Sweating the small stuff---the influence of metabolite extraction and separation on metabolomic studies

Transcription

1 Sweating the small stuff---the influence of metabolite extraction and separation on metabolomic studies Andrew D. Patterson, PhD Associate Professor of Molecular Toxicology Penn State University

2 Resources Metabolomics Workbench XCMS Institute Great tutorials on chromatography, platforms, databases Twitter

3 References Mass spectrometry-based metabolomics Xenobiotic metabolomics: major impact on the metabolome The intestinal metabolome: an intersection between microbiota and host

4 "What have we here? A man or a fish? Dead or alive? He smells like a fish; a very ancient and fish-like smell..." W. Shakespeare The Tempest N GENOME FMO3 Mutation FISHY EGGS PROTEOME METABOLOME USE METABOLITE CHANGES TO INFORM ABOUT MECHANISM

5 Metabolomics Metabolomics is the systematic analysis of the unique chemical fingerprints left behindby specific cellular processes These smallmoleculemetaboliteprofiles provideinsightintocellular status. All -omics based scientific disciplines aim at the collective characterization and measurement of their particular constituent molecules A comprehensive approach to study complete pools of biological molecules Defines the structure, function and dynamics of an organism. Vast chemical diversity among small molecule metabolites has made extended coverage of the metabolome challenging Size ( Da) Concentration ( pm mm) Physicochemical properties (diverse log P values) Stereochemistry (distinct biologicalactivity)

6 Metabolite Extraction Currently no analytical technique exists that is capable of in-situ measurementofall classes of cellular metabolites Metabolite extraction therefore becomes a crucial step in any type of metabolomics study Criticalto both targeted andglobal basedprofilingstrategies. Optimized extraction methodologyshould fulfill severalcriteria: Extractthe largest numberof metabolites Unbiased and non-selective - physical or chemical properties of a molecule Non-destructive- no modification of metabolites

7 Separation of Metabolites Mass spectrometry usually requires some form of chromatographic separation Most systemsuse either liquid or gas chromatography CE-MS gaining popularity Fractionation of sample components simplifies the resulting mass spectrawhile ensuringmoreaccuratecompound identification Capacity factor (k) is critical to optimizing resolution Increased resolution allows longer MS dwell times resulting in better signal/noiseratios Inadequate chromatographic separation of metabolites results in: signal suppression ion suppression compromised metabolite quantification reduced metabolitecoverage

8 Metabolite Class Separation Mode Stationary Phase Solvent System Bile Acids - general RPLC Waters BEH C18 acetonitrile/water/formic acid (0.01%) Bile Acids - MCA isomers RPLC Waters BEH C18 acetonitrile/water/formic acid (0.01%) Bile Acids - taurine conjugated resolving power, and dirtying isomers RPLC Restek Raptor biphenyl MeOH/water (ph=4.6) Waters Amide column used for HILIC will rapidly and dramatically clog with salt over time unless flushed extensively at the end of each run with methanol/isopropanol/water. Phenomenex Luna NH2 silica will break down over time at high ph thus reducing column life, reducing the source. Carnitines RPLC Waters BEH C18 acetonitrile/water/formic acid (0.01%) Carnitines RPLC Waters CSH C18 acetonitrile/water/formic acid (0.01%) Coenzyme A thioesters RPLC Waters BEH C18 acetonitrile/water/ammonium acetate (ph=10.0) Coenzyme A thioesters RPLC Waters CSH C18 acetonitrile/water/ammonium acetate (ph=10.0) Thermo Hypercarb is robust, provides excellent separation, but is extremely sticky. Coenzyme A thioesters HILIC Waters BEH Amide acetonitrile/water/ammonium acetate (ph=10.0) Eicosanoids RPLC Waters BEH C18 acetonitrile/water/formic acid (0.01%) Keto-prostaglandins RPLC Waters BEH C18 acetonitrile/water/formic acid (0.01%) Nucleotides HILIC Waters BEH Amide acetonitrile/water/ammonium acetate (ph=9.0) Nucleotides HILIC Waters BEH HILIC acetonitrile/water/ammonium acetate (ph=9.0) Nucleotides - NAD(P)+/NAD(P)H RPLC Thermo - Hypercarb graphite acetonitrile/water/ammonium acetate

9 Hypothesis Extraction and separation of metabolites may influence metabolomic studies as much as the disease process being investigated Rationale Developing optimized protocols for extraction efficiency and chromatographic resolution based on metabolite class and/or characteristics will dramatically improve accuracy and reproducibility of metabolomic data sets.

