Reconstruction and Analysis of Metabolic Networks

Transcription

1 1 Reconstruction and Analysis of Metabolic Networks

2 2 Outline What is a Reconstruction? Data Collection Interactions Between Network Components Special Considerations Applications

3 Genome-scale Metabolic Model 3 Genome Annotation - by homology, location Biochemical Data - protein characterized Physiological Data - indirect, pathway known Inferred Reactions - indirect, inferred from biomass requirements Quantitative Analysis - simulate cell behavior - drive experimental studies Reconstruction ORGANISM Genome Annotation Biochemistry Network Reconstruction Metabolic Model Cell Physiology Inferred Reactions Quantitative Analytical Methods New Predictions Emergent Properties

4 Genome-scale Metabolic Model 4 Model Development: an iterative process Reconstruction ORGANISM - Biochemical data -Revised ORF assignments Genome Annotation Biochemistry Cell Physiology Inferred Reactions Network Reconstruction Computational, Biochemical Investigation Metabolic Model New Predictions Emergent Properties Quantitative Analytical Methods

5 What is in a reconstruction? 5 Genome: Annotated genes Gene location Regulatory regions Wobble base pairs Biochemistry: Stereochemistry ph and pka (charge) Elemental balance Charge balance Multiple reactions/enzyme Multiple enzymes/reaction Transcription/translation: Gene to transcript to protein to reaction association Transcript half-lives trna abundances Ribosomal capacities Physiology: Flux data Knock-outs Balanced functions Overall phenotypic behavior Location of gene product compartmentalization

6 Defining Metabolic Reactions 6 ydbh hslj ldha 1st level: Primary metabolites LAC 2nd level: Neutral Formulas C 3 H 6 O 3 Charged Formulas C 3 H 5 O 3 1-3rd level: 4th level: 5th level: prokaryotes Lactate Dehydrogenase Metabolite Specificity PYR Metabolite Formulas C 3 H 4 O 3 C 3 H 3 O 3 1- Stoichiometry C 21 H 26 N 7 O 14 P 2 1 LAC + 1 NAD? 1 PYR + 1 NADH + 1 H Thermodynamic Considerations: Directionality Localization eukaryotes NAD C 21 H 26 N 7 O 14 P 2 1- Coenzymes 1 LAC + 1 NAD 1 PYR + 1 NADH + 1 H NADH C 21 H 27 N 7 O 14 P 2 C 21 H 27 N 7 O 14 P 2 1- [c]: cytoplasm [n]: nucleus [m]: mitochondria [e]: extracellular [g]: golgi aparatus [x]: peroxisome [p]: periplasm [v]: vacuole [h]: chloroplast [l]: lysosome [r]: endoplasmic reticulum 1 LAC [c] + 1 NAD [c] 1 PYR [c] + 1 NADH [c] + 1 H [c] STEPWISE INCORPORATION OF INFORMATION

7 Sources of Information 7 Genome Sequence & Annotation Physiological Data Databases OD 600 Growth Measurements Time (Hours) Available Literature Phylogenetic Data Bacteria Archaea Eukarya Localization Signal sequences...pllllpisgsalp...

8 Network Assembly and Representation 8 Reconstruction of Glycolytic Pathway Abbr. HEX1 PGI PFK FBA TPI GAPD PGK PGM ENO PYK Glycolytic Reactions [c]glc +atp g6p + adp [c]g6p f6p [c]atp + f6p adp + fdp + h [c]fdp dhap + g3p [c]dhap g3p [c]g3p + nad + pi 13dpg + h + nadh [c]13dpg + adp 3pg + atp [c]3pg 2pg [c]2pg h2o + pep [c]adp + h + pep atp + pyr Genes glk pgi pfka,pfkb fbaa,fbab tpia gapa,gapc_1,gapc_2 pgk gpma,gpmb eno pyka,pykf HEX1 PGI PFK FBA TPI GAPD PGK PGM ENO PYK atp glc adp g6p h f6p fdp dhap g3p nad pi dpg nadh pg pg pep h2o pyr PYK: IF pyka OR pykf ENO: IF eno GAPD: IF gapa OR (gapc_1 AND gapc_2)

9 Network Evaluation 9 Precursor Metabolite Formation Filling Network Gaps Incorporating Biomass Composition ATP Maintenance Calculation ATP (mmol/gdw/h) ATP m ATP biomass D (1/h) Physiological Data Comparison Knockout Data Comparison qco 2 (mmol/gdw/h) Experiment Model qo 2 (mmol/gdw/h) Experiment Model D (1/h) D (1/h) P/O Ratio Calculation 1.5 H + 3 H + Outer Membrane Inner Membrane I II III IV ATPase ETS 3 H +

10 10 Data Collection I. Genome Annotation II. Biochemistry III. Physiology

11 I. Genome Annotation ,238 protein sequences derived from whole genomes (expected to reach 1 million by 2005) 101,602 entries in Swiss Prot High-quality annotation requires substantial effort

