Tài liệu HPLC for Pharmaceutical Scientists 2007 (Part 20) pdf

During the stop-flow mode, the time to acquire anNMR spectrum on each peak was limited to two hours to avoid excess broad-ening of the remaining chromatographic peaks.. Deuterated chlorof

Copyright © 2007 by John Wiley & Sons, Inc.

*This chapter is an update reprinted from the reference 40, reprinted with permission from Elsevier, copyright 2003.

structural problems of mixtures of unknown compounds LC-MS has been one

of the most extensively applied hyphenated techniques for complex mixturesbecause MS is more compatible with HPLC and has higher sensitivity thanNMR [1–3] Recent advances in NMR technology have made NMR morecompatible with HPLC and MS and have enabled LC-NMR and even LC-MS-NMR (or LC-NMR-MS or LC-NMR/MS) to become routine analyticaltools in many laboratories in the pharmaceutical environment The presentchapter provides an overview of the LC-NMR and LC-MS-NMR hyphenatedanalytical techniques with (a) a description of their limitations together withexamples of LC-NMR and LC-MS-NMR to illustrate the data generated bythese hyphenated techniques and (b) extensive references toward the appli-cation in the pharmaceutical industry (drug discovery, drug metabolism, drugimpurities, degradation products, natural products, food analysis, and pharma-ceutical research) This chapter is not meant to imply that LC-MS-NMR willreplace LC-MS, LC-NMR, or NMR techniques for structural elucidation ofcompounds LC-MS-NMR together with LC-MS, LC-NMR, and NMR aretechniques that should be available and applied in appropriate cases based ontheir advantages and limitations

The first part of this section (Section 20.2.1) will provide the reader with torical overview of NMR and with a brief description of the most typicalexperiments used in NMR for the structural elucidation of organic com-pounds The second part of this section (Section 20.2.2) will focus mainly onthe improvements carried out in the NMR as a hyphenated analytical tech-nique for the elucidation of organic compounds and an understanding of theneed to develop LC-NMR for the analysis of complex mixtures

his-20.2.1 Historical Development of NMR

In 1945 NMR signals in condensed phases were detected by the physicistsBloch [4] at Stanford and Purcell [5] at Harvard, who received the first NobelPrize in NMR Work on solids dominated the early years of NMR because ofthe limitations of the instruments and the incomplete development of theory.Work in liquids was confined to relaxation studies A later development wasthe discovery of the chemical shift and the spin–spin coupling constant In 1951the proton spectrum of ethanol with three distinct resonances showed thepotential of NMR for structure elucidation of organic compounds [6] Scalarcoupling provides information on spins that are connected by bonds Spindecoupling or double resonance, which removes the spin–spin splitting by asecond radiofrequency field, was developed to obtain information about thescalar couplings in molecules by simplifying the NMR spectrum [7] Initialmanipulation of the nuclear spin carried out by Hahn [8] was essential forfurther development of experiments such as insensitive nuclei enhanced by

polarization transfer (INEPT) [9], which is the basis of many modern pulsesequence experiments During the 1960s and 1970s the development of super-conducting magnets and computers improved the sensitivity and broadenedthe applications of the NMR spectrometers The Fourier transform (FT) tech-nique was implemented in the instruments by Anderson and Ernst [10] in the1960s, but it took time to become the standard method of acquiring spectra.Another milestone which increased the signal-to-noise (S/N) ratio was the dis-covery of the nuclear Overhauser effect (NOE) by Overhauser [11], whichimproves the S/N in less sensitive nuclei by polarization transfer The three-fold enhancement generally observed for the weak carbon-13 (13C) signals was

a major factor in stimulating research on this important nuclide Several yearslater, the proton–proton Overhauser effect was applied to identify protonsthat are within 5 Å of each other In the 1970s Ernst [12] implemented the idea

of acquiring a two-dimensional (2D) spectrum by applying two separateradiofrequency pulses with different increments between the pulses, and aftertwo Fourier transformations the 2D spectrum was created Two-dimensionalexperiments opened up a new direction for the development of NMR, andErnst obtained the second Nobel Prize in NMR in 1991 2D correlation exper-iments are of special value because they connect signals through bonds Exam-ples of these correlation experiments are correlation spectroscopy (COSY)[12], total correlation spectroscopy (TOCSY) [13], heteronuclear correlationspectroscopy (HETCOR) [14], and variations Other 2D experiments such

as nuclear Overhauser effect spectroscopy (NOESY) [15] and rotating frame Overhauser effect spectroscopy (ROESY) [16] provide information onprotons that are connected through space to establish molecular conforma-tions In 1979 Müller [17] developed a novel 2D experiment that correlates thechemical shift of two spins, one with a strong and the other with weak mag-netic moment Initially the experiment was applied to detect the weak 15Nnuclei in proteins, but was later modified to detect the chemical shift of 13Cnuclei through the detection of the protons attached directly to the carbons[18] The heteronuclear multiple quantum correlation (HMQC) experimentgives the same data as the HETCOR, but with greater sensitivity Heteronu-clear single quantum correlation (HSQC) [19] is another widely used experi-ment that provides the same information as the HMQC and uses twosuccessive INEPT sequences to transfer the polarization from protons to 13C

or 15N Heteronuclear multiple bond correlation (HMBC) [20] experimentgives correlations through long-range couplings, which allows two and three