10 organic phase organic phase HA org HA org K D K D HA aq K a HA aq + H 2 0 H A - aqueous phase aqueous phase

11 Definitions Isomer same chemical formula, different chemical structure Stereoisomer same chemical formula, same order/sequence of bonded atoms, different 3- dimensional orientation Isobar same mass, but different chemical formula

12 Resolution of Bile Acid Metabolites by RPLC using Waters BEH C18 1. TCA 2. GCA 3. TCDCA 4. CA 5. UDCA 6. GCDCA 7. CDCA 8. DCA

13 Resolution of Taurine Conjugated MCA Isomers by RPLC on WATERS BEH C ) Tauro-ω-MCA 2) Tauro-α-MCA 3) Tauro-β-MCA

14 Resolution of Taurine Conjugated MCA Isomers by RPLC on Restek Rapture Biphenyl ) Tauro-α-MCA 2) Tauro-β-MCA 3) Tauro-ω-MCA

15 Resolution of Taurine Conjugated MCA Isomers by RPLC on Restek Ultra AQ C Tauro-ω-MCA 2. Tauro-α-MCA 3. Tauro-β-MCA

16 Tauro-β-Muricholic Acid An Typical Isomer Example in Metabolomics TβMCA TβMCA TβMCA TαMCA TαMCA

17 TβMCA is an Farnesoid X Receptor Antagonist Shp (FXR target gene) induction in hepatocytes Sayin et al Cell Metabolism 2013 Li F et al Nature Communications 2013

18 METABOLOMICS FOR UNDERSTANDING DRUG TOXICITY---ACETAMINOPHEN

19 Bernard Brodie APAP NAPQI Toxicity N-acetyl-p-benzoquinone imine

20 NORMAL MOUSE LIVER NECROTIC MOUSE LIVER (400 mg/kg APAP 6 HOURS)

21 Contained in 100s of products One of the most common pharmaceuticals associated with accidental and intentional poisoning (>7 g per adult per day) APAP overdose serves as a model for drug-induced liver toxicity Excess NAPQI (with reduced glutathione levels) leads to oxidative damage and inflammation leading to hepatocellular death/necrosis

22 Acetaminophen Metabolomics Unlabeled Ring - labeled LC-MS Acetyl - labeled Data Analysis Metabolite Identification

23 Mouse Urinary Proteins (MUPs) Dilute equal volume of mouse urine with an equal volume of 50% methanol MUPs

24 Mouse Urinary Proteins Dilute equal volume of mouse urine with an equal volume of 100% methanol TOTAL REMOVAL OF MUPS = IMPROVED SIGNAL, REDUCED ION SUPPRESSION

25 APAP Metabolism Study #4 Score Scatter Plot PCA model Urine

26 APAP Metabolism Loading Scatter Plot PCA Model Drug Metabolites Depleted with APAP Treatment Enriched with APAP Treatment

27 X-Variable Trend Plot for L-Carnitine (m/z= ) Control Acetaminophen Depleted with APAP Treatment

28 X-Variable Trend Plot for Propionylcarnitine(m/z= ) Control Acetaminophen Depleted with APAP Treatment

29 X-Variable Trend Plot for Acetylcarnitine (m/z= ) Control Acetaminophen Enriched with APAP Treatment

30 X-Variable Trend Plot for Decanoylcarnitine (m/z= ) Control Acetaminophen Enriched with APAP Treatment

31 INFLUENCE OF EXTRACTION PROTOCOL Carnitines and CoAs

32 CONTROL APAP Control APAP

33 Workflow for Acyl-Carnitines liver 2-propanol acidified MeOH:H 2 O ph= propanol basic n-butanol H 2 O saturated MeOH:H 2 O ph=4.0 RPLC BEH C18 CSH C18 ESI-MS-MS

