11 NMR Spectroscopy

7/31/2019 11 NMR Spectroscopy

1/19

11 NMR Spectroscopy

By B. A. SALVATORE

Department of Chemistry and Biochemistry, University of South Carolina, Columbia,

SC 29208, USA

1 Introduction

It appears that most recent breakthroughs in NMR are driven by research involving

biological macromolecules. The fact that many of these developments are occurring in

neighboring fields does not preclude their application within the realm of organic

chemistry. A survey of the recent literature reveals that many new NMR techniques are

indeed important to chemists who study relatively small organic molecules. Thus, it isinstructive to distill the recent literature from a cross-section of disciplines and

pinpoint those NMR techniques which are of interest to the organic community. It is

in that spirit that this review focuses on recent advances in NMR from a variety of

areas. The material pertains to both isotropic and oriented NMR sample systems, and

the author has adopted a selective (rather than comprehensive) approach in presenting

some of the most significant developments.

2 Monitoring chemical reactions by NMR

The utility of NMR in monitoring the progress of chemical reactions is continually

being enhanced, with recent advances demonstrating its analytical power in both

solution and solid-phase chemistry. This includes the development and refinement of

new techniques and of existing hardware.

Solution-phase chemistry

NMR is particularly well-suited for detecting and identifying intermediates in chemi-

cal reactions. Organocuprate chemistry is one area in which mechanistic details

remain obscure. A recent study of several cuprates helped clarify the dichotomous

reactivity of these agents, which often participate in both electron-transfer and conju-

gate addition processes. In the reaction between Me

CuLi

and trimethyl ethylene-

tricarboxylate, the cuprate was found to be a particularly strong reducing agent, with

the reduction product predominating over the conjugate addition adduct by more

than 4 : 1. Standard H-decoupled C spectra of Me

CuLi

and trimethyl ethylene-

361


2/19

tricarboxylate [C-enriched at C(2)], in ether at 0 C, revealed a single carbon

resonance, representing the intermediate which follows single electron-transfer from

the cuprate. However, no resonances from an intermediate along the conjugate addi-

tion pathway were detectable by NMR, for reasons that remain unclear. This is

puzzling, in light of the fact that the conjugate addition adduct still accounts for nearly20% of the total product yield, and thus further investigations are warranted.

Chemically Induced Dynamic Nuclear Polarization (CIDNP) is a powerful NMR-

based method for probing the molecular structure of species in which a photochemi-

cally generated free-radical electron spin polarizes a nuclear spin on the same atom.

Using this technique, Giese et al. provided the first spectroscopic proof for the

existence of a radical cation intermediate in a chemical reaction that models the

C,O-bond scission process of 4-DNA radicals. Such radical species are believed to be

important intermediates in the cleavage of DNA by bleomycin and enediyne-based

natural products.NMR is also a powerful tool in assessing the formation of organometallic com-

plexes. Particularly important, are new metallo-NMR methods for characterizing the

structure of ionic species in solution. Koga et al. described the Li and N analysis of a

labeled chiral bidentate lithium amide. LiN couplings established the basis for

studying these complexes in solution.Note that Li is quadrupolar (nuclear spin: 1).

Such experiments should help in explaining the dependence of enantioselectivity on

solvent composition in proton transfer of asymmetric carbon. These investigators

concluded that an observed drop in enantioselectivity in certain solvents was caused

by formation of a dimer of the chiral lithium amide species.Other atom-pairs have also recently been investigated by NMR. Gunther and

Bohler reported the first 2D heteronuclear shift correlation experiment for the spin

pair, LiSi, which they used to study the structure of a silyl-substituted or-

ganolithium anion. The experiments were performed unlocked, on a doubly-tuned

probe, with proton decoupling. The retuned deuterium lock coil served as the Li

channel and the X-coil was used to pulse Si nuclei (natural abundance:5%). One-

and three-bond correlations were visible in the 2D HSQC spectrum, which displayed

two sets ofLiSi scalar couplings. Such experiments are very powerful, because they

can establish the site of chelation within an ionic complex.New hardware developments have also broadened the applications of NMR in

monitoring chemical reactions. Baumann et al. described a simple apparatus which is

particularly applicable (but not limited) to kinetic and mechanistic studies of reactions

between dissolved gases and organometallic complexes. Such studies make it possible

to monitor the appearance and disappearance of intermediates that ordinarily cannot

be isolated. Several reactions have been studied with this apparatus, including an

unusual one, in which two molecules of ethylene add to a zirconocenealkyne complex,

displacing the alkyne.

Other researchers have sought to ally NMR with liquid chromatographic instru-

mentation. LC-NMR consists of linking an HPLC in series with a specially-develop-

ed NMR probe, capable of detecting flow-through samples. A temporary pause in the

flow, as the compound moves through the probe, allows the sample to remain within

the NMR coil long enough to obtain adequate signal averaging. Other implementa-

tions of this technique have also been introduced. For example, elimination of the

HPLC column and introduction of an autoinjector establishes the basis for another

362 B. A. Salvatore


3/19

new analytical system known as Versatile Automated Sample Transport (VAST)

NMR, which promises to be a very powerful tool in the analysis of combinatorial

libraries.

Applications in solid-phase synthesisOne long-standing problem in solid-phase synthesis (SPS) involves monitoring the

progress of reactions during synthesis. FT-IR is often employed to accomplish this, but

it lacks the analytical power of H and C NMR. The primary obstacle to NMR

analysis of SPS is local discontinuities in magnetic susceptibility that result from the

heterogeneous nature of resin-bound samples. This heterogeneity causes variations in

the effective magnetic field around the attached molecules, resulting in chemical shift

dispersions (i.e. broad lines!). As a result, the solid-phase synthetic chemist often must

cleave some product from the resin after each step, which is used for solution NMR or

mass spectrometric analyses. This is time-consuming and wastes synthetic intermedi-ates.

Fortunately, this is changing with the development of fast, non-destructive NMR

methods for characterization of the products of SPS. C NMR of solvent-swelled

resin samples (gel-phase NMR) is suitable in certain situations. Solvating the resin

sample as much as possible before acquiring spectra increases motional freedom of the

resin-bound molecules, thereby reducing linewidths. This is a convenient adaptation of

conventional solution NMR methods, requiring no specialized equipment. Yet, even

though this method often provides carbon NMR spectra suitable for analysis of the

products of chemical syntheses, the problem of broad lines remains and renders themethod unsuitable for high-resolution H NMR studies.