1H–13C connectivities to be observed for organic compounds In 1981 a 2Dincredible natural abundance double quantum transfer experiment (INADE-QUATE) [21] was developed and defines all the carbon–carbon bonds, thusestablishing the complete carbon skeleton in a single experiment However,due to the low natural abundance of adjacent 13C nuclei, this experiment is notvery practical All of these experiments became available with the develop-ment of computers in the 1980s With the accelerated improvements in elec-tronics, computers, and software in the 1990s, the use of the pulsed field

gradients as part of the pulse sequences was developed [22] and applied toimprove solvent suppression and to decrease the time required to acquire 2Dexperimental data.

This brief historical introduction is intended to give a simplified overview

of some of the critical milestones of NMR mainly in chemical applications,excluding the innovations in the field of proteins, solid state, and magnetic resonance in clinical medicine To find out more details, see the articles written by Emsley and Feeney [23], Shoolery [24], and Freeman [25], and theirincluded references

20.2.2 Historical Development of LC-NMR

As mentioned at the end of the historical development of NMR section, thedevelopment of the pulse field gradients extended the applications of NMR.One of the areas not mentioned is the hyphenated techniques NMR is one ofthe most powerful techniques for elucidating the structure of organic com-pounds Before undertaking NMR analysis of a complex mixture, separation

of the individual components by chromatography is required LC-MS is tinely used to analyze mixtures without prior isolation of its components Inmany cases, however, NMR is needed for an unambiguous identification Eventhough hyphenated LC-NMR has been known since the late 1970s [26–33], ithas not been widely implemented until the last decade [34–40]

rou-The first paper on LC-NMR was published in 1978 [26] using stop-flow toanalyze a mixture of two or three known compounds At that time, the limi-tations in the NMR side—for example, sensitivity, available NMR solvents,software and hardware, and resolution achieved only with sample-spinning—made direct coupling to the HPLC difficult Watanabe and Niki [26] modifiedthe NMR probe to make it more sensitive, introducing a thin-wall teflon tube

of 1.4 mm (inner diameter) and thereby transforming it into a flow-throughstructure The effective length and volume of this probe were about 1 cm and

15µL, respectively Two three-way valves connected this probe to the HPLCdetector This connection needed to be short to minimize broadening of thechromatographic peaks During the stop-flow mode, the time to acquire anNMR spectrum on each peak was limited to two hours to avoid excess broad-ening of the remaining chromatographic peaks The authors also mentionedthat use of tetrachloroethylene or carbon tetrachloride as solvents, along withETH-silica as a normal-phase column, limited the applications for this tech-nique Because solvent suppression techniques were not available at that time,the authors [26] recognized that more development was required in the soft-ware and hardware of the NMR side to include the use of reverse-phasedcolumns and their solvents, which in turn would broaden the range of appli-cations A year later, Bayer et al [27] carried out on-flow and stop-flow exper-iments with a different flow-probe design on standard compounds They usednormal phase columns and carbon tetrachloride as solvent One of their obser-vations was that the resolution of the NMR spectra in the LC-NMR system

was poorer than for the uncoupled NMR system, which made the ment of small coupling constants difficult The first application of on-flow LC-NMR was carried out in 1980 to analyze mixtures of several jet fuel samples[28] Deuterated chloroform and Freon-113 and normal-phase columns werethe common conditions used for LC-NMR [29–33], limiting the application ofthis technique.

measure-The use of reversed-phase columns in LC-NMR complicates the NMRanalysis because of (1) the use of more than one protonated solvent, whichwill very likely interfere with the sample, (2) the change in solvent resonancesduring the course of the chromatographic run when using solvent gradients,and (3) small analyte signals relative to those of the solvent In 1995 Small-combe et al [41] overcame these problems by developing the solvent-suppression technique, which greatly improved the quality of the spectraobtained by on-flow or stop-flow experiments The optimization of the WET(water suppression enhanced through T1 effects) solvent suppression tech-nique generates high-quality spectra and effectively obtains 1D on-flow andstop-flow spectra and 2D spectra for the stop-flow mode, such as WET-TOCSY, WET-COSY, WET-NOESY, and others [41]