34 Influence of ph on Metabolite Extraction from Mouse Liver CONTROL LIVER ph = 10 ph = 7 ph = 4 Combined

35 Influence of ph on Metabolite Extraction from Mouse Liver CONTROL ph = 10 APAP ph = 7 ph = 10 ph = 4 ph = 7 ph = 4

36 Influence of ph on Metabolite Extraction from Mouse Liver ph = 4/7/10 Pooled ph = 4/7/10 Pooled ph = 10 ph = 4/7 ph = 10 CONTROL ph = 4/7 APAP

37 Extraction Efficiency of L-Carnitine from Mouse Liver C a r n itin e E x tr a c tio n E ffic ie n c y T e s tin g % % p e a k a re a /m g tis s u e % % % 0 n -b u ta n o l a c id ifie d IP A b a s ic IP A M e O H : w a te r(p H 4.0 ) M e O H : w a te r(p H )

38 Matrix Effects and Extraction LIVER SERUM 2.0! !! p e a k a r e a 1.5! ! n-butanol IPA ph 4 IPA ph 8 MeOH:Water ph 4 MeOH:Water ph 8 n -b u ta n o l a c id ifie d IP A b a s ic IP A M e O H :w a te r (p H 4.0 ) M e O H :w a te r (p H ) p e a k a r e a p e a k a r e a 6! 1 0 6! ! 1 0 4! n -b u ta n o l a c id ifie d IP A b a s ic IP A M e O H :w a te r (p H 4.0 ) M e O H :w a te r (p H ) 5.0! ! ! C 0 C 2 C 3 C 3 :1 C 3 -O H C 4 C 4 -O H C 4 :1 C C 0 C 2 C 3 C 3 :1 C 3 -O H C 4 C 4 -O H C 4 :1 C 5 :1 C 6 C 6 -O H C 6 :1 C 8 C 8 :1 C 9 C 1 0 C 1 0 :1 6! ! 8 1! CARNITINES C0 C5 1.5! 6 1!

39 6! b a s ic IP A M e O H :w a te r (p H 4.0 ) M e O H :w a te r (p H ) 1.5! b a s ic IP A M e O H :w a te r (p H 4.0 ) M e O H :w a te r (p H ) 1! p e a k a r e a 4! Matrix Effects and Extraction 1.0! p e a k a r e a 5! ! ! LIVER SERUM 0 8! ! C 0 C 2 C 3 C 3 :1 C 3 -O H C 4 C 4 -O H C 4 :1 n-butanol IPA ph 4 IPA ph 8 MeOH:Water ph 4 MeOH:Water ph ! ! ! C 0 C 2 C 3 C 3 :1 C 3 -O H C 4 C 4 -O H C 4 :1 C 5 C 5 C 5 :1 C 6 C 6 -O H C 6 :1 C 8 C 8 :1 C 9 p e a k a r e a 4! p e a k a r e a 1.0! ! ! ! C 1 0 C 1 0 :1 C 1 0 :2 C 1 2 C 1 2 :1 C 1 2 -D C C 1 4 C 1 4 :1 C 1 4 :1 -O H 0 CC1 01 :2 6 C C1 61 -O 2 H C 1 26 :1 :1 C 16 2:1 -D-O C H C 164 :2 C C1 61 :2 4 :1 -O H C 1 4 C:1 1-O 8 H C :2 :1 C :2 :1-O -OH CARNITINES C10 C14

40 C 0 C 2 C 3 C 3 :1 C 3 -O H C 4 C 4 -O H C 4 : A n -b u ta n o l a c id ifie d IP A b a s ic IP A M eo H :w ater (ph 4.0 ) M eo H :w ater (ph 10.0 ) B p e a k a r e a p e a k a r e a C 5 C 5 :1 C 6 C 6 -O H C 6 :1 C 8 C 8 :1 C C D p e a k a r e a p e a k a r e a C 1 0 C 1 0 :1 C 1 0 :2 C 1 2 C 1 2 :1 C 1 2 -D C C 1 4 C 1 4 :1 C 1 4 :1 -O H 0 C 1 6 C 1 6 -O H C 1 6 :1 C 1 6 :1 -O H C 1 6 :2 C 1 6 :2 -O H C 1 8 C 1 8 :1 C 1 8 :1 -O H

