of the unlabelled isotopomer (derived from unlabelled glucose in the medium) and the fully 13C-labelled isotopomer due to de novo synthesis of alanine from [U-13C3]pyruvate made from [U-13C6]glucose via glycolysis.
(A) Experimental raw data of alanine (Ala)
C Atom Metabolite
C Atom Fragment
(B) Alanine experimental raw data analysis 1
(C) Isotopomer calculation results (example: alanine)
N a %
R a %
As shown in Figure 7, abundances of four different alanine isotopomers are observed. For the isotopomer 000 (i.e. 12C only), the estimated natural abundance is 96.7%, whereas the same isotopomer accounts for 70.0% in the labelled compound. The isotopomer group with one 13C-atom has 3.26% or 0% abundance in the unlabelled or labelled compound, respectively. The relative abundances for the isotopomer group comprising two 13C-atoms are 0.037% or 0% in the unlabelled or labelled alanine sample, respectively, whereas the abundances for the fully labelled isotopomer are 0.00014% or 30.0% in the unlabelled or labelled sample, respectively. On this basis, the observed absolute enrichment value of 13C in the labelled sample from S. enterica results in 30.0%. This is in line with our expectations and calculations also using other software tools for isotopomer analysis.
The resulting spectrum is shown in Figure 7. At constant m/e values the peaks of the drawn spectrum may vary according to their molecular composition . The strongest observed relative intensity 13C isotopomer peak in this example is at 263.1 in the correct range of M o , M -1 , M max so the values for 260.1, 263.1 and 264.1 are shown in Figure 7 for the alanine mass spectrum. For more results of labelled TBDMS-amino acids from the same labelling experiment [20, 21], please have a look at Additional file 3: Table S2-S3.
For the pathway analyses of Salmonellae mentioned above and in similar studies on other bacteria, we tested the processing of different data sets; input data file preparation and management, experimentation and data analysis. This allowed also an analysis of error rates. Accuracy of the calculation itself is quite high (error less then 1 part per billion). This estimate is based on several hundred test runs of the software and meticulous testing for bugs and unexpected behavior. Regarding errors from the non-trivial coefficients of matrix A involving experimental errors from atoms of the derivatization agent present and overlap of fragment spectra as well as the mode of ionization (positive or negative), we expect error rates of less than 1%.
Another inherent source of error is the analysis of network fluxes for complex biological system that typically result in notably higher deviations when looking at the results from replicates. As these are indirectly inferred from the changes of measured isotopomer concentrations, these data already carry the error due to the mathematical procedures. In addition, experimental errors regarding growth conditions must be taken into account (cell number, actual glucose concentration, sample preparation). Indeed, the resulting overall error for flux analyses considering all these effects can be as high as 10% based on our experience. Some limitations arise from the fact that not in all pathways metabolized label is present, but this is an inherent prediction limitation for the approach.
13C Labelling of metabolites has proven to be a powerful method in quantifying metabolic routes and fluxes, especially, if there are alternative pathways between two metabolites. Isotopomer balancing provides the basis for deducing metabolic pathways and fluxes.
In own studies, we have shown that 13C-incorporation studies coupled to software-based isotopomer calculation allowed us to identify the pathways of amino acid biosynthesis  under relevant non-standardized conditions (e.g. proline biosynthesis in Listeria monocytogenes and its modulation by the transcription factor PrfA ). Another example concerns nutrient supply for Salmonella in the Salmonella containing vacuole . Again, the unequivocal determination of nutrient flow across the vacuole to Salmonella depends on isotopologue data with the use of suitable processing software.
In general, without transforming the mathematics into applied software doing the calculations, none of these and other insights [6, 20, 21] regarding flux modifications and usage of different metabolic pathways in different organisms is possible.
Despite this potential power of the methodological approach, it is still difficult to perform metabolic flux analyses due to the lack of user-friendly and open-source software tools. This limitation also demands the development of mathematical modelling of metabolism for each substrate to obtain more detailed and accurate results. Before the implementation of LS-MIDA, we relied on the usage of a lab-specific Excel/Solver-based software doing the required calculations. However, this approach did not provide user-friendly output formats nor included a database allowing extensive comparative studies. As an alternative to this lab-specific solution, commercial packages may also be used [12, 13, 15]. However, these software packages are not freely available.
In order to establish tools that can be widely distributed, we have established the LS-MIDA software. Briefly, Brauman´s least square algorithm is used and developed in the form of a versatile software application iteratively analyzing the estimated abundance resonances  after binomial expansion for the calculation of isotopomer enrichments in labelled metabolites.
Furthermore, LS-MIDA provides a file-based data management system for fast and accurate MS-based isotopomer analyses.