Magic-angle-spinning (MAS) of solvent-swelled resin beads drastically reduces line

widths by averaging out magnetic susceptibility differences within the sample, thus

improving resolution and making it suitable for high-resolution H NMR applica-

tions.This technique is useful in monitoring the progress of SPS, and efforts are being

made to optimize its utility. It was found that poly(ethylene glycol) (PEG)-tethered

resins, commonly called TentaGel, give the narrowest NMR line-widths. Spin echo

experiments and spin-locking are commonly employed to attenuate unwanted poly-

mer peaks and enhance resolution obtained with other resins. However, one disadvan-

tage of spin echoes is that J-couplings are lost as the echo refocuses. If it is important to

retain the J-coupling information, it can be recovered through 2D J-resolved spectros-

copy. Shapiro et al. reported a useful experiment, in which an untilted 2D J-resolved

spectrum is projected along a single axis.This technique capitalizes on the benefits of

the spin echo experiment by attenuating unwanted polymer resin peaks, while retain-

ing proton scalar coupling information along the chemical shift axis of a standard

one-dimensional H-spectrum.

MAS NMR is an excellent tool for analyzing the products from peptide SPS. Opella

et al. have shown that it is now possible to determine the three-dimensional structural

characteristics of resin-bound molecules. They acquired 2D NOESY spectra and

established the conformation of a resin-bound hexapeptide. Such information is

relevant to drug design, because binding assays are sometimes performed on resin-

bound libraries of ligands.

Lippens et al. employed similar techniques to investigate the structural basis for the

difficulties encountered during the SPS of certain peptide sequences. Specifically,

363NMR Spectroscopy


4/19

Fig. 1 A schematic representation of the H nano-probe. The plug for the nanoprobe

cell and the RF leads have been omitted for clarity.

(Reproduced with permission from J. Magn. Reson., Series A, 1996, 119, 65

they used NOE and chemical shift data to establish a correlation between coupling

difficulties and the degree of interchain aggregation, as the synthesis progressed.

Advances have been made in hardware development, as well. Keifer et al. inves-

tigated magic-angle-spinning with new, high-resolution probes (Fig. 1) that were

optimized for very small sample volumes (ca. 40l). The excellent resolution ob-

tained with these nano-NMR probes demonstrates the important benefits of mini-

mizing magnetic susceptibility discontinuities in probe design, as well as in the sample.

Shapiro et al. have adapted MAS NMR methods to analyze products on multipincrowns, thus extending its utility in parallel combinatorial synthesis. Application of

MAS NMR in the solid-phase synthesis of oligosaccharides and other non-peptide-

based combinatorial libraries has also been documented.

Supramolecular chemistry

NMR is a powerful tool for probing non-covalent molecular assemblies, along with

the dynamic exchange processes that occur in those assemblies. Lehn et al. have used

Ag-NMR spectroscopy to monitor the formation of a rectangular supramolecular

grid, assembled from the combination of tritopic ligand 1, ditopic ligand 2, and silver

triflate in a 2: 3: 6 stoichiometric ratio. The Ag NMR spectrum displayed two

signals in a 2: 1 ratio, corresponding to four peripheral and two central silver ions,

respectively. This provides evidence for formation of the complex 3 in which the signal

from the peripheral silver ions is shifted downfield by 59 ppm from the resonance

generated by the central silver ions. Thus, it was consistent with the formation of a

2;3 rectangular grid, via mixed-ligand recognition.

364 B. A. Salvatore


5/19

N

N

N

N

N

N

Me

Me

N

N

N

N

Me

Me

1 2

NN NN

MeMe

N

N

N

N

N

N

Me

Me

N

N

N

N

N

N

Me

Me

NN NN

MeMe

NN NN

MeMe

3

+6

Another elegant example of NMRs utility in supramolecular chemistry was re-

ported in an investigation of the reversible dimerization and guest exchange in

C

-symmetric calixarenes. Off-resonance rotating frame NMR experiments provide

a useful way of discriminating between exchange and the direct cross-relaxation

transfers that are commonly witnessed in standard NOE experiments. Off-resonance

spectroscopy also eliminates complications from TOCSY transfers that sometimes

arise during on-resonance rotating frame experiments. Spin locking was performed in

these experiments, such that the angle between the effective field and the external

magnetic field (:35.6) was in between those used in standard NOE experiments

(: 0) and on-resonance rotating frame experiments (: 90). That value was

365NMR Spectroscopy


6/19

chosen because it resulted in a cancellation of the cross-relaxation contributions from

the laboratory frame and the on-resonance rotating frame for the observed protons.

Thus, only exchange cross-peaks were detected.

Determination of acid dissociation constants by NMRIt is often a challenge to accurately measure dissociation constants of organic acids in

non-aqueous media. Pehk et al. have devised a general approach for measuring

relative acid dissociation constraints by C NMR. This technique is based on the

measurement of frequency differences at varying degrees of protonation between a

known reference compound and a compound whose dissociation constant is un-

known. The degree of protonation of the reference compound (e.g. acetic acid) is

known exactly at each stage of the process. The ratio of dissociation constants for the

acid under study and the reference compound can be determined from the following

relationship in eqn. (1),

K/K: (9

)(9)/(

9)(9

) (1)

where represents the chemical shift for the partly protonated reference acid, while

and

represent the chemical shifts of its fully protonated and fully deprotonated

species, respectively. The corresponding chemical shift values for the acid under

investigation are represented by ,

, and , accordingly. In the most convenient

implementation of this technique, one measures the chemical shift for the reference

acid and for the compound under investigation and then plots the difference 9

against the degree of protonation (n), according to eqn. (2),

9 :9

9 n(

9

);nK/K(

9

)/[1;n[(K/K)9 1]] (2)

where K/K represents the ratio of dissociation constants between the reference acid

and the acid under study. The derivation of this equation (based on the law of mass-

action and two expressions relating NMR chemical shift to the distribution of molecu-

lar populations during the exchange process) was presented by the authors. Plotting

the experimental data for 9 vs. n, one obtains bell-shaped plots which are fit to

obtain the sought quantity K/K, according to eqn. (2). In principle, the dissociation

constant of a particular acid can be determined by this technique when it differs fromthat of the reference acid by as little as 4 J mol\. This technique has minimal

requirements for sample purity, and it can be carried out without any determination of

pH. Additional methods are available for treating the experimental data which sim-

plify the data interpretation when the chemical shift differences between the fully

protonated and deprotonated forms of the compounds are not equal. The authors

illustrate several practical examples, involving branched carboxylic acids.