During the last few years, more progress has been achieved by ing LC-NMR to MS The LC-NMR-MS or LC-NMR/MS (referred to as LC-MS-NMR in this chapter) has expanded the structure-solving capabilities byobtaining simultaneously MS and NMR data from the same chromatographicpeak There are some compromises that have to be taken into account because

hyphenat-of the differences between MS and NMR, such as sensitivity, solvent ibility, and destructive versus nondestructive technique, discussed below LC-

compat-MS has been used for many years as a preferred analytical technique; however,with the development of electrospray ionization techniques, LC-MS has beenroutinely used for the analysis of complex mixtures in the pharmaceuticalindustry LC-MS-NMR is a combination of LC-MS with electrospray and LC-NMR presented below

20.3.1 Introduction

The decision to use either NMR or LC-NMR for the analysis of mixtures inthe pharmaceutical industry depends on factors related to their chromato-graphic separation and the ability of NMR to elucidate the structure of organiccompounds whether hyphenated or not The major technical considerations ofLC-NMR, discussed below, are NMR sensitivity, NMR and chromatographi-cally compatible solvents, solvent suppression, NMR flow-probe design, andLC-NMR sensitivity or compatibility of the volume of the chromatographicpeak with the volume of the NMR flow cell for better detection Figure 20-1shows the schematic setup of the LC-NMR connected to other devices, such

as radioactivity detector and MS (see Section 20.4)

20.3.1.1 NMR Sensitivity. NMR is a less sensitive technique compared to

MS and hence requires much larger samples for structural analysis MS sis is routinely carried out in the picogram range Modern high-field NMRspectrometers (400 MHz and higher) can detect proton signals from puredemonstration samples well into the nanogram range (MW 300 Da) With thecryoprobes (for Bruker NMR instruments) or cold probes (for Varian NMRinstruments), depending on the NMR vendor currently available, the sensi-tivity of NMR markedly improves The samples in the low nanogram rangecan be detected In the high nanogram range, structural analysis can be carriedout For real-world samples, however, purity problems become more intrusivewith diminishing sample size and can be overwhelming in the submicrogramdomain, even by the interference of the impurities from the deuterated solventused for the NMR studies This places a current practical lower limit for moststructural elucidation by NMR, which is estimated by the writer to be close to

analy-500 nanograms (MW 300 Da)

Although several other important nuclides can be detected by NMR,proton (1H) remains the most widely used because of its high sensitivity, highisotopic natural abundance (99.985%), and ubiquitous presence in organiccompounds Of comparable importance is carbon (13C), 1.108% abundance,which, because of substantial improvements in instrument sensitivity, is nowutilized as routinely as proton Fluorine (19F), 100% abundance, is less usedsince it is present in only about 10% of pharmaceutical compounds Anotherconsequence of the intrinsic low sensitivity of NMR is that virtually all samplesrequire signal averaging to reach an acceptable signal-to-noise level Depend-

Figure 20-1 Schematic setup for the LC-MS-NMR system (Reprinted from reference

40, copyright 2003, with permission from Elsevier.)

ing on sample size and amount of sample for the structural analysis, signalaveraging may range anywhere from several minutes to several days Formetabolites in the 1- to 10-µg range, for example, overnight experiments aregenerally necessary.

20.3.1.2 NMR and Chromatographically Compatible Solvents. LiquidNMR requires the use of deuterated solvents Conventionally the sample isanalyzed as a solution using a 5- or 3-mm NMR tube depending on the NMR

probe, which requires ca 500 or 150µL respectively of deuterated solvents.The increased solvent requirements for LC-NMR make this technique highlyexpensive Deuterium oxide (D2O) is the most readily available, reasonablypriced solvent (over $300/L) The cost of deuterated acetonitrile (CD3CN) isdecreasing and varies depending on the percentage of included D2O, but isstill over $1000/L Deuterated methanol (CD3OD) is even more expensive.Deuterated solvents for normal-phase columns are not readily available, butthose that are readily available have even more prohibitive prices This neces-sitates the use of reversed-phase columns.Another factor to be concerned with

is compatibility of the HPLC gradient-solvent system with the NMR tions An HPLC gradient-solvent system greater than 2–3%/min causes prob-lems in optimizing the magnetic field homogeneity (shimming) due to solventmixing in the flow cell A gradient-solvent system greater than 3%/min maytake days for the mixture to equilibrate in the flow cell before NMR experi-ments can be carried out Recently, with the new technology developments

opera-in solid-phase extraction (SPE) as SPE-NMR and capillary-based HPLC ascapLC-NMR or microflow NMR (see Section 20.3.3), the amount of deuter-ated solvents needed is much less and is in the microliter to milliter range

to pump the analyte of interest to the flow cell for the NMR analysis Thesedevelopments make the hyphenated NMR techniques economically moreaccessible

