·406·生物化学与生物物理进展2006:33 6)Prog. Biochem . Biophys.resolution magic-angle spinning (HRMAS)Both require substantial in provem ents to obtain richNMRmolecular information(as discussed earlier)Om ethodsremovethe linebroadening factorsresultingascertaintheidentitiesandpropertiesofbiomarkersfrom residualdipole-dipoleinteraction,chemical shiftFor M S-based m ethods, equipment can also befairlyanisotropyandsusceptibility,achievinghighresohuitionexpensive (e.g.,FTMS).Itisimpossibleto obtainspectroscopyfortissueswhichiscomparabletothatinthe liquid state NM R B81~38]The intoduction of thiscompartmentationinformationwithchromatographyandMSbasedmethodseventhoughsuchinformationmethodmakesitpossibletodirectlyinvestigatetissuesand even whole living organisn s Bo01 w ithout anyisextrem elyin portant Forboth techniques,althoughMoreover,NMRisaquantification can be done by constructing in situdestructionof samples.standard calibrationcurves,thiswill requireenormousquantitativeand simultaneousdetection method thusamountsofworkbecause,not biasedfor anymolecules.To quantifytheinmanycasesmetabonom es consist of hundreds of metabolites andm etabolites,NMR methods only require a knowntheir standards are often not readilyavailable.Evenconcentrationcompoundoranelectronicsignalasnew biomarkers,bydefaultarereference.TheNMRbasedmethodis also highmoreimportantly,often unknownmetabolites.Someofthedisadvantagesthroughput and400samplespercanmeasureuroin ection technology E0]W ith the richcanbeovercomsomeextentbytheirhvphenation24hwithflowForexam ple, GCM S has shown greatprom ise l1027obtained,suchmethodsinfomationhavelowsinceGCiswellestablishedandreproducibilityisrecurrent expenditure so that the analysis cost isusually low on costper-sam ple basis even thoughreasonably high.H ow ever, therem ainingproblem s arestill associated withthe aspects of quantification,NMR spectrometers are expensive.InfactNMR(biased)selectivedetection,molecularmethod is the onlycurrentmetabonomics detectionpoorinfomation,tin em ethod which can be used to undertake in vivo and inconsumingsam plepreparationprocedures and invasive/destructive nature. Recently,situ studies at the levels of cells, m ulticellular tissueshyphenatedultrah ighperfomanceliquidorgans and w hole organ isn s N M R-based m ethods canUPLC-chrom atographyandmassspectrom etryalsobeemployedtoseparatesignalswithoutseparateM S)28~301 has been develbped,which substantiallysam ples via. spectral editing techniques by takingchrom atographicresolutionimprovedandadvantages of differences inrelaxation propertiesanddiffusivities ofm etabolites3, further favouring thereproducibility,show ingpotentialingreatmetabonomic applications.Itmay be possible that,spectral sim plification and signal assignm ents.Thewith somemore improvementsin both facilitiesandmapr disadvantage for the conventional NM R is itscolumnstationaryphases,UPLCMSwillplayevenlowintrinsicsensitivity.However,therearetwowaysgreaterroles inthefuturemetabonom icsresearch.ThisinproveNMRsensitivity,namely,improvingtotechniqueand GC-M Swill probablybecomethem ainsignato-noise ratio by reducing electronic noise andtechnology especiallychem ical noiselevels.In order to reduce electronicstreamwhenvivolinsitustudiesstrictlyrequiredandthenoiselevelscryogenic probe technology hasbeenarenotis not critically vitalcom partm entationinfomationintroducedwhich thenum.berofvearsHoweverattentionsneededelectronicdetectioncircuits wasmenttemperaturcomprehensivelydealwiththeproblem sof"knownreducedaboutt20Khenceincreaseddetectionhunknown markers”forbothchromatography-M Ssensitivityabout fourfoldsbyChem ical noisesbased technolbogies.resulting fromsignal overlappingcanbereducedbyIn contrastthe NMRbasedmethod isuserincreasing the m agnetic field strength.Thiswillalsoindependentrequireslittleorno samplepreparationincrease detection sensitivity and spectral resoluition.(e.g.,forbiofluids of mammals),and has excellentCurrentlythestandard600MHzspectrometerresoluitionandreproducibility.Inaddition,NMRequipped with optimised cryogenic probe technologymethod offers rich molecular infomation includingcan detectmetabolites at nanogram (10-g)levelmetabolitestructure,concentration,molecularSincethepolarisation rateinNMR obeysBoltzmanndynam ics,interactions, pH andcompartnentationdistribution law and such rate is extremely low underwhen diffusion-editingtechniques areemployed.nomal circum stances,methods are also currentlyFurthermoreNMRisnon-invasiveandnon-destructiveunderdevelopm entto increase thepolarisationratebyto sam ples, m ak ing it possible to facilitate in vivo andusing dynam ic nuclearpolarisation techniques.It ism situ studies. For exam ple, recently developed highthus conceivablethatfurther developments of this21994-2014 China Academic Journal Electronic Publishing House.All rights reserved.http://www.cnki.net