3 Defining structure and conformation through NMR

Correlation spectroscopy

Rychnovsky et al. reported a strategy for assigning the configuration of 1,3-skipped

polyols. This method was demonstrated by analysis of a pair of polyacetonide

acetate derivatives of the natural product dermostatin A, which are frame-shifted with

respect to each other (4 and 5, Fig. 2). Thus, this strategy is particularly well-suited for

366 B. A. Salvatore


7/19

O

O O

O

OOOOOOAcO

O

O OAc

O

OOOOOOO

4

5

15

15

16

16

17

17

19

1921

21

23

23

25

25

27

27

29

29

31

31

Fig. 2 Two isomeric (frame shifted) polyacetonide acetate derivatives (4 and 5) of

dermostatin A

polyols which contain an odd number of hydroxy groups. The technique relies on acombination of HH COSY, HC DQF-HSQC, and HH ROESY experi-

ments. After assigning all the protons from COSY spectra, HSQC and ROESY spectra

are acquired to determine which of the acetonides in 4 are syn and which are anti (Fig.

3). The relative stereochemistry between each pair of acetonides in 4 is then established

by analyzing the HSQC and ROESY spectra from an isomeric frame-shifted aceton-

ide (5, Fig. 2). A small level of C-enrichment within the acetonides (ca. 10%) is

optimal. This is a very powerful NMR-based method for assigning the relative

stereochemistry within polyol chains. Mosher ester analysis can then be used to

subsequently establish the absolute stereochemistry.Gervay et al. have explored long-range CH NMR connectivity in carbohy-

drates.They applied a 1D Inverse Detected Single Quantum Long Range (INSQLR)

experiment to establish the existence of a highly-labile sialic acid lactone moiety.

Selective excitation at the C-labeled carbonyl in one of the possible lactones resulted

in magnetization transfer through the lactone oxygen, which was detected at a proton,

three bonds away. This signified the presence of the lactone. Had that lactone been

absent, this correlation would not have been observed, since it would have required

CH magnetization transfer through seven -bonds.

One of the major problems in three-dimensional structure determination of

oligosaccharides by solution NMR results from the limited number of distance and

angular restraints, which are generally defined from HH NOEs, and three-bond

HC coupling constant measurements. To alleviate this problem, Homans et al.

have enhanced existing NMR methods for deriving information from exchangeable

protons (i.e. -OH, -NH protons). This technique was demonstrated on N-acetyllac-

tose with three experiments (TOCSY-HSQC, ROESY-HSQC, and NOESY-HSQC).

367NMR Spectroscopy


8/19

Fig. 3 ROESY (left) and HMQC (right) of the acetonide methyl group region for

compound 4. There are three syn- and two anti-acetonides [two of the axial methyl

groups in distinct syn acetonides coincidentally have the same HSQC chemical shifts

(H: 1.44 ppm and C: 20 ppm)]. No ROE correlations appear with the equatorial

methyl groups in the syn acetonides.

(Reproduced with permission from J. Org. Chem., 1997, 62, 2925)

The samples were prepared in an H

O[H

]acetone mixture, which was cooled

(917 C) to minimize proton exchange. The investigators employed 3D C-editing

techniques to overcome chemical shift overlap of non-exchangeable protons, a com-

mon problem in most proton-detected 2D NMR experiments involving carbohy-

drates. Water suppression in these experiments was performed without pulsed-field

gradients (vide infra). These experiments produced a vast number of additional dis-

tance restraints which were useful in conformational analysis of the disaccharide.

Symmetry often poses an obstacle in conformational studies with NMR, since

NOEs and ROEs normally cannot be detected between chemically-equivalent proto-ns. Thus, fewer distance constraints are available for analysis. Wagner and Berger

reported an effective solution to this problem, building upon prior work, in which

HMQC-ROE transfers were performed between two chemically equivalent protons on

a C/C atom-pair. Unwanted signals resulting from ROEs between protons on

C-atoms were successfully suppressed with pulsed field gradients. These authors also

reported an improved 1D version of this experiment, based on the selective excitation

of a single C resonance, bearing one member of a chemically-equivalent pair of

protons (the other proton being on a C-atom within the same molecule). Enclosing

the selective 180 pulses within a gradient sandwich facilitates calibration of the pulses.

A negatively-phased signal appears in the center of each CH doublet (representing

two equivalent protons), where an NOE exists (Fig. 4). Jimenez-Barbero et al. have

reported a similar 1D experiment for extracting NOE (ROE) information from chemi-

cally equivalent anomeric protons in the C

symmetric disaccharide, trehalose.Their

technique is based on the selective inversion of one anomeric C resonance with a

DANTE-Z pulse train, during which time the proton magnetization is spin-locked.

368 B. A. Salvatore


9/19

Fig. 4 1D HSQC-NOESY spectrum of phenanthrene, obtained through a selectivepulse on C(4). A negative peak in the centre of the HC splitting for H(4) is apparent

above. This represents an NOE between the two chemically-equivalent H(4) protons.

(Spectrum reproduced with permission from Magn. Reson. Chem., 1997, 35, 199)

Deriving information from scalar coupling constants

It is possible to garner a tremendous amount of structural and conformational

information through the accurate measurement of scalar (J) coupling constants. In a

recent review, Thomas reminds the chemical community not to lose sight of theimportance of coupling constants in conformational analysis. He attributes recent

neglect, in part, to the explosive development of new multidimensional NMR experi-

ments, which have relegated J-couplings to an uninteresting role. Investigators who

simply view J-couplings as the basis for multidimensional correlations, rather than a

direct source of information, as well, may be ignoring potentially valuable data.

Conformational analysis, based on scalar couplings, has, in fact, been undergoing a

renaissance recently, with many investigators gathering information about HC

and CC J-values. Serianni et al. have developed synthetic and spectroscopic

methods for measuring the complete set of one-, two- and three-bond HC and

CC scalar couplings in --ribofuranose and 2-deoxy--ribofuranose rings.