20.3.1.3 Solvent Suppression. During the LC-NMR run, the solvent signal

in the chromatographic peak is much larger than those of the sample andneeds to be suppressed This applies even with deuterated solvents In the case

of acetonitrile, the two 13C satellite peaks of either the protonated or residualprotonated methyl group for CH3CN or CD3CN also require suppressionbecause they are typically much larger than signals from the sample With theoptimization of the WET solvent suppression technique by Smallcombe et al.[41] in 1995, the quality of spectra generated during LC-NMR has been greatlyimproved and is routine The WET solvent suppression technique is the stan-dard technique for LC-NMR because it has the capability of suppressingseveral solvent lines without minimum baseline distortions, compared withothers such as presaturation or watergate One disadvantage of suppressingthe solvent lines is that any nearby analyte signal will also be suppressed,resulting in loss of structural information With the development of SPE-NMRand capLC-NMR or microflow NMR (see Section 20.3.3), the solvent

suppression is not as dramatic as for conventional LC-NMR improving thequality of the NMR data.

20.3.1.4 NMR Flow-Probe Design. Conventional NMR flow cells have anactive volume of 60µL (i.e., corresponds to the length of the receiver coilaround the flow cell) and a total volume of 120µL This means that NMR willonly “see” 60µL of the chromatographic peak If the flow rate in the HPLC is

1 mL/min, when 4.6-mm columns are used, only 3.6 sec of the chromatographicpeak will be “seen” by NMR Chromatographic peaks are generally muchwider than 4 sec, indicating that less than half of the chromatographic peakwill be detected This is one of the disadvantages of LC-NMR compared withconventional 3-mm NMR probes where the amount of sample “seen” by theNMR receiver coil is independent of the width of the chromatographic peak.Recently, NMR flow cells with an active volume of 10, 30, 60, and 120µL arecommercially available Applications using solid-phase extraction (SPE) asSPE-NMR will be more appropriate for 10- or 30-µL flow cells (see Section20.3.3) Microcoil NMR flow cells for capLC-NMR or microflow NMR have

an active volument of 1.5µL for applications of samples in low concentration(see Section 20.3.3)

20.3.1.5 LC-NMR Sensitivity. Because NMR is a low-sensitivity technique,which requires samples in the order of several micrograms, analytical HPLCcolumns have to be saturated when injecting samples in that range This willaffect the chromatographic resolution and separation since resolution oftendegrades when sample injection is scaled-up to that level Another factor thatcan affect chromatographic performance is the use of deuterated solvents Inmany cases, analytes show broad chromatographic peaks and occasionally dif-ferent retention times when using deuterated solvents due to different polar-ity and hydrogen bonding of deuterated versus nondeuterated solvents Whenthis occurs, more chromatographic development is required in order to obtainreasonable resolution One way to increase the LC-NMR sensitivity is bydecreasing the flow rate to less than 1 mL/min At flow rate lower than

1 mL/min, a greater portion of the chromatographic peak will be “seen” byNMR However, this is only possible if the pump of the LC system is accurate

at rates lower than 1 mL/min In the case of the SPE-NMR, the LC-NMR sitivity can be improved by concentrating the chromatographic peak into theSPE cartridge by injecting the sample several times (see Section 20.3.3) ForcapLC-NMR or microflow NMR, the LC-NMR sensitivity can improved if thesample is concentrated in a volume of 5µL

sen-20.3.2 Modes of Operation for LC-NMR

The HPLC is connected by red polyether ether ketone (PEEK) tubing to theNMR flow cell which is inside the magnet With shielded cryomagnets or ultra-shielded magnets the HPLC can be as close as 30–50 cm to the magnet versus

1.5–2 m for conventional magnets Normally a UV detector is used in theHPLC system to monitor the chromatographic run Radioactivity or fluores-cent detectors can also be used to trigger the chromatographic peak(s) ofinterest.