41 Resolution of Acyl Carnitine Standards by RPLC on Waters BEH C L-carnitine 2. Propionylcarnitine 3. Hexanoylcarnitine 4. Octanoylcarnitine 5. Decanoylcarnitine 6. Lauroylcarnitine 7. Myristoylcarnitine 8. Palmitoylcarnitine

42 Acylcarnitine Extraction in Acidified IPA c o n tro l A P A P -tre a te d p e a k a re a /m g tis s u e * LONG CHAIN ACYL CARNITINES * * * * * L -c a r n itin e a c e ty l p r o p io n y l h e x a n o y l o c ta n o y l d e c a n o y l la u r o y l m y r is to y l p a lm ito y l Increasing Chain Length

43 Workflow for Coenzyme A thioesters liver MeOH:H 2 O (80:20) ph=4.0 MeOH:H 2 O (80:20) ph=7.0 MeOH:H 2 O (80:20) ph=10.0 CHCl 3 :MeOH:H 2 O HILIC RPLC amide SiOH BEH C18 CSH C18 MeCN:H20 ph=4.6 MeCN:H2O ph=10.0 ESI-MS-MS

44 Resolution of Coenzyme A (CoA) Thioester Metabolites by RPLC using Waters BEH C Coenzyme A 2. Acetyl CoA 3. 3-dephospho CoA 4. Propionyl CoA 5. 2-butenyl CoA 6. Isobutyrl CoA 7. Butyrl CoA 8. Phenylacetyl CoA 9. Hexanoyl CoA 10. Octanoyl CoA 11. Decanoyl CoA 12. Lauroyl CoA 13. Myristoyl CoA

45

46

47 Chromatographic Overlay for Coenzyme A (CoA) Metabolites 1/ Coenzyme A 2. Acetyl CoA 3. Propionyl CoA 4. Isobutyrl CoA 5. Butyrl CoA 6. Phenylacetyl CoA 7. Hexanoyl CoA 8. Octanoyl CoA 9. Decanoyl CoA 10. Lauroyl CoA 11. Myristoyl CoA 12. Lineleoyl CoA 13. Oleoyl CoA

48 Chromatographic Overlay for Coenzyme A (CoA) Metabolites Coenzyme A 2. Acetyl CoA 3. 3-dephospho CoA 4. Propionyl CoA 5. 2-butenyl CoA 6. IsobutyrlCoA 7. Butyrl CoA 8. Phenylacetyl CoA 9. HexanoylCoA 10. Octanoyl CoA 11. DecanoylCoA 12. LauroylCoA 13. Myristoyl CoA 14. Lineleoyl CoA 15. Oleoyl CoA

49 AcylCoa Extraction via Modified Bligh/Dyer A c y l-c o A A n a ly s is in M e O H E x tra c ts o f A P A P -tre a te d M u rin e L iv e r E x p t # c o n tro l LONG CHAIN ACYL A P A P -tre a te d * p e a k a re a /m g tis s u e COAs * * * * * * * * * 0 C o e n z y m e A 3 -d e p h o s p h o a c e ty l * p r o p io n y l c r o to n y l 2 -b u te n o y l b u ty r l is o b u ty r l m a lo n y l h e x a n o y l Increasing Chain Length m e th y l m a lo n y l o c ta n o y l d e c a n o y l * la u r o y l * m y r is to y l p a lm ito le o y l p a lm ito y l h e p ta d e c a n o y l lin o le o y l o le o y l s te a r o y l

50 APAP NAPQI Liver Toxicity PPARα Serum Acylcarnitines MITOCHONDRIAL DYSFUNCTION

51 PPARα

52 AcylCoA Extraction in Acidified IPA PPARα Long Chain CoA Carnitine c o n tro l A P A P -tre a te d * p e a k a re a /m g tis s u e * * * * * * * * * 0 C o e n z y m e A 3 -d e p h o s p h o a c e ty l * p r o p io n y l c r o to n y l 2 -b u te n o y l b u ty r l is o b u ty r l m a lo n y l h e x a n o y l m e th y l m a lo n y l o c ta n o y l d e c a n o y l * la u r o y l * m y r is to y l p a lm ito le o y l p a lm ito y l h e p ta d e c a n o y l lin o le o y l o le o y l s te a r o y l