Since all the furanosyl rings within DNA (or RNA) have the same chemical structure,

comparisons between related J-couplings from the sugars within discrete segments can

yield important information about the topological structure of nucleic acids.

This same group has devised an empirical method for predicting the magnitude and

sign of two-bond CC scalar coupling constants (J!!

) in aldopyranosyl rings.

369NMR Spectroscopy


10/19

Fig. 5 Seriannis projection rule for determining J!!

. To determine J!!

for this

carbohydrate, one must first inspect the angles made by each oxygen substituent on C1

and C2 with the projection anti to the C(2)C(3) bond [viewed along C(1)C(2)] and

then inspect the angles made by each oxygen substituent on C2 and C3 with the

projection anti to the C(1)C(2) bond [viewed along C(2)C(3)]. Summing the cosines

of all of these angles provides a resultant (90.5), which predicts a small negative

coupling constant (ca. 92 Hz). The measured value is 92.4Hz

The values of such couplings have been shown to depend on the orientation of

electronegative substituents relative to the CC bond. Seriannis projection rule for

estimating J!!

is based on an inspection of the angle that each electronegative

substituent on each of the two CC bonds makes with a projection anti to the other

CC bond (Fig. 5). Then, the cosines of all these angles are simply added together. A

small positive sum (:1.0) or a negative sum is indicative of a negative J!!

, while a

larger positive value (91.0) predicts a positive J

!!

. This report includes data for

several different sugars and demonstrates that ab initio calculations ofJ!!

in modelcompounds agreed with the authors predictions. This empirical rule also applies to

J!!

couplings through oxygen (i.e. COC), and this should prove particularly useful

in the conformational analysis ofO-glycosidic linkages in oligosaccharides.

In structural studies of complex natural products, it is often desirable to detect

long-range HC scalar couplings. The standard 2D HMBC experiment is one of the

most powerful methods for accomplishing this. Yet, it is often hampered by low

sensitivity for some long-range correlations, due to the difficulty in setting a univer-

sally-optimal delay time after the first 90 C-pulse. As a compromise, a 5060ms

delay is generally prescribed, but this is often not optimal, particularly for couplings

involving fast-decaying signals (e.g. methylene signals). Seto et al. have proposed a way

to overcome this commonly encountered problem, which entails displaying a 2D

projection from a three-dimensional HMBC experiment. The delay time following

the first carbon pulse in the conventional 2D experiment becomes variable and is

increased uniformly in 4 ms increments from 20 ms up to 80 ms, thus establishing the

third dimension. The results of this 3D experiment are viewed as a projection of the

370 B. A. Salvatore


11/19

sum of sixteen separate 2D spectra onto the f2,f3 plane. By employing gradient-based

coherence selection, requiring only 4 scans per t

point, this experiment takes no

longer than a standard 2D HMBC experiment, performed with 64 scans per t

point.

In a comparison, run with the natural product, monazomycin, the results of a 3D

HMBC were far better than those derived from a standard 2D HMBC. This experi-ment is particularly useful for detecting long-range J-couplings, involving protons that

are broadened in complex spin systems.

Fructose exhibits a complex mutarotational equilibrium between five isomeric

forms, including the pyranose, furanose, and open (straight chain) forms. In an effort to

develop receptors that will selectively bind to one isomeric form in solution, Eggert

and Norrild characterized the various boronic acid complexes of fructose on the basis

of one-bond CC coupling constants (J!!

). Their work is based on the import-

ant observation that exceptionally low J!!

(3540 Hz) result when the OCCO

fragment in a vicinal diol is incorporated into a five-membered ring (as in vicinal cyclicboronic esters). It is believed that this effect results from a change in the orientation of

the oxygen lone pairs with respect to the CC bond. Since this species is selectively

bound (albeit, in DMSO), as its 2,3: 4,5-bis(p-tosylboronate) ester, the authors believe

that they may be able to design a bis-boronic acid-based receptor that selectively binds

to the --fructopyranose anomer in water.

Structureactivity relationships by NMR (SAR by NMR)

Individuals engaged in drug discovery today have the choice of pursuing rational

design methods or concentrating on the many recently-developed combinatorialapproaches. The latter techniques have rapidly gained popularity, because the rational

approach to drug design continues to be hampered with difficulties, such as those

involved in predicting the enthalpic and entropic effects of ligand binding to drug

targets. These factors are key in determining the stability of most drug-protein com-

plexes. For example, water molecules are often released upon ligand binding or,

alternatively, they move to fill gaps at the binding site in unpredictable ways. Problems

of this sort make life difficult for those pursuing a strictly rational approach to drug

discovery. However, combinatorial chemists also encounter difficulties, due to the

limited sensitivity of most biological assays, which generally facilitate the identification

of only the most active compounds in a given library. Weaker binding is often

obscured by background signals that result from high ligand concentrations. Thus, it is

likely that many key lead compounds are missed in high throughput combinatorial

assays. While the debate continues between those espousing combinatorial methods

for drug discovery and those who remain firmly entrenched with rational drug design,

Fesik et al. devised a strategy which blends these two strategies into a very powerful

new approach. This is the so-called, StructureActivity Relationship by NMR (SAR

by NMR) approach to drug discovery. This technique, which has been previously

reviewed, is very powerful, yet elegant in its simplicity. It blends the advantages of

rational drug design and combinatorial chemistry with NMR studies ofN-enriched

proteins.

The HN pairs in folded proteins generally yield well-resolved HSQC spectra.