There are four general modes of operation for LC-NMR: on-flow, stop-flow,time-sliced, and loop collection These modes described below are automated

by software that controls the valves of the HPLC to stop the flow whenneeded, depending on the mode of operation selected for LC-NMR

20.3.2.1 On-Flow. On the on-flow or continuous-flow mode, the matographic run continues without stopping at any point of the run The chro-matographic peaks are flowing through the NMR flow cell while NMR spectraare being acquired In this mode, the NMR experiments require more amount

chro-of sample to analyze “on the fly” because the resident time in the NMR flowcell is very short (3.6 sec at 1 mL/min) during the chromatographic run, whichlimits this approach to 1D NMR spectra acquisition only This mode can beused to analyze the major components of the mixture and, in many cases, torapidly identify the major known compounds of the mixture

20.3.2.2 Stop-Flow. On the stop-flow mode, the chromatographic peak isanalyzed under static conditions The chromatographic peak of interest is sub-mitted directly from the HPLC to the NMR flow cell Stop-flow requires thecalibration of the delay time, which is the time required for the sample to travelfrom the UV detector of the HPLC to the NMR flow cell, which depends inturn on the flow rate and the length of the tubing connecting the HPLC withthe NMR Because the chromatographic run is automatically stopped whenthe chromatographic peak of interest is in the flow cell, the amount of samplerequired for the analysis can be reduced compared to the on-flow mode and2D NMR experiments, such as WET-COSY, WET-TOCSY, and others [41],can be obtained since the sample can remain inside the flow cell for days It ispossible to obtain NMR data on a number of chromatographic peaks in aseries of stops during the chromatographic run without on-column diffusionthat causes loss of resolution, but only if the NMR data for each chromato-graphic peak can be acquired in a short time (30 min or less if more than fourpeaks have to be analyzed, and less than two hours for the analysis of no more than three peaks) The use of commercially available cryoprobes or cold probes improves the sensitivity of the stop-flow mode (see Section20.3.1.1)

For instance, stop-flow is the preferred mode for the analysis of lites when the chromatography is reasonable or the metabolite is unstable.One example is the analysis of the major metabolites of compound I (Figure

metabo-20-2), a ras farnesyl transferase inhibitor in rats and dogs [42] Preliminary

studies by LC-NMR using a linear solvent gradient [5–75% B 0–25 min,75–95% B 25–35 min, A: D2O, B: ACN (acetonitrile), 1 mL/min, 235 nm, BDSHypersil C18 column 15 cm × 4.6 cm, 5 µm] indicated that even with the use of

protonated acetonitrile in the solvent mixture, all the resonances were visible(Figure 20-3) Figures 20-4A and 20-4B are the UV chromatograms from asmall injection of dog bile and dog urine for metabolites M9 (retention time

10 min) and M11 (retention time 21 min), respectively These small injections

Figure 20-2 Structure of compound I, a ras farnesyl transferase inhibitor in rats and

dogs, and proposed structures by MS of its major metabolites in dog bile (M9) and dogand rat urine (M11) (Reprinted from reference 40, copyright 2003, with permissionfrom Elsevier.)

Figure 20-3. 1H NMR spectrum of compound I in stop-flow (Reprinted from ence 40, copyright 2003, with permission from Elsevier.)

refer-were carried out to identify the UV chromatographic peaks of the analytes ofinterest to determine if there were other chromatographic peaks that couldinterfere the NMR studies by stop-flow Metabolite M11 was also found in raturine To analyze the structures of M9 and M11 by NMR, larger injections ofdog bile, dog urine, and rat urine were carried out for the stop-flow experi-ments.The 1H NMR spectrum on the LC-NMR system (Varian Inova 500 MHzequipped with an 1H–13C pulse field gradient indirect detection microflowNMR probe with a 60-µL flow cell, Palo Alto, CA) of M9 (Figure 20-5)revealed the presence of a 1,2,4-trisubstituted aromatic ring in the 3-chlorophenyl ring and the glucuronide moiety Neither of the two possibilitiesfor the position of the glucuronide moiety ring, positions 4 or 6, could be dis-tinguished NOE experiments on the LC-NMR were not successful because

of problems with the solvent suppression The sample was collected and theNOE was performed (Varian Unity 400 MHz, equipped with a 3-mm 1H–13Cpulse field gradient indirect detection Nalorac probe, Palo Alto, CA) over aweekend (Figure 20-6) Even though the collected sample contained moreimpurities, the NOE experiment showed that the glucuronide moiety wasattached at C-4 by irradiating the methylene at i which elicited NOE signalsfrom H-2 and H-6, thus eliminating the C-6 possibility (Figure 20-6) LC-MS

on M11 indicated it to be only the 1-(3-chlorophenyl)piperazinone moietywith an additional oxidation on the piperazinone ring The 1H NMR spectrum

on the LC-NMR system of M11 lacked the isolated methylene signal on thepiperazine ring (Figure 20-7), indicating it to be the (1-(3-chlorophenyl)piper-azine-2,3-dione)

Recently, a radioactive volatile metabolite M3 with a small molecularweight was studied using LC-NMR [43] Conventional NMR was not possiblebecause the radioactivity of the sample was lost when the fraction containingthe metabolite was evaporated to dryness prior to the NMR studies In this

Figure 20-4 UV chromatograms from small injections of the dog bile containing

metabolite M9 (A) and dog urine containing metabolite M11 (B) (Reprinted from erence 40, copyright 2003, with permission from Elsevier.)

ref-912 LC-NMR OVERVIEW AND PHARMACEUTICAL APPLICATIONS

Figure 20-5. 1H NMR spectrum of metabolite M9 from dog bile in stop-flow mode.(Reprinted from reference 40, copyright 2003, with permission from Elsevier.)