53 INFLUENCE OF EXTRACTION PROTOCOL Ceramides

54 Ceramide Physicochemical Properties Ceramides are a family of waxy lipid molecules. Name derived from the latin word: cera = waxy + amide Ceramides are comprised of: sphingosine: 18 carbon unsaturated amino alcohol fatty acid moiety amide bond Ceramides are not water soluble: Very hydrophobic Confined to cellular membranes Participate in lipid raft formation >200 structurally distinct species have been identified in mammalian cells

55 Ceramide General Structure Ceramide (d18:1/16:0) 2-amino-1,3-octadec-4-ene-diol Amino alcohol (sphingoid) backbone Palmitic acid Fatty acyl group Ceramide (d18:1/24:1(15z)) 2-amino-1,3-octadec-4-ene-diol Amino alcohol (sphingoid) backbone 15-tetracosenoic acid Fatty acyl group

56 Ceramide Biochemistry Ceramides are found in high concentration in the membrane of cells. Structural component of the lipid bilayer Bioactive lipid - implicated in a variety of physiological functions including: ü Apoptosis and cell growth arrest ü differentiation and cell senescence ü cell migration and adhesion Ceramides are converted rapidly to more complex sphingolipids: Sphingomyelin Glycosylceramides Little accumulation observed Except for the skin (50% of total lipids can be ceramides)

57 Biosynthesis of Ceramides 1. De novo biosynthesis: Ceramide synthases couple sphinganine + long chain fatty acid to form dihydroceramide. Double bond introduced into position 4 of the sphingoid base ceramide synthases 5 and 6 generate are specific for palmitic acid C16 ceramide ceramide synthases 1 (brain and skeletal muscle) specific for stearic aid (C18:0 & C18:1 ceramides) ceramide synthases 2 specific for very long chain CoA-thioesters (C 20 -C 26 ) (C20:0, C22:0, C24:0, C24:1 etc) ceramide synthases 3 unusual ceramides of skin & testes 2. Catabolism of complex sphingolipids: Sphingomyelinases/phospholipase C breakdown sphingomyelin in animal tissues Many factors can stimulate the hydrolysis of sphingomyelin to produce ceramide: Cytokines :TNF-α, IFN-γ & various interleukins 1,25-dihydroxy-vitamin D 3 endotoxin nerve growth factor ionizing radiation & heat

58 Ceramides and Disease Ceramide metabolites have been implicated in various pathological conditions including: Ø Cancer Ø Diabetes Ø Obesity Ø Inflammation Ø Neurodegeneration Although not understood, the structure of individual ceramides aids in defining their physiological function. Ø Ceramides containing specific fatty acids are generated in response to particular stimuli.

59 LC Method Development: Where to Start? Designing and optimizing an LC method involves choosing appropriate: 1. Separation mechanism: NPC, RPLC, HILIC, size exclusion ion, exchange etc 2. Column chemistry: C2, C4, C8, C18, cyanopropyl, phenyl, biphenyl, amide, SiOH etc 3. Column properties: pore size, particle size & column dimensions 4. Stationary and mobile phase combinations Critical to optimizing the chromatographic efficiency, retention, resolution & selectivity of analytes.

60 Ceramide Scouting Gradients on Waters BEH C18 Steep gradient; large change in organic modifier. Provides initial information about analyte retention & resolution. Aids in expediting the development of an optimized LC method Poor Resolution (R) Long Retention (T r ) Time

61 Fractionation of Ceramide Metabolites on Waters CSH C18 Column UPLC-ESI-QTOF-MS C16:0 2. C18:1 3. C18:0 4. C20:0 5. C24:1 6. C22:0 7. C24:0 Column: 100x2.1mm 1.7u Solvent System: A= MeCN:water (60:40 v/v) B= 2-propanol:MeCN (90:10 v/v) 10mM NH 4 OAc + 0.1% formic acid Ballistic gradient 55 o C

62 G r a d ie n t C o m p a r is o n b a llis tic s ta n d a r d 8 0 % o rg a n ic Ballistic Gradient: Ultra-fast gradient Rapid change in organic modifier Separations performed at high flow rates High linear velocity tim e (m in )