Addition of a substrate with a moderate affinity for the protein results in a shift of the

HSQC NMR signals for all the HN atom pairs within the binding site. The chief

advantage of SAR by NMR is that it allows one to obtain reliable SAR for compounds

371NMR Spectroscopy


12/19

N

OO

OCH3

OCH3

H3CO

OCH3

O

N

O

H

OH

HO

N

OO

OCH3

OCH3

H3CO

O

O

N

O

H

OH

O

Kd = 2 M Kd = 100 M

Kd = 19 nM

Fig. 6 Two fairly weak-binding FKBP ligands discovered through SAR by NMR

screening (top). These two ligands bind to two distinct sites on FKBP. Linkage of thetwo ligands produced a much stronger-binding ligand (in box)

which bind to the target with low affinity (millimolar range). In their seminal paper on

this technique, Fesik et al. designed a ligand which binds to FKBP in the nanomolar

range by first identifying two ligands that bind to distinct sites on FKBP with

moderately weak binding constants (Fig. 6). A key feature of this technique is the use

ofHN HSQC spectra to detect the binding of small ligands, and to differentiate

between multiple binding sites on the protein surface. Due to N-spectral editing, no

signal from the ligands is observed in the spectra, just changes in the proton and

nitrogen resonances for protein HN atom pairs within the binding site(s). NOEs

between the ligands and specific HN pairs on the protein can also be used to assess

the manner in which the ligands bind to the protein. Fesik et al. acquired HSQC data

with help from an automated sample-changer. Selection of the strongest binding

ligands was then accomplished by considering a weighted average of the chemical shift

changes for H and N, upon addition of each ligand. The overall strategy involves

372 B. A. Salvatore


13/19

five steps: (1) screening of binding of first ligand by NMR, (2) optimization of binding

of the first ligand, (3) screening of binding of second ligand by NMR, (4) optimization

of binding of the second ligand, (5) linking the first and second ligands. When two or

more ligands which bind to a protein at distinct sites are linked together, one obtains a

total free energy of binding based on the sum of the two individual binding energies,plus an additional negative free-energy term associated with the entropy decrease that

results from linking the two ligands together [eqn. (3)].

G:G;G

;G

(3)

A similar approach was adopted to identify powerful inhibitors of stromelysin, a

zinc-dependent matrix metalloproteinase. Fesik et al. also applied SAR by NMR to

find initial leads for inhibitors of the E2 protein from the human papilloma virus,

which binds DNA at a single site. In this case, rather than physically linking the two

independently-binding ligands together, SAR from two distinct series of weakly-binding lead compounds were gathered and merged in the design of a single lead that

displayed an IC

in the micromolar range.

The two main disadvantages of SAR by NMR are that it requires a minimum of

200 mg ofN-labeled protein, overall, and the size of the protein should not exceed

the 30 kDa limit imposed by solution NMR. However, those problems are largely

overcome with relaxation- and diffusion-edited NMR screening techniques. These

new methods are complementary to the HSQC-based technique, since spectra of the

ligands (instead of the protein) are observed. The spectra acquired by these methods do

not allow characterization of the protein binding site. However, they minimize theamount of protein required, eliminate the need for isotopically labelled protein and

they are amenable to binding studies with very large proteins.

Advances in water suppression

For NMR structural studies involving water-soluble compounds (e.g. carbohydrates,

peptides) from which information about exchangeable protons is gathered, H

O is

often a necessary component of the sample, as well as a significant source of problems

during spectral acquisition. For studies in which the sample concentration is in the m

range, the water proton concentration may be up to five orders of magnitude higher.

This large difference in concentrations complicates the acquisition process. Most

problematic is the fact that the huge water resonance makes it impossible to set the

receiver gain at a level suitable for analyzing the sample. Secondly, the water resonance

often overlaps with some of the sample resonances. For these reasons, the development

of efficient water suppression methods continues to be an active research area.

Today, the most popular solvent suppression techniques involve selective excitation

of the water resonance. This is particularly useful in (but not limited to) protein NMR

applications, since protein resonances have shorter relaxation times than water and

thus return to equilibrium much faster. Radiation damping can be a serious problem.

This phenomenon results when transverse water magnetization induces a current in

the receiver coil. The induced current results in a magnetic field that causes the water

proton magnetization to return to equilibrium at a rate much faster than that pre-

scribed by its true T

, thus ruining the solvent suppression. For this reason, a lot of

research has been done to develop gradient-enhanced, frequency-selective water sup-

pression techniques. The general utility of pulsed-field gradients in NMR has been

373NMR Spectroscopy


14/19

recently reviewed by Canet. Gradient-based solvent suppression methods are gen-

erally classified into two categories: (1) frequency-selective excitation followed by

dephasing (i.e. spoiling) with gradient pulses, and (2) frequency-selective refocusing

flanked by gradient pulses which dephase unwanted transverse magnetization.

A recent straightforward application of the first category is the WANTED (water-selective DANTE using gradients) sequence. Here, radiation damping is suppressed

during a water-selective pulse train by keeping the transverse water magnetization

defocused during the period in between each DANTE pulse. In homonuclear 2D

NMR, however, frequency selective refocusing of the solvent resonance is more

common. This is usually achieved by inserting a WATERGATE sequence near the end

of a pulse sequence, prior to acquisition. However, splicing this segment within the

middle of a pulse sequence often complicates the phase cycling and timing within the

entire sequence. Thus, each experiment must be developed and optimized indepen-

dently. For example, Ni et al. recently reported a new gradient-enhanced WATER-GATE-TOCSY experiment in which pulsed-field gradients were used to maintain

precise control of the water magnetization vector. This experiment demonstrated

marked improvements over a standard z-filtered TOCSY, which used water presatura-

tion, but one should not underestimate the effort needed to develop and optimize

similar experiments of this nature.

Several more sophisticated methods for solvent suppression have been developed

which, like WATERGATE, are based on the frequency-selective refocusing of the

water resonance, where the refocusing RF pulses are flanked by gradients that dephase

unwanted transverse magnetization. These are the so-called, excitation sculpting andMEGA techniques and they are less sensitive to flip-angle errors than WATER-

GATE. This alleviates the phase problems commonly encountered when water mag-

netization is spin-locked. In these experiments, the water magnetization is fully re-

turned to equilibrium prior to each acquisition. This improves water suppression,

alleviates attenuation of the sample signal from saturation, and eliminates radiation

damping.

As an aside, it is instructive to point out that excitation sculpting has other useful

applications besides solvent suppression. Frenkiel et al. have extended it to selective

excitation of other (i.e. non-solvent) resonances, and it is particularly applicable inidentifying correlations between specific protons in small molecules. Such tech-

niques often allow one to get the same amount of information from a 1D experiment

that would otherwise require 2D NMR.

Depending on the sample, however, simpler water-suppression techniques are some-

times adequate or even more desirable. As previously mentioned, Homans et al.

opted not to use a gradient-based water suppression technique. This was due to the

small difference in chemical shift between water and the anomeric protons in the

selected carbohydrates. This made it virtually impossible to selectively suppress the

water resonance without also suppressing the anomeric protons in the disaccharide.