12 Genome Annotation: how to 12 Open Reading Frame (ORF) Identification - Start & Stop codons, GLIMMER. Traditional Annotation Methods - Experimental (direct) - Sequence homology - Generally covers 40-70% of new genomes New Annotation Methods - Protein-protein interactions - Correlated mrna expression levels - Phylogenetic profile clustering - Protein fusion - Gene neighbors (operon clustering)

13 Genome Databases: TIGR 13 The Comprehensive Microbial Resource (CMR) 353 completed bacterial genomes 28 completed archaeal genomes Single-genome analysis: Genome overview, list by role category (eg amino acid biosynthesis) analysis methods, searches Multi-genome analysis also available

14 Genome Databases: TIGR 14

15 Genome Databases: NCBI 15 Microbial Genomes Resources 595 completed microbial genomes (47 archael) FTP sites for Protein Annotations (ptt files)

16 II. Biochemical Data: Reactions stoichiometry and reversibility 16 Gene: glk Enzyme: Glucokinase Reaction: ATP + D-Glucose > ADP + D-Glucose 6-phosphate E.C.: Human ß-cell glk

17 Trust the E.C. Nomenclature! 17 EC 1 Oxidoreductases EC 1.1 Acting on the CH-OH group of donors EC With NAD or NADP as acceptor EC With a cytochrome as acceptor EC With oxygen as acceptor EC With a disulfide as acceptor EC With a quinone or similar compound as acceptor EC With other acceptors EC 1.2 Acting on the aldehyde or oxo group of donors EC With NAD or NADP as acceptor EC With a cytochrome as acceptor EC With oxygen as acceptor EC With a disulfide as acceptor EC With an iron-sulfur protein acceptor EC With other acceptors EC 1.3 Acting on the CH-CH group of donors EC With NAD or NADP as acceptor EC With a cytochrome as acceptor EC With oxygen as acceptor EC With a quinone or related compound as acceptor EC With an iron-sulfur protein as acceptor Not widely available for other types of gene products (T.C. numbers are being developed) Kudos to enzymologists Make sure to balance elements when writing reaction

18 KEGG 18

19 How are these reconstructions 19 Reaction Entries different than KEGG? Not elementally or charge balanced Network Maps Don t show which compounds participate in the reactions Compound Entries Don t include details on ionization state Don t show localization info

20 Charge Determination on 20 Metabolite at neutral ph Identify compound & look up in KEGG Determine and identify ionizable group Determine acid and base forms Determine pka values based on the identifiable group (in the Table) If pka > ph, acid form dominant If pka < ph, base form dominant

21 21 Ionizable Groups 1 GROUP ACID BASE + H + pka Terminal Carboxyl -COOH COO - + H + ~4 Primary (Secondary, Tertiary) Amine -NH 3 + NH 2 + H + > 9 Thiol -SH S - + H + ~8.5 O - Phenol + H + ~10

22 22 Ionizable Groups 2 GROUP ACID BASE + H + pka Primary Alcohol -CH 2 OH CH 2 O - + H + ~15 Acetamide (Amide) + H + ~0 OH + Urea (Carbamide) OH + + H + ~1 Guanido Group H -N-C-NH 2 H -N-C-NH 2 + H + ~12 NH 2 + NH

23 23 Ionizable Groups 3 GROUP ACID BASE + H + pka Pyridine H+ + H + ~5 Pyrole + H + ~ -1 H 2 + Imidazole H+ + H + ~7 Pyrimidine H+ + H + ~0

24 24 Ionizable Groups 4 GROUP ACID BASE + H + pka Pyrrolidine (Like Sec Amine) H+ + H + ~10 Aniline H+ + H + ~5 -O Benzoic Acid + H + ~4

25 Ionizable Groups 5: Purines & Pyrimidines 25 Adenine Guanine Thymine Uracil Cytosine Net Charge = 0!

26 26 Example : Arginosuccinate Example: Argininosuccinate Primary Amine pka ~ 9 Guanido pka ~ 12 Neutral MF: C10H18N4O pka: 1.62, 2.70, 4.26, 9.58, >12 Net Charge: -1 Charged MF: C10H17N4O6

27 Biochemical Data: 27 Curation and Expansion of the Network H. pylori Glycolysis according to KEGG: Glucose G-6-P F-6-P FDP H. pylori Glycolysis according to Hoffman et al. (1996): Glucose G-6-P F-6-P FDP

28 28 Organism-specific Textbooks Great starting point Broad view of the organism s metabolism, biochemistry, physiology, uses, etc.