生物化学与生物物理进展 Prog. Biochem. Biophys. 2006; 33 (5) Both require substantial improvements to obtain rich molecular information (as discussed earlier) to ascertain the identities and properties of biomarkers. For MS-based methods, equipment can also be fairly expensive (e.g., FTMS). It is impossible to obtain compartmentation information with chromatography and MS based methods even though such information is extremely important. For both techniques, although quantification can be done by constructing in situ standard calibration curves, this will require enormous amounts of work because, in many cases, metabonomes consist of hundreds of metabolites and their standards are often not readily available. Even more importantly, new biomarkers, by default, are often unknown metabolites. Some of the disadvantages can be overcome to some extent by their hyphenation. For example, GC-MS has shown great promise [10,26,27] since GC is well established and reproducibility is reasonably high. However, the remaining problems are still associated with the aspects of quantification, selective (biased) detection, poor molecular information, time consuming sample preparation procedures and invasive/destructive nature. Recently, hyphenated ultrahigh performance liquid chromatography and mass spectrometry (UPLCMS)[28～30] has been developed, which substantially improved chromatographic resolution and reproducibility, showing great potential in metabonomic applications. It may be possible that, with some more improvements in both facilities and column stationary phases, UPLC-MS will play even greater roles in the future metabonomics research. This technique and GC-MS will probably become the main stream technology especially when in vivo/ in situ studies are not strictly required and the compartmentation information is not critically vital. However, some urgent attentions are needed to comprehensively deal with the problems of “known unknown markers” for both chromatography-MS based technologies. In contrast, the NMR-based method is user independent, requires little or no sample preparation (e.g., for biofluids of mammals), and has excellent resolution and reproducibility. In addition, NMR method offers rich molecular information including metabolite structure, concentration, molecular dynamics, interactions, pH and compartmentation when diffusion-editing techniques are employed. Furthermore NMR is non-invasive and non-destructive to samples, making it possible to facilitate in vivo and in situ studies. For example, recently developed high resolution magic-angle spinning (HRMAS) NMR methods remove the line broadening factors resulting from residual dipole-dipole interaction, chemical shift anisotropy and susceptibility, achieving high resolution spectroscopy for tissues which is comparable to that in the liquid state NMR[31～38] . The introduction of this method makes it possible to directly investigate tissues and even whole living organisms [39] without any destruction of samples. Moreover, NMR is a quantitative and simultaneous detection method thus not biased for any molecules. To quantify the metabolites, NMR methods only require a known concentration compound or an electronic signal as reference. The NMR-based method is also high throughput and can measure up to 400 samples per 24 h with flow-injection technology[40] . With the rich information obtained, such methods have low recurrent expenditure so that the analysis cost is usually low on cost-per-sample basis even though NMR spectrometers are expensive. In fact, NMR method is the only current metabonomics detection method which can be used to undertake in vivo and in situ studies at the levels of cells, multicellular tissues, organs and whole organisms. NMR-based methods can also be employed to separate signals without separate samples via. spectral editing techniques by taking advantages of differences in relaxation properties and diffusivities of metabolites[41～43] , further favouring the spectral simplification and signal assignments. The major disadvantage for the conventional NMR is its low intrinsic sensitivity. However, there are two ways to improve NMR sensitivity, namely, improving signal-to-noise ratio by reducing electronic noise and chemical noise levels. In order to reduce electronic noise levels, cryogenic probe technology has been introduced for a number of years in which the temperature of the electronic detection circuits was reduced to about 20K, hence increased detection sensitivity by about four folds. Chemical noises resulting from signal overlapping can be reduced by increasing the magnetic field strength. This will also increase detection sensitivity and spectral resolution. Currently the standard 600 MHz spectrometer equipped with optimised cryogenic probe technology can detect metabolites at nanogram (10 - 9 g) level. Since the polarisation rate in NMR obeys Boltzmann distribution law and such rate is extremely low under normal circumstances, methods are also currently under development to increase the polarisation rate by using dynamic nuclear polarisation techniques. It is thus conceivable that further developments of this · 406 ·