Thus, these investigators opted for a technique which used long water-selective purge

pulses, after all the disaccharide proton magnetization had been temporarily transfer-

red to C.

Other solvent suppression techniques sometimes have distinct advantages over

pulsed-field gradient-based methods. The water-PRESS sequence is deployed just

before the main part of a pulse sequence. In this technique, a -RF pulse inverts all of

374 B. A. Salvatore


15/19

the magnetization (i.e. sample and water resonances), and then any transverse magnet-

ization is removed with a homospoil. During a delay, both the water and protein

resonances relax along the z-axis. However, since the T

for water is inherently much

longer than the T

for the sample, the delay is chosen such that the water magnetiz-

ation is approximately zero when the protein resonances are almost all fully relaxed.The homospoil succeeds in dephasing spurious amounts of water magnetization in the

transverse plane, thus preventing the onset of damping during the delay. A subsequent

read pulse is then applied to observe the sample magnetization, while flipping the

water magnetization back to the 9z-axis. Since the water-PRESS suppression module

is appended in front of the pulse sequence, and not spliced within it, the main part of

the experiment begins with the sample magnetization at thermal equilibrium. This

eliminates the phase errors often seen with gradient-based techniques, particularly

when some of the pulse-widths or gradients are not calibrated properly. This technique

is useful because it facilitates the observation of sample resonances that lie under-neath the water resonance, something which cannot be achieved with selective-

inversion methods, like WATERGATE. In contrast to most other methods, the

Water-PRESS technique is extremely simple to implement and optimize. It does not

require accurately calibrated RF pulses, nor excellent lineshape. Moreover, there is no

loss of sample intensity from diffusional effects, which are especially problematic for

small molecules. One disadvantage of the Water-PRESS method is the length of time it

adds to each acquisition, but this may be outweighed by some of the above advantages.

Using computers in structure determinationTraditional automated-NMR-resonance-assignment strategies are based on grouping

resonances into spin systems that represent distinct components of a molecule (e.g.

amino acid residues within a protein). This is followed by the identification of these

segments and the sequential connection of the spin systems, ultimately allowing

assignment of all the NMR spectral resonances. Advances in double isotopic labeling

(C, N) techniques have greatly facilitated this process for proteins. Now, the

efficiency is being further increased by computers. Montelione et al. have developed a

computer program for assigning NMR resonances in proteins, called AUTOAS-

SIGN. It requires the amino acid sequence and input data generated from HN

HSQC, as well as data from the eight most common 3D triple-resonance NMR

experiments [H(CA)(CO)NH, CA(CO)NH, CBCA(CO)NH, HNCO, H(CA)NH,

CANH, CBCANH, and HN(CA)CO]. The program employs five sequential stages of

analysis. Depending on the stage, different methods and criteria are used to designate

chemical shifts and establish sequential links between individual spin systems. Each

stage uses constraint-based matching, which progressively relaxes the criteria used in

designating chemical shifts and sequential links along the protein carbon backbone.

Prestegard et al. have pursued an alternative approach which employs a neural

network to make connections between input data and output structural assignments

from N-edited TOCSY-HSQC spectral data. This approach is more flexible, in that

probabilities are evaluated at each stage, resulting in several choices, instead of just one

definitive choice. One disadvantage of a neural network approach, however, is the

need for a large number of correctly assigned examples to train the network. This can

ultimately be offset by the small amount of data required to make the actual assign-

ments.

375NMR Spectroscopy


16/19

Emerenciano et al. have applied a similar heuristic approach to the structure

determination of natural products by C NMR. This has resulted in a collection of

modular programs, based on the main program, called SISTEMAT. This program

operates on the premise that virtually all natural products are divisible into three

components: the skeletal atoms, the heteroatoms directly bound to these atoms, andthe carbon side-chains. These researchers recently reported a subroutine which is

suited for the identification of common side-chains (e.g. angelate, tiglate, etc.), attached

to any of the atoms in a natural product. This module identifies subspectra of the

carbon atoms representing specific substituent groups, amid the raw C NMR

spectral data. Thus, side-chain peaks are identified and distinguished from the skeletal

carbons, whose values, in turn, can be fed to SISTEMAT to identify the carbon

skeleton. SISTEMAT, itself, was recently upgraded to identify aromatic molecules,

and it was trained with a library of over 700 flavinoids, which represent 72 distinct

skeletal types. Given a set of C chemical shifts for an unknown flavinoid, theprogram was able to suggest a list of probable carbon skeletons, eliminating 70 of the

72 possible carbon skeletons in one demonstration.

Another useful application of computers in the structure identification of natural

products (albeit one that does not utilize AI methods) involves the simulation of

complex NMR spectra, based on molecular mechanics-derived structural data. Laa-

tikainen et al. have developed one such spectral-simulation program, called

PERCH. They demonstrated this programs usefulness by performing a complete

H NMR spectral analysis of-pinene, which possesses a highly complex spin system,

containing 16 coupled protons.

Oriented-sample NMR techniques

It is ultimately desirable to study the properties of a natural molecule in an environ-

ment that resembles its native environment. Unfortunately, this is more easily said

than done. Natural oligosaccharides, for example, are often fixed on the outer phos-

pholipid bilayer of cells, and yet conformational information is still generally derived

from these molecules through solution NMR experiments, in which they tumble

isotropically. However, new NMR methods for studying such molecules in membrane-

like environments have been developed over recent years. Isotropic micelles offer one

option for studying these molecules by NMR, and magic angle spinning of multi-

lamellar liposomes has facilitated the acquisition of some high-resolution NMR

spectra. Most appealing, though, are techniques which not only allow one to study

such molecules at membrane surfaces, but to also extract information about structure

and conformation which is not available from solution or MAS NMR experiments.