Figure 20-6. 1

H NMR (bottom) and 1D NOE spectra at i (top) of M9 from dog bilerecovered from LC-NMR (Reprinted from reference 40, copyright 2003, with permis-sion from Elsevier.)

example, the LC-MS was not informative, suggesting a molecular weight lessthan 200 Da LC-NMR was one of the alternatives used to solve this structuralproblem To be able to identify the UV chromatographic peak corresponding

to the radioactive metabolite, a radioactivity detector equipped with a liquidcell (Radiomatic C150TR, Packard) was connected on-line to the LC-UVsystem of the LC-NMR Figure 20-1 shows the schematic diagram for thissetup Small injections were carried out initially to identify the metabolite UVchromatographic peak with the radioactive peak prior to the stop-flow exper-iments (Figure 20-8) Stop-flow experiments were triggered by UV becausethe transfer delay from the UV to the NMR was shorter than from the radioac-tive detector to the NMR, due to the thicker tubing used in the liquid cell ofthe radioactivity detector.1H NMR spectrum revealed the presence of the p-

fluorophenyl ring with the characteristic splitting pattern, indicating that

the compound was drug-related The downfield shift of the ortho protons at

7.91 ppm suggested the presence of a carbonyl substituent (Figure 20-8) Thepresence of a singlet at 4.85 ppm, integrating for approximately two protons,was consistent with a methylene that was flanked by the carbonyl and ahydroxyl group (Figure 20-8) These features thus led to proposing the struc-

ture for M3 as the p-fluoro-α-hydroxyacetophenone (Figure 20-8).

20.3.2.3 Time-Sliced Mode. The time-sliced mode involves a series of stopsduring the elution of the chromatographic peak of interest.A time-sliced mode

is used when two analytes elute together or with close retention times, or whenthe separation is poor Depending on the NMR vendor, the software can be

Figure 20-7. 1H NMR spectrum of metabolite M11 from dog urine in stop-flow mode.(Reprinted from reference 40, copyright 2003, with permission from Elsevier.)

designed to automate this mode, but sometimes the analyst may prefer to do

it manually

20.3.2.4 Loop Collection. On the loop collection mode, the graphic peaks of interest are automatically stored in loops controlled by thesoftware for later off-line NMR study Then the stored chromatographic peaksare transferred to the NMR flow cell individually for NMR studies The soft-ware is designed to send the stored chromatographic peaks to the NMR flowcell in the same or different order as they were stored from the chromato-graphic run Loop collection can be used when there is more than one chro-matographic peak of interest in the same run In this case the analytes must

chromato-be stable inside the loops during the extended period of analysis Capillarytubing should be used to avoid peak broadening with concomitant loss ofanalyte “seen” by the NMR spectrometer Loop collection can be used in con-nection with SPE for SPE-NMR analysis (see Section 20.3.3)

20.3.3 Other Analytical Separation Techniques Hyphenated with NMR

Recently, other chromatographic techniques have been coupled on-line toNMR for additional applications in the pharmaceutical environment, such as

Figure 20-8 UV-radioactive (C-14) chromatograms of the fraction containing

metabo-lite M3 (top) and expanded sections of the 1H NMR spectrum of metabolite M3acquired for one day (bottom)

size-exclusion chromatography (SEC) as SEC-NMR for the characterization

of polymer additives [44], capillary electrophoresis (CE) as CE-NMR for smallvolume samples [45–47], capillary electrochromatography (CEC) as CEC-NMR, capillary zone electrophoresis (CZE) as CZE-NMR for on-flow iden-tification of metabolites with small volume samples [46, 48–52], andgel-permeation chromatography (GPC) as GPC-NMR and supercritical fluidchromatography (SFC) as SCF-NMR for polymer separation and identifica-tion [53] as examples CE-NMR and CEC-NMR are techniques that work withvery small-volume NMR probes with capillary separations Solid-phase extrac-tion (SPE) as SPE-NMR is becoming a popular technique for trace analysis