唐惠儒等：代谢组学：一个迅速发展的新兴学科2006;336)·407.techniquewillmakethe“low sensitivity"nature ofspecial biofluids such as C SF, m ilk and dialysates canNMRrapidlybecomea history.alsobeimportantBycomparingwiththemetabonomeItisclearthatthebestchoiceatpresentmaylieatofthecontrolsamples,theeffectsofintemalandthe com bination of a number of existing techniquesextemal stimuli on thesub ects can be easily analysedThehyphenation of chromatography,MSandNMRin a holistic and quantitative way using statistical too ls.LC-NMRMS)playedanimportantrolealreadyintheForplantsystems,mostcurrentlyusedmethodsmetabolite identifications.Itisparticularly exciting toare based on solventextractsthus are in vitro bynature. This is ow ing to that the plant “biofluids"note that a statistical heterospectoscopy m ethod hasbeen reported E4l w hich takes advantages of both N M Rare not as readily available as in the case of anin als.and UPLC-M S. It is predictable that such correlationHowever,tissues and sometimes whole organismtechniqueswill becom e increasingly inportantin thestudies are possible by employing high resolutionmagic-angle spinning (HRMAS)NMR.Thismethodfuture.2.2can alsobe applied to m icrobiota and cultured cells, inSamplesformetabonomicsstudiesIn theory,any biological samples includingwhichboth cellsand m ediaare important carriers forbiofluids, tissues and even whole organism s can bem etabonomic information.The analysis of media isalso im portantand often referred to as cellm etabolismused form etabonomics studies.In practice,however,thesamplerequirements aredependent onthepurpose“footprintings”of studies and m ethodsused.Forboth chromatography2.3How todometabonomicsstudies?and M S based m ethods, the sam ples have to be in theThere are three general steps for conductingmetabonom ics studies,namely,samplepreparation,liquidfom andoften requireremovalofproteins orextractions.ForNMR-basedmethods,ontheotherdata acquisition anddatam iningin orderto extractthehand,both biofluids (e.g.urineand blood plasm a)andbiologicalsignificances.However,for sucha com plextissues (e.g., liver,kidney and brainbiopsies)aresubectextra carehasto be takenineachstep.Forgenerally employed directly withoutmuch sampleexample,thesam ple collectionsandpreparation stepispreparation or extractions, so that in vivo and in situfiundamentallymportantManyfactorsaffectinginform ation is readily available.In such cases, only asm etabonomes have to be carefiully thought throughlittle as 2 μlbiofluids5] and about 10 m g tissues aresuch as variations in species, strains, gender, age, dietrequiredto obtain nform ation in themannersofeitheracclimatisation,timingofsamplesandthestrengthofInformation on thestimuliInthecase ofusing extracts,theextractioninvivo,exvivo or in situ.m etabolite com partn entation isalsoreadilyretrievablemethods(completeness of extraction)have to beNMRm ethods [16 ~49]diffusion-editedoptimisedsothatmaximum metabonomic informationusingFurtherm ore, with such a small volumeof biofluidsis available.To acquire data for the metaboliterequired,thetime course ofm etabonom ic changes canone can use chromatography,complementmassalso be m onitored w ith m inim um orno invasion to thespectroscopytoobtain digitalspectrometryNMRsubjects [5] Such studies can also be caried outinfomationthemetabonomecompositionandondirectly on w hole an im alsbo]stimulitheexertedThe choice ofresponsesdetectionFormammalianmodels,themostcommonlyusedmethodshastobeappropriatefortheThebiofluids arebloodplasm a and urine.Blood plasmaofstudies.dataanalysisandpurposescarries the snap shot infomationaboutthegiveninterpretationthe most crucial step,in whichISphysiologicalandpathologicalprocessesandhowwellappropriate jistifications and validations have to be"supervised”the hom eostasis is m aintained;urine sam ples, on thecarefully considered especially whenotherhand,contains information aboutthemetabolitesdata analysis methods are employed.Bjologicalexcreted in aperiod of time,which mayreflectthem eaningfulnessistheultimateaim form etabonomicsstates of physiolbgy, biological age, possibility ofstudieswhich can be clearlyillustrated inthefollowingdiseasesincludingtheinbomandenorsapplications.Tissuesenvironmentallyinducedpathology.3Applicationsofmetabonomicsmetabonome provides a unique opportunity forM etabolite analysis is im portant because anydetecting m olecularinform ation of an organ where theperturbationtoalivingsystem,whetherit isbiochem ical processes occur, hence the underlyingphysiological or pathological in nature,will causemechanism of itsbiochemicalresponsetostimulicanperturbations of concentration orflux of endogenousbe understood.For some specific purposes,other21994-2014 China Academic Journal Electronic Publishing House.All rights reserved.http:/www.cnki.net