One such technique is liquid crystal NMR, where molecules are studied in a field-

orientable liquid crystalline matrix that facilitates the measurement of dipolar coup-

lings. These through-space couplings do not exist in isotropic or MAS samples,

because they average to zero as the molecules tumble randomly or spin at the magic

angle. Liquid crystal samples are often prepared by adding the compound of interest to

a concentrated aqueous lipid-micelle solution. Dimyristoyl phosphatidylcholine

(DMPC) and dihexanoyl phosphatidylcholine (DHPC), (or alternatively DMPC and

the bile salt, CHAPSO) are commonly used to form orientable micelle solutions. The

resulting bilayer micelles (i.e. bicelles) are planar and thus possess an anisotropic

magnetic susceptibility which causes them to orient in a magnetic field, such that the

376 B. A. Salvatore


17/19

HO

O

NH

O

OH

O

OHNH

OHO

HOHOO

O

HO

O

NH

OH

O

OHNH

OHO

HOHO

O

O

Obilayer normal

Bo

Fig. 7 Incorporation of GM4-lactam glycolipid molecules into a DMPCCHAPSO

bicelle. Note that the bicelles bilayer normal aligns perpendicular to the static magnet-

ic field of the spectrometer (B

)

normal to the bicelle surface, on average, lies perpendicular to the external magnetic

field (Fig. 7).In liquid crystal bicelle systems, unlike solid crystals, dipolar couplings are partially

averaged (scaled down) by the motion of the bicelles, and are hence termed, residual

dipolar couplings. The size of a particular residual dipolar coupling D

is defined by

eqn. (4),

D:

9h

2rS

S

3cos 9 1

2 (4)where

and

are the gyromagnetic ratios of the two nuclei, r is the distance between

the nuclei,

is the angle that the internuclear vector makes with the external magneticfield, and the two S terms are order parameters whose product is sample-dependent

and can be measured experimentally. The angular quantity (

) is the key term on the

right side of the equation, where the brackets denote the average value of the enclosed

quantity on the NMR timescale. The presence of this angular term indicates that a

relationship exists between D

and

. Thus one can infer that a correlation also exists

between D

and a molecules orientation/conformation in an oriented system. This is

the basis for liquid crystal NMR. The data are generally analyzed in terms of an order

matrix, using NMR dipolar (or quadrupolar) coupling data along with structural

models from which distance information (r) is derived.

Prestegard et al. employed oriented planar bilayer micelles (i.e. bicelles) to determine

oligosaccharide headgroup orientation at membrane surfaces. Recent studies explored

the conformations of GM4-lactam glycolipid (Fig. 7), a ganglioside analog with

potential applications in the development of cancer vaccines, as well as sul-

foquinovosyldiacylglycerol, a glycolipid with strong inhibitory activity against HIV-

1. These investigations were based on measurements of HC, CC, and

377NMR Spectroscopy


18/19

HN residual dipolar coupling measurements, as well as site-specific C- and

N-chemical shift anisotropy measurements. The dipolar coupling values were de-

rived from labeled samples, mainly via direct detection of C spectra (proton-de-

coupled 1D CC INADEQUATE and 2D CC DQF-COSY), and molecular

mechanics-minimized structures were used to obtain distance (r) values.Vold and Prosser reported a modified bicelle system for which the diamagnetic

anisotropy, and hence the sense of bicelle orientation, is flipped. Thus the bilayer

normal axis aligns parallel to the magnetic field (S

9 0). This was achieved by

doping a DMPCDHPC solution with lanthanide salts (e.g. EuCl

), and it resulted in

a two-fold increase in the order parameter (S

), as determined by quadrupolar

splitting measurements in oriented-chain perdeuterated DMPC. This modified bicelle

system enhances the resolution of dipolar couplings in NMR spectra and makes bicelle

systems more applicable to the study of large, slow tumbling molecules, like proteins.

Bax and Tjandra found that even greater resolution spectra are attainable, evenwithout metals, by simply diluting the bicelle concentration (down to 3% w/v). This

substantial decrease in bicelle concentration does not disrupt orientation, but rather,

results in much higher resolution spectra which facilitate the measurement ofHN,

HN, and HC dipolar couplings. Moreover, this bicelle system is amenable to

indirect detection, which provides far greater sensitivity than the direct C- and

N-detection methods employed previously. In such dilute bicelle systems, it is

unlikely that the sample molecules are incorporated into the individual bicelles, but

they are still influenced, and thus oriented, by the cooperative alignment effects within

the bicelle matrix. These results expand the power of oriented-bicelles, making themmore applicable to small molecules which are not necessarily membrane-anchored.

The same investigators also reported a new pulse sequence for determining the residual

dipolar contributions to CH splittings within methinyl groups, based on the quanti-

tative measurement of peak intensities in HC HSQC spectra. They accom-

plished this with a modified constant time HSQC experiment, in which the proton

pulse is fixed in at the center of the constant time evolution period, and only the carbon

pulse is stepped through this period.

4 Miscellaneous

On a final note, one paper which did not conveniently fall into any one of the above

sections, and yet is potentially applicable to all of them, is a comprehensive list of the

H- and C-chemical shift data of virtually all common laboratory solvents as trace in

impurities in a variety of deuterated NMR solvents. Such a useful compendium will

find application in the interpretation of a myriad of NMR spectra of organic com-

pounds.

References

1 T. Chounan, H. Horino, T. Ibuka and Y. Yamamoto, Bull. Chem. Soc. Jpn., 1997, 70, 1953.2 A. Gugger, R. Batra, P. Rzadek, G. Rist and B. Giese, J. Am. Chem. Soc., 1997, 119, 8740.3 D. Sato, H. Kawasaki, I. Shimada, Y. Arata, K. Okamura, T. Date and K. Koga, Tetrahedron, 1997, 53,

7191.

378 B. A. Salvatore


19/19

4 B. Bohler and H. Gunther, TetrahedronLett., 1996, 37, 8723.5 W. Baumann, S. Mansel, D. Heller, and S. Borns, Magn. Reson. Chem., 1997, 35, 463.6 S. Mansel, D. Thomas, C. Lefeber, D. Heller, R. Kempe, W. Baumann and U. Rosenthal, Organometallics,

1997, 16, 2886.7 P. A. Keifer, Drug Discov. Today, 1997, 2, 468.8 C. P. Holmes, J. Org. Chem., 1997, 62, 2370.9 P. A. Keifer, J. Org. Chem., 1996, 61, 1558.