In SPE-NMR, the chromatographic peaks are trapped into trap cartridgesusing multiple injections to increase the concentration of the chromatographicpeaks, and then the cartridges are dried with nitrogen to remove all residualsolvents With this technique, deuterated solvents are only used to flush eachpeak from the cartridge to the NMR flow cell, creating a sharp eluting band(25- to 30-µL eluting volume) that requires the use of small NMR flow cells,such as 10- or 30-µL flow cells SPE-NMR allows increasing the sensitivitycompared with regular LC-NMR The recent use of cryogenic flow probe withthe SPE-NMR application improves tremendously the sensitivity of NMR[54] SPE-NMR has been applied for trace analysis [55], microbial metabolites[56], and natural products [54, 57, 58] Lately, more developments have beencarried out to hyphenate capillary-based HPLC (capLC) with NMR as capLC-NMR or microflow NMR and the use of commercial microcoil NMR probes[46, 59–61] With microcoil NMR probes, the range of sample used in capLC-NMR could reach the nanogram level (low nanogram level only for detectionlimit but not for structural analysis) [46, 59–61] With this technique, thevolume of the chromatographic peak is comparable to the volume of themicrocoil NMR flow cell The volume observed for a commercial microcoilNMR flow cell is approximately 1.5µL, and there is a wider range of solventgradient variation than in the standard LC-NMR CapLC-NMR can be usedwithout a column for analysis of low concentrated pure compounds, such as

1µg, or with the column to study mixtures of compounds One of the ments for capLC-NMR is that the sample has to be soluble in a volume ofapproximately 5µL or less, which is not always possible The delay timebetween the UV detector of the cap-LC and the NMR flow cell has to be cal-ibrated for all chromatographic conditions due to the changes of viscosity ofthe different solvent compositions, which has an effect on the pump of the cap-LC More recently, the development of multiple coils connected in paral-lel may be applicable to acquire NMR data of several samples at the sametime [39, 62–64] So far, four samples can be run at the same time, but recentdevelopments are going toward analysis of 96-well plates emulating tech-niques such as LC-MS [39] CapLC-NMR with single or multiple solenoidalmicrocoils can also be used with other capillary techniques such as capillaryelectrophoresis (CE) [63, 64], capillary isotachophoresis [63, 65, 66], and others [63]

require-20.3.4 Applications of LC-NMR

There are many examples in the literature for applications of LC-NMR in thepharmaceutical industry In the area of natural products, LC-NMR has beenapplied to screen plant constituents from crude extracts [54, 57, 67, 68] and toanalyze plant and marine alkaloids [69–72], flavonoids [73], sesquiterpene lac-tones [74, 75], saponins [58, 76], vitamin E homologues [77], and antifungal andbacterial constituents [56, 78, 79] as examples In the field of drug metabolism,LC-NMR has been extensively applied for the identification of metabolites[42, 80–88] and even polar [89] or unstable metabolites [43] And finally, LC-NMR has been used for areas such degradation products [90–93], drug impu-rities [94–102], drug discovery [103, 104], and food analysis [105–107]

20.4 LC-MS-NMR (OR LC-NMR-MS OR LC-NMR/MS)

20.4.1 Introduction

The capability of analyzing a complex mixture in a chromatographic run bythe hyphenation of several techniques, such as NMR and MS, to HPLC isbecoming more popular in the pharmaceutical industry NMR and MS data

on the same analyte are crucial for structural elucidation When different lates such as metabolites are analyzed by NMR and MS, one cannot always

iso-be certain that the NMR and the MS data apply to the same analyte, cially when the analytes have been isolated using analytical columns and prepcolumns for the MS and NMR analysis, respectively HPLC conditions are notalways reproducible when analytical and prep-HPLC columns are used toisolate different amounts of the analytes of interest To avoid this ambiguity,LC-MS and LC-NMR are combined MS data should be obtained initiallybecause with NMR, data collection in the stop-flow mode can take hours ordays, depending on the complexity of the structure and the amount of sample.This is why it is preferable to designate this operation as LC-MS-NMR ratherthan LC-NMR-MS or LC-NMR/MS

espe-Since MS is considerably more sensitive than NMR, a splitter is rated after the HPLC to direct the sample to the MS and NMR units sepa-rately In the example below, the MS used in these studies is a classic LCQinstrument (ThermoFinnigan, CA) A custom-made splitter was used with

incorpo-a splitting rincorpo-atio of 1/100 (Acurincorpo-ateTM, LC Packings, CA) It was designed todeliver 1% of the sample initially to the MS and the balance 20 seconds later

to the NMR With a flow rate of 1 mL/min, the final flow rate going to theNMR will be 0.990 mL/min, and the final flow rate going to the MS will be0.010 mL/min Electrospray is the only source of ionization that will work withsuch low flow rate (10µL/min) in LCQ Figure 20-1 depicts the scheme of theLC-MS-NMR system used in the example for this chapter The technical con-siderations of LC-MS-NMR are the same as LC-NMR (see Section 20.3) plusthe effect of using deuterated solvents for the MS of the LC-MS-NMR

For the last 4–5 years, the LC-NMR-MS system has been commerciallyavailable only for the Bruker NMR instruments For the Varian NMR instru-ments, the system has recently become available The work presented here hasbeen carried out by the author using a custom design of the LC-MS-NMRsystem on a Varian NMR instrument as explained above.