63 Fractionation of Ceramide Metabolites on Waters CSH C18 Column UPLC-ESI-MS-MS QqQ MS Detector 1. C16:0 2. C18:1 3. C18:0 4. C20:0 5. C24:1 6. C22:0 7. C24:0 Column: 100x2.1mm 1.7u Solvent System: A= MeCN:water (60:40 v/v) B= 2-propanol:MeCN (90:10 v/v) 10mM NH 4 OAc + 0.1% formic acid Ballistic gradient 55 o C

64 Ceramide Fragmentation Patterns m/z = 262 or 284

65 C e ra m id e A n a ly s is in C H C l 3 :M e O H E x tra c te d M u rin e S e ru m C o n tro l S e ru m S p ik e d S e ru m p e a k a r e a Very little endogenous ceramide signal Poor recovery of spiked ceramide analytes What went wrong?? Sub-optimal extraction method Solubility issues Matrix effects 0 C 1 6 :0 C 1 8 :0 C 1 8 :1 C 2 0 :0 C 2 2 :0 C 2 4 :0 C 2 4 :1

66 C e ra m id e A n a ly s is in C H C l 3 :M e O H E x tra c te d M u rin e S e ru m S e ru m S a m p le # 1 S e ru m S a m p le # 2 S p ik e d C o n tro l p e a k a r e a C 1 6 :0 C 1 8 :0 C 1 8 :1 C 2 0 :0 C 2 2 :0 C 2 4 :0 C 2 4 :1

67 C e ra m id e A n a ly s is in C H C l 3 :M e O H E x tra c te d M u rin e L iv e r L iv e r S a m p le # 1 L iv e r S a m p le # 2 L iv e r S a m p le # 3 S p ik e d C o n tro l c o r r e c te d p e a k a r e a C 1 6 :0 C 1 8 :0 C 1 8 :1 C 2 0 :0 C 2 2 :0 C 2 4 :0 C 2 4 :1

68

69 Conclusions Ceramides can be effectively resolved using reverse-phase liquid chromatography (RPLC) methodologies: C18 column chemistry sufficient; but particle properties important (BEH vs CSH) Stronger eluotropic series needed; MeOH or MeCN & water no good Higher column temp required to compensate for increased backpressure (solvent viscosity & increased flow rates) Various ceramide metabolites can be detected using multiple UPLC-MS platforms: Global metabolite profiling approach - UPLC-ESI-QTOF-MS Targeted metabolite approach - UPLC-ESI-MS-MRM Poor detection of ceramides from bio-fluids and serum. Low levels of endogenous ceramides? Rapidly converted to more complex sphingolipids? Ineffective extraction method? Matrix effects - Ion suppression?

70 Conclusions Extraction protocols can impact metabolomic data sets considerably Solventsystem composition and ph exhibit the most dramatic effects on metabolite recovery The magnitude ofthese effects depend on metabolite class Some classes ofmetabolites The number of extraction repetitions also plays a role in enhancing metabolite recovery Tradeoff - longer sample prep time Larger sample volumes to process (evaporate)

71 Conclusions Traditional RPLC methods can provide efficient separation of acyl-carnitine, bile acid and CoA thioester mixtures. Advancements in hybrid particle technologies Allowing for extremes in mobile phase ph and temperature manipulate selectivity Complex ligand stationary phase interactions HILIC methods are superior at separating highly polar metabolites. Nucleotides and derivatives Small polar metabolites sugars, organic acids, amino acids, hydrophilic vitamins Advanced column chemistries (amide, aminopropyl, biphenyl, graphite, phenyl-hexyl) and alternative chromatographic methodologies (HILIC) can provide enhanced coverage of the metabolome.

72 Future Plans There s no one perfect extraction or LC method available capable of efficiently resolving all componentsor features in the metabolome Therefore, our goal is to continue to develop optimized extraction and chromatography protocols for variousclasses of liver metabolites

73 Acknowledgments Penn State University Chris Chiaro Philip Smith Jared Correll MRC Julian Griffin Elizabeth Stanley National Cancer Institute Frank J. Gonzalez Kris Krausz Changtao Jiang Fei Li Curt Harris Ewy Mathe Majda Haznadar NIEHS R01 ES022186

74 Metastars