In comparison with other existing approaches [24-26], the combined features in LS-MIDA are not available in standard packages for metabolite modelling such as Metatool , Yanasquare , Gepasi  or FiatFlux  (here fluxes are predicted after the isotopologue data have been processed). There are two software solutions available for isotopomer data processing, Envelop  and Isotope Pattern Calculator , but none uses binomial expression for data extension. The implementation of Brauman’s least square method with the inclusion of binomial expression allows rapid and accurate calculation of isotopomer data.
LS-MIDA was compared with in-house software demonstrating its robustness. This showed that LS-MIDA can be used as an open-source platform for many (even non-expert) users in consortia in research programs such as the ongoing priority DFG program dealing with “host adapted metabolism of pathogens” in our example as well as other interested academic groups (German and foreign) for which we did provide and also will provide in future the software free of charge. Training courses and service are offered to support the use of the software within the program and for other users. In consequence, the software is freely available for the world-wide academic community.
LS-MIDA is a versatile, open-source, and user friendly software with rapid calculation, integrated overview on all isotopomers, least square correction and database management system, with good application potential for biology and biotechnology such as studying the core metabolism of organisms, the pathways and fluxes leading to desired products in biotechnology, and complementing methods from genomics, proteomics or metabolomics.
Availability and requirements
LS-MIDA is free available software for all academic users with open license; a commercial license can be obtained on request.
LS-MIDA is developed using the Microsoft C# (sharp) programming language and Microsoft Dot Net framework 2008. It is compatible (install and use) for all Microsoft Windows operating systems. Moreover, LS-MIDA automatically adopts the language of the installed operating system and presents numerical values accordingly e.g. in case of English language decimal values are ‘.’ (dot) separated and in case of German language decimal values are ‘,’ (comma) separated. Numerical values are separated by ‘#’ (hash) symbol for all languages.
Further details are available in the Additional file 3 Tutorial (installation, evaluation, further data, glossary).
We would like to thank Dr. U. Rapp-Galmiche for native speaker corrections and the Deutsche Forschungsgemeinschaft (grant Da 208/10-2, TR34/Z1 and EI-384/5-1) for funding. This publication was funded by the German Research Foundation (DFG) and the University of Wuerzburg in the funding programme Open Access Publishing.
- Nicola Z, Eliane F, Uwe S: FiatFlux - a software for metabolic flux analysis from 13C-glucose experiments. BMC Bioinforma. 2005, 6: 209-10.1186/1471-2105-6-209.View ArticleGoogle Scholar
- Dauner M, Bailey J, Sauer U: Metabolic flux analysis with a comprehensive isotopomer model in Bacillus subtilis. Biotechnol Bioeng. 2001, 76: 144-156. 10.1002/bit.1154.View ArticlePubMedGoogle Scholar
- Wiechert W: 13C metabolic flux analysis. Metab Eng. 2001, 3: 195-206. 10.1006/mben.2001.0187.View ArticlePubMedGoogle Scholar
- Rantanen A, Rousu J, Jouhten P, Zamboni N, Maaheimo H, Ukkonen E: An analytic and systematic framework for estimating metabolic flux ratios from 13C tracer experiments. BMC Bioinforma. 2008, 9: 266-10.1186/1471-2105-9-266.View ArticleGoogle Scholar
- Ahmed Z, Majeed S, Dandekar T: Computational Feature Performance and DSA Evaluation of Applications towards MFA. Recent Patents Comput Sci. 2012, 5: 3-View ArticleGoogle Scholar
- Dandekar T, Fieselmann A, Majeed S, Ahmed Z: Briefings in Bioinformatics, First online published. Software Applications toward Quantitative Metabolic Flux Analysis and Visualization. 2012, November 9Google Scholar
- Brenninkmeijer CA, Janssen C, Kaiser J, Röckmann T, Rhee TS, Assonov SS: Isotope effects in the chemistry of atmospheric trace compounds. Chem Rev. 2003, 103: 5125-5161. 10.1021/cr020644k.View ArticlePubMedGoogle Scholar