20.4.1.1 The Use of Deuterated Solvents. Another consideration for theLC-MS-NMR is the use of deuterated solvents needed for NMR Analyteswith exchangeable or “active” hydrogens can exchange (i.e., equilibrate) withdeuterium (2H) at different rates The analyst should be alert to this possibil-ity because it could result in the appearance of several closely spaced molec-ular ions with pseudo-molecular ions increased, depending on the number

of exchangeable hydrogens being deuterated If the compound of interestexchanges all the active hydrogens for deuteriums, the pseudo-molecular mol-ecular ion will be [M +2H]+or [M −2H]−in positive or negative mode, respec-tively, where M is the molecular weight with all the exchangeable hydrogensdeuterated When buffers or other compatible solvents for MS are needed, it

is recommendable to use deuterated buffers to avoid the suppression of tional solvent lines in the NMR spectra (see Section 20.3.1.3)

addi-20.4.2 Modes of Operation for LC-MS-NMR

As mentioned in the section of modes of operation for LC-NMR (Section20.3.2), with the use of shielded cryomagnets, the location of the MS instru-ment will follow the same rule as for the HPLC The most common modes ofoperation for LC-MS-NMR are on-flow and stop-flow With stop-flow, the MSinstrument can also be used to stop the flow on the chromatographic peak ofinterest that is to be analyzed by NMR These two modes are presented herewith an example In the loop collection mode, the MS of the LC-MS-NMRsystem may also monitor the trapping of the chromatographic peak inside the loop

In the last few years, there have been relatively few examples in the ture dealing with the application of LC-MS-NMR in the pharmaceutical indus-try.The author of this chapter has been interested in evaluating this technology

litera-to determine the pros and cons and litera-to decide which cases are suitable for this

application To illustrate these modes of operation, a group of flavonoids waschosen Eight flavonoids were selected to mimic a real complex mixture ofcompounds of similar structure that may present some ambiguity in theiranalysis that can be resolved by this hyphenated technique versus the indi-vidual nonhyphenated techniques Figure 20-9 shows the eight flavonoids(Aldrich) chosen for this example These compounds have simple structurescomposed primarily of aromatic protons; some have low-field aliphaticprotons which would not be hidden under the NMR solvent peaks Phenolicprotons exchange rapidly with D2O, so that each compound will only show one pseudo-molecular ion Flavonoids are natural products with important

biological functions acting as antioxidants, free radical scavengers, and metalchelators and are important to the food industry.

The chromatographic conditions are as follows: 35–50% B 0–10 min,50–80% B 10–15 min; A, D2O; B, ACN; 1 mL/min, 287 nm, Discovery C18column 15 cm × 4.6 cm, 5 µm Stock solutions of each compound were prepared

at 1µg/µL in ACN : MeOH 1 : 1

A Varian Unity Inova 600-MHz NMR instrument (Palo Alto, CA) equippedwith a 1H{13C/15N} pulse field gradient triple resonance microflow NMR probe(flow cell 60µL; 3 mm O.D.) was used Reversed-phase HPLC of the sampleswas carried out on a Varian modular HPLC system (a 9012 pump and a 9065photodiode array UV detector) The Varian HPLC software was also equippedwith the capability for programmable stop-flow experiments based on UVpeak detection An LCQ classic MS instrument, mentioned in the previoussection, was connected on-line to the HPLC-UV system of the LC-NMR bycontact closure The 2H resonance of the D2O was used for field-frequencylock, and the spectra were centered on the ACN methyl resonance Suppres-sion of resonances from HOD and methyl of ACN and its two 13C satelliteswas accomplished using a train of four selective WET pulses, each followed

by a Bogradient pulse and a composite 90-degree read pulse [41]

20.4.2.1 On-Flow. The on-flow experiment was carried out on a mixture ofeight flavonoids (Figure 20-9) (20µg each) MS and NMR data were obtainedduring this on-flow experiment The UV chromatogram is depicted in Figure20-10 Table 20-1 and Figures 20-12A–D show the pseudo-molecular ion infor-mation [M − 2H]−, where M is the molecular weight with all the hydroxyl

Figure 20-9 Structures of eight flavonoids used for the LC-MS-NMR technology

development studies (Reprinted from reference 40, copyright 2003, with permissionfrom Elsevier.)