10 M. J. Shapiro, J. Chin, R. E. Marti and M. A. Jarosinski, Tetrahedron Lett., 1997, 38, 1333.11 R. Jelinek, A. P. Valente, K. G. Valentine and S.J. Opella, J. Magn. Reson., 1997, 125, 185.12 C. Dhalluin, C. Boutillon, A. Tartar and G. Lippens, J. Am. Chem. Soc., 1997, 119, 10 494.13 P. A. Keifer, L. Baltusis, D. M. Rice, A. A. Tymiak and J. N. Shoolery, J. Magn. Reson. Series A, 1996, 119,

65.14 J. Chin, B. Fell, M. J. Shapiro, J. Tomesch, J. R. Wareing and A. M. Bray, J. Org. Chem., 1997, 62, 538.15 P. H. Seeberger, X. Beebe, G. D. Sukenick, S. Pochapsky and S. J. Danishefsy, Angew. Chem., Int. Ed. Engl.,

1997, 36, 491.16 P. N. W. Baxter, J.-M. Lehn, B. O. Kneisel and D. Fenske, Angew. Chem., Int. Ed. Engl., 1997, 36, 1978.17 O. Mogck, M. Pons, V. Bohmer and W. Vogt, J. Am. Chem. Soc., 1997, 119, 5706.18 T. Pehk, E. Kiirend, E. Lippmaa and U. Ragnarsson, J. Chem. Soc., Perkin Trans. 2, 1996, 2351.

19 S. D. Rychnovsky, T. I. Richardson and B. N. Rogers, J. Org. Chem., 1997, 62, 2925.20 J. Gervay, N.N. Mamuya and R.A. Barber, TetrahedronLett., 1997, 38, 1865.21 R. Harris, T. J. Rutherford, M. J. Milton and S. W. Homans, J. Biomol. NMR, 1997, 9, 47.22 R. Wagner and S. Berger, Magn. Reson. Chem., 1997, 35, 199.23 A. Poveda, C. Vicent, S. Penades and J. Jimenez-Barbero, Carbohydr. Res., 1997, 301, 5.24 W. A. Thomas, Prog. Nucl. Magn. Reson. Spectrosc., 1997, 30, 183.25 T. Bandyopadhyay, J. Wu, W. A. Stripe, I. Carmichael and A. S. Serianni, J. Am. Chem. Soc., 1997, 119, 1737.26 T. J. Church, I. Carmichael and A. S. Serianni, J. Am. Chem. Soc., 1997, 119, 8946.27 T. Church, I. Carmichael and A. S. Serianni, Carbohydr. Res., 1997, 280, 177.28 K. Furihata and H. Seto,Tetrahedron Lett., 1996, 37, 8901.29 J. C. Norrild and H. Eggert, J. Chem. Soc., Perkin Trans. 2, 1996, 2583.30 H. Kessler, Angew. Chem., Int. Ed. Engl., 1997, 36, 829.

31 S. B. Shuker, P. J. Hajduk, R. P. Meadows and S.W. Fesik, Science, 274, 1996, 1531.32 P. J. Hajduk, G. Sheppard, D. G. Nettesheim, E. T. Olejniczak, S. B. Shuker, R. P. Meadows, D. H. Stein-

man, G. M. Carrera, Jr., P. A. Marcotte, J. Severin, K. Walter, H. Smith, E. Gubbins, R. Simmer, T. F.Holzman, D. W. Morgan, S. K. Davidsen, J. B. Summers and S. W. Fesik, J. Am. Chem. Soc., 1997, 119,5818.

33 P.J. Hajduk, J. Dinges, G.F. Miknis, M. Merlock, T. Middleton, D. J. Kempf, D. A. Egan, K. A. Walter, T. S.Robins, S. B. Shuker, T. F. Holzman and S. W. Fesik, J. Med. Chem., 1997, 40, 3144.

34 P. J. Hajduk, E. T. Olejniczak and S. W. Fesik, J. Am. Chem. Soc., 1997, 119, 12 257.35 D. Canet, Prog. Nucl. Magn. Reson. Spectrosc., 1997, 30, 101.36 A. Bockmann and E. Guittet, J. Biomol. NMR, 1996, 8, 87.37 B. D. Fulton, R. Hrabal and F. Ni, J. Biomol. NMR, 1996, 8, 213.38 M. Mescher, A. Tannus, M. ONeil Johnson and M. Garwood, J. Magn. Reson., Series A, 1996, 123, 226.

39 M. J. Gradwell, H. Kogelberg and T. A. Frenkiel, J. Magn. Reson., 1997, 124, 267.40 W. S. Price, K. Hayamizu and Y. Arata, J. Magn. Reson., 1997, 126, 256.41 D. E. Zimmerman, C. A. Kulikowski, Y. Huang, W. Feng, M. Tashiro, S. Shimotakahara, C. Chien, R.

Powers and G. T. Montelione, J. Mol. Biol., 269, 1997, 592.42 K. Huang, M. Andrec, S. Heald, P. Blake and J. H. Prestegard, J. Biomol. NMR, 1997, 10, 45.43 G. V. Rodrigues, I. P. A. Campos and V. P. Emerenciano, Spectroscopy, 1997, 13, 191.44 V. P. Emerenciano, L. D. Melo, G.V. Rodrigues and J.P. Gastmans, Spectroscopy, 1997, 13, 181.45 R. Laatikainen, M. Niemitz, U. Weber, J.-P. Sundelin, T. Hassinen and J. Vepsalainen, J. Magn. Reson.,

Series A, 1996, 120, 1.46 E. Kolehmainen, K. Laihia, R. Laatikainen, J. Vepsalainen, M. Niemitz and R. Suontamo, Magn. Reson.

Chem., 1997, 35, 463.47 G. Cerichelli and G. Mancini, Curr. Opin. Colloid Interface Sci., 1997, 2, 641.48 M. Traikia, D. B. Langlais, G. M. Cannarozzi and P. F. Devaux, J. Magn. Reson., 1997, 125, 140.49 B. A. Salvatore, R. Ghose and J. H. Prestegard, J. Am. Chem. Soc., 1996, 118, 4001.50 K. P. Howard and J. H. Prestegard, Biophys. J., 1996, 71, 2573.51 R. S. Prosser and R. R. Vold, Biophys. J., 1997, 72, WPMC4.52 A. Bax and N. Tjandra, J. Biomol. NMR, 1997, 10, 289.53 N. Tjandra and A. Bax, J. Magn. Reson., 1997, 124, 512.54 H. E. Gottlieb, V. Kotlyar and A. Nudelman, J. Org. Chem., 1997, 62, 7512.

379NMR Spectroscopy

11 NMR Spectroscopy

Documents