- Rousu JA, Rantanen RA, Ketola C, Juha T, Kokkonen C: Isotopomer distribution computation from tandem mass spectrometric data with overlapping fragment spectra. Spectroscopy. 2005, 19: 53-67. 10.1155/2005/575686.View ArticleGoogle Scholar
- Bequette BJ, Sunny NE, El-Kadi SW, Owens SL: Application of stable isotopes and mass isotopomer distribution analysis to the study of intermediary metabolism of nutrients. J Anim Sci. 2006, 84: E50-E59.PubMedGoogle Scholar
- Christensen B, Nielsen J: Isotopomer analysis using GC-MS. Metab Eng. 1999, 1: E8-E16.View ArticleGoogle Scholar
- Previs SF, Fernandez CA, Yang D, Soloviev MV, France D, Brunengraber H: Limitations of the Mass Isotopomer Distribution Analysis of Glucose to Study Gluconeogenesis. J Biol Chem. 1998, 277: 16853-16859.View ArticleGoogle Scholar
- Rantanen A, Rousu J, Ketola RA, Kokkonen JT, Tarkiainen V: Computing positional isotopomer distributions from tandem mass spectrometric data. Metab Eng. 2002, 4: 285-294. 10.1006/mben.2002.0232.View ArticlePubMedGoogle Scholar
- Winden WV, Wittman C, Heinzle E, Heijnen J: Correcting mass isotopomer distributions for naturally occurring isotopes. Biotechnol Bioeng. 2002, 80: 477-479. 10.1002/bit.10393.View ArticlePubMedGoogle Scholar
- Lee WN, Byerley LO, Bergner EA, Edmond J: Mass isotopomer analysis: theoretical and practical considerations. Biol Mass Spectrom. 1991, 20: 451-458. 10.1002/bms.1200200804.View ArticlePubMedGoogle Scholar
- Dauner M, Sauer U: GC-MS analysis of amino acids rapidly provides rich information for isotopomer balancing. Biotechnol Prog. 2000, 16: 642-649. 10.1021/bp000058h.View ArticlePubMedGoogle Scholar
- Brauman JI: Least Squares Analysis and Simplification of Multi-Isotope Mass Spectra. Anal Chem. 1966, 38: 607-610. 10.1021/ac60236a020.View ArticleGoogle Scholar
- Korzekwa K, Howald WN, Trager WF: The use of Brauman's least squares approach for the quantification of deuterated chlorophenols. Biomed Environ Mass Spectrom. 1990, 19: 211-217. 10.1002/bms.1200190402.View ArticlePubMedGoogle Scholar
- Pfeiffer T, Sanchez Valdenebro I, Nuno JC, Montero F, Schuster S: METATOOL: for studying metabolic networks. Bioinformatics. 1999, 15: 251-257. 10.1093/bioinformatics/15.3.251.View ArticlePubMedGoogle Scholar
- Ahmed Z, Majeed S, Dandekar T: Unified Modeling and HCI Mockup Designing towards MIDA. IJES. 2012, 2: 361-382.Google Scholar
- Dandekar T, Fieselmann A, Popp J, Hensel M: Salmonella enterica: a surprisingly well-adapted intracellular lifestyle. Front Microbiol. 2012, 3: 164-PubMed CentralView ArticlePubMedGoogle Scholar
- Eisenreich W, Dandekar T, Heesemann J, Goebel W: Carbon metabolism of intracellular bacterial pathogens and possible links to virulence. Nat Rev Microbiol. 2010, 8: 401-412. 10.1038/nrmicro2351.View ArticlePubMedGoogle Scholar
- Papageorgopoulos C, Caldwell K, Shackleton C, Schweingrubber H, Hellerstein MK: Measuring Protein Synthesis by Mass Isotopomer Distribution Analysis (MIDA). Anal Biochem. 1999, 267: 1-16. 10.1006/abio.1998.2958.View ArticlePubMedGoogle Scholar
- Baverel G, Conjard A, Chauvin MF, Vercoutere B, Vittorelli A, Dubourg L, Gauthier C, Michoudet C, Durozard D, Martin G: Carbon 13 NMR spectroscopy: a powerful tool for studying renal metabolism. Biochimie. 2003, 85: 863-871. 10.1016/j.biochi.2003.10.001.View ArticlePubMedGoogle Scholar
- Sykes MT, Williamson JR: Envelope: interactive software for modeling and fitting complex isotope distributions. BMC Bioinforma. 2008, 9: 446-10.1186/1471-2105-9-446.View ArticleGoogle Scholar
- Massila K, Soong HC, Azlianor AA, Muhammad SS: Reinforcing the concept of calculating isotope pattern using theoretical isotope generator (TIG). WSEAS Trans Inf Sci Appl. 2008, 5: 949-Google Scholar
- Jennings ME, Matthews DE: Determination of complex isotopomer patterns in isotopically labeled compounds by mass spectrometry. Anal Chem. 2005, 77: 6435-6444. 10.1021/ac0509354.PubMed CentralView ArticlePubMedGoogle Scholar
- Schwarz R, Liang C, Kaleta C, Kuhnel M, Hoffmann E, Kuznetsov S, Hecker M, Griffith G, Schuster S, Dandekar T: Integrated network reconstruction, visualization and analysis using YANAsquare. BMC Bioinforma. 2007, 8: 313-10.1186/1471-2105-8-313. 10 ppView ArticleGoogle Scholar
- Mendes P: GEPASI: A software package for modelling the dynamics, steady states and control of biochemical and other systems. CABIOS. 1993, 9: 563-571.PubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.