29 29 III. Physiological Information and Inferred Reactions: Filling in the Gaps based on indirect evidence

30 Filling in the Gaps an Example 30 Experiments determine which amino acids are taken up by H. pylori vs. which can be produced in vivo Missing steps of amino acid biosynthesis are added if necessary on the basis of this physiological evidence Amino Acid Requirements AA Reynolds Model Ala - - Arg - - Asn + + Asp + + Cys + + Gln + + Glu + + Gly + + His - - Ile - - Leu - - Lys + + Met - - Phe - - Pro + + Ser + + Thr + + Trp + + Tyr + + Val - - in vivo in silico

31 31 Inferred Reactions Some reactions are included based on indirect physiological evidence (by inference) Assumption: the cell must be able to produce all biomass components to grow Reactions are added if necessary Generally transporters, etc. Most tentative; should be examined more carefully

32 Reaction Confidence: Sources of 32 Evidence Increasing Confidence Biochemical Enzyme has been tested biochemically. Genetic Gene overexpression and purification, gene deletions. Sequence There is significant sequence similarity to another gene with known function. Physiological There is physiological data to support inclusion in the model. Modeling Reaction is included to improve simulation results. Kinetic Assay Overexpression 86% Homology Grows on Ascorbate

33 33 Gene to Reaction Connections

34 Escherichia coli Metabolism 34

35 From Genes to Reactions 35 Not all genes have a one-to-one relationship with their corresponding enzymes or reactions Many genes, one reaction: frdabcd Four subunits combine to form fumarate reductase enzyme, catalyzing FUM + FADH 2 SUCC + FAD One gene, many reactions: tkta One gene encodes transketolase I enzyme, catalyzing R5P + X5P T3P1 + S7P E4P + X5P T3P1 + F6P E. coli frdabcd

36 Integrating -omics Data 36 ORF gene Genomics ORF annotation Transcriptomics mrna levels protein reactions Proteomics protein levels Fluxomics flux measurements

37 Example of Isozymes 37 fructose-1,6-bisphosphate aldolase

38 A More Complex Example 38 Pyruvate Metabolism

39 39 Enolase 1 gene 1 reaction (342/904 Genes are one-to-one) Succinate Dehydrogenase 4 subunits 2 reactions (135/904 Genes interact as Subunits) Xylose ABC Transporter Protein complex (3 proteins) 1 reaction (153/904 Genes interact as PCs) Pyruvate Kinase 2 isozymes 1 reaction (149/931 Reactions have multiple isozymes)

40 Special Considerations 40

41 Biomass Composition 41 Indicates demands of the system (more detail in modeling section of class) Precursors may also be used for smaller networks Approximation of Biomass composition for less-characterized organisms (H. pylori, H. influenzae) Metabolite NADPH G6P F6P R5P E4P GA3P 3PG PEP PYR ACCOA OXA AKG SUCCOA Demand (mmol) ATP 41.3 NAD (trace)

42 42 Generally: Nutrient uptake Growth rate

43 43 Charge Considerations An underappreciated aspect of building reaction networks electrical charge should be conserved in all reactions Phosphofructokinase (from KEGG): ATP + F6P => ADP + FDP charges = -6 = -7 = -6 + H + + 1

44 44 Finding Compound Charges Consult diagram and look at each chemical group independently Determine if H s are attached or dissociated at cellular ph (attached if pka<ph) can find pka values F6P: Both H s dissociate at cellular ph, leaving a charge of -2

45 Balancing Reactions: An 45 Solution # 1: Example pyruvate decarboxylase pyr acald + co 2 C3H3O3 C2H4O CO2 ELEMENTALLY IMBALANCED (H and O) (-1) (0) (0) CHARGE IMBALANCED Solution # 2: h 2 o + pyr acald + co 2 + ½ o 2 + h + H2O C3H3O3 C2H4O CO2 O2 H ELEMENTALLY BALANCED (0) (-1) (0) (0) (0) (+1) CHARGE IMBALANCED Solution # 3: h + + pyr acald + co 2 H C3H3O3 C2H4O CO2 ELEMENTALLY BALANCED (+1) (-1) (0) (0) CHARGE BALANCED

46 Compartmentalization 46 extracellular space cytosol endoplasmic reticulum nucleus peroxisome mitochondria vacuole Golgi apparatus There are 8 compartments included in our yeast model May need to infer transport reactions between compartments H+, ATP, NADH, NADPH must be balanced within each compartment

47 47 Protein Localization Compartmentalization is a key part of network reconstruction Both a component (static localization) and state (dynamic localization) data type Techniques typically based on GFP-tagging of proteins or isolation of specific organelles Potential problems: Effect of the GFP tag on localization Usually human assistance is required in image analysis Condition dependence: e.g. mitochondrial localization agrees only in 30% of cases in yeast between three different data sets nucleus nucl. per. ER bud neck mitochon. lip. ves Huh et al. Nature 425:686 (2003)

48 48 Mitochondrial Isolation: Protein Identification 657 distinct proteins 498 (81%) functionally classified into 15 cellular processes. 153 unique enzymatic activities Glycolysis, TCA cycle, oxidative phosphorylation, urea cycle, fatty acid oxidation, lipid and heme biosynthesis. Taylor S. et al, Nature Biotech (21), 2003

49 Metabolic Databases 49

50 50 Genome-Scale Reconstructions