2006; 33 (5) 唐惠儒等：代谢组学：一个迅速发展的新兴学科 technique will make the “low sensitivity” nature of NMR rapidly become a history. It is clear that the best choice at present may lie at the combination of a number of existing techniques. The hyphenation of chromatography, MS and NMR (LC-NMR-MS) played an important role already in the metabolite identifications. It is particularly exciting to note that a statistical heterospectroscopy method has been reported[44] which takes advantages of both NMR and UPLC-MS. It is predictable that such correlation techniques will become increasingly important in the future. 2.2 Samples for metabonomics studies In theory, any biological samples including biofluids, tissues and even whole organisms can be used for metabonomics studies. In practice, however, the sample requirements are dependent on the purpose of studies and methods used. For both chromatography and MS based methods, the samples have to be in the liquid form and often require removal of proteins or extractions. For NMR-based methods, on the other hand, both biofluids (e.g. urine and blood plasma) and tissues (e.g., liver, kidney and brain biopsies) are generally employed directly without much sample preparation or extractions, so that in vivo and in situ information is readily available. In such cases, only as little as 2 !l biofluids[45] and about 10 mg tissues are required to obtain information in the manners of either in vivo, ex vivo or in situ. Information on the metabolite compartmentation is also readily retrievable using diffusion-edited NMR methods[46～49] . Furthermore, with such a small volume of biofluids required, the time course of metabonomic changes can also be monitored with minimum or no invasion to the subjects [45] . Such studies can also be carried out directly on whole animals[50] . For mammalian models, the most commonly used biofluids are blood plasma and urine. Blood plasma carries the snap shot information about the given physiological and pathological processes and how well the homeostasis is maintained; urine samples, on the other hand, contains information about the metabolites excreted in a period of time, which may reflect the states of physiology, biological age, possibility of diseases including the inborn errors and environmentally induced pathology. Tissues" metabonome provides a unique opportunity for detecting molecular information of an organ where the biochemical processes occur, hence the underlying mechanism of its biochemical response to stimuli can be understood. For some specific purposes, other special biofluids such as CSF, milk and dialysates can also be important. By comparing with the metabonome of the control samples, the effects of internal and external stimuli on the subjects can be easily analysed in a holistic and quantitative way using statistical tools. For plant systems, most currently used methods are based on solvent extracts thus are in vitro by nature. This is owing to that the plant “biofluids” are not as readily available as in the case of animals. However, tissues and sometimes whole organism studies are possible by employing high resolution magic-angle spinning (HRMAS) NMR. This method can also be applied to microbiota and cultured cells, in which both cells and media are important carriers for metabonomic information. The analysis of media is also important and often referred to as cell metabolism “foot-printings”. 2.3 How to do metabonomics studies? There are three general steps for conducting metabonomics studies, namely, sample preparation, data acquisition and data mining in order to extract the biological significances. However, for such a complex subject, extra care has to be taken in each step. For example, the sample collections and preparation step is fundamentally important. Many factors affecting metabonomes have to be carefully thought through such as variations in species, strains, gender, age, diet, acclimatisation, timing of samples and the strength of stimuli. In the case of using extracts, the extraction methods (completeness of extraction) have to be optimised so that maximum metabonomic information is available. To acquire data for the metabolite complement, one can use chromatography, mass spectrometry or NMR spectroscopy to obtain digital information on the metabonome composition and responses to the exerted stimuli. The choice of detection methods has to be appropriate for the purposes of studies. The data analysis and interpretation is the most crucial step, in which appropriate justifications and validations have to be carefully considered especially when “supervised” data analysis methods are employed. Biological meaningfulness is the ultimate aim for metabonomics studies which can be clearly illustrated in the following applications. 3 Applications of metabonomics Metabolite analysis is important because any perturbation to a living system, whether it is physiological or pathological in nature, will cause perturbations of concentration or flux of endogenous · 407 ·