NMR in Chemistry and the Life Sciences

New Challenges and Opportunities

  • Fig. 1: [13C,13C]-Inadequate of 0.8 M menthol (3.8 mg in 30 µl DMSO, 1.7 mm probe) (by courtesy of Bruker AG)Fig. 1: [13C,13C]-Inadequate of 0.8 M menthol (3.8 mg in 30 µl DMSO, 1.7 mm probe) (by courtesy of Bruker AG)
  • Fig. 1: [13C,13C]-Inadequate of 0.8 M menthol (3.8 mg in 30 µl DMSO, 1.7 mm probe) (by courtesy of Bruker AG)
  • Fig. 2: Workflow during automatic structure elucidation from NMR data (by courtesy of Bruker AG)
  • Fig. 3: Protein structure determination. Left 1H-15N correlation spectra of the N-terminal YM2 fragment in absence (top) or presence (bottom) of the unlabeled C-terminal fragment. Center Non-interfacial (top) and interfacial (bottom) medium- and long-range NOEs. Right Conformers of the protein complex and a scheme of the YM3A protein.

NMR (Nuclear magnetic resonance spectroscopy) has always constituted a central analytical technique in chemistry. Previously, experiments were performed and interpreted by NMR experts. Keeping the instruments running was a central task of dedicated technicians. Teaching how to interpret NMR spectra based on chemical shifts and scalar couplings was a central task in the curriculum. The investigated structures usually fell into the small-molecules category. This has all dramatically changed in the meantime.

Students in the Department of Chemistry at the University of Zurich (UZH) record NMR spectra on their own. They elucidate structures from first principles using correlations in 2D spectra. Instruments are used in equal share by researchers from inorganic, organic or physical chemistry. Most of the investigated molecules still belong to the class of small molecules but a large portion of instrument time is used for biological macromolecules. Studies comprise projects to determine protein or RNA structures, interactions of drug-like small molecules with proteins or interactions of metals with RNA. Chemical reactions are observed in the NMR tube, and variable-temperature NMR is exploited to extract activation energies. In short: The whole field has changed completely.
NMR has also sneaked quietly into other branches of science, in particular biochemistry and biology or materials science. Biology students here at UZH have been through chemistry and physics courses and are well-prepared to enter the field of NMR if desired.

A basic NMR setup requires a number of steps like locking, shimming and probe tuning/matching as well as calibration of pulses. These steps used to be fairly complicated, but run more or less in automatic fashion nowadays. However, students should have a basic understanding of these procedures. Because 2D NMR experiments play an essential role they also need to understand how these experiments work.

Practical Aspects of Analyzing NMR Spectra

The literature is full of excellent NMR books but often go far beyond of what is required for a typical user in chemistry or biology, while regularly failing to cover the practical aspects of performing NMR experiments.

Another shortage is often that they do not really suggest strategies for assigning spectra and for establishing covalent structures. In principle the experiments used for analyzing spectra of peptides, steroids and oligosaccharides are similar, but the approach to assign them is often rather different, and the success depends on using the best set of spectra, and starting the assignment from the proper place. Based on lecture courses at the UZH a book has been published recently [1], in which a thorough description of the basic NMR phenomenon, a detailed account of frequently used NMR experiments, a description of the practical aspects of NMR, and a section on assignments strategies for natural products and biopolymers are provided.

At UZH students once they have received a basic training are allowed to use the web-based NMR booking schedule, and are free to measure their own spectra. The philosophy is to provide a good training and then encourage them to measure their own spectra. In the following experiments are described which are applied to typical problems from chemistry and biochemistry in the NMR department at UZH.

Technical Advances

The principle of NMR spectrometers is still the same but a number of technical advances have made their operation much easier. For example, shimming, the adjustment of field homogeneity, frequently required time-consuming optimization. Nowadays, gradient-shimming, a method that directly images field inhomogeneity within the sample, works in automatic fashion. The introduction of pulsed magnetic field gradients [2], has had a large impact on the quality of certain spectra, and now helps obtaining nearly artifact-free spectra. Another consequence of using gradients is that the number of scans per increment in 2D experiments is mostly limited by the signal/noise. Single-scan proton-carbon correlations often give nice spectra and can be recorded in less than 10 minutes. Recently, methods for fast acquisition of multidimensional spectra have been introduced. One set of experiments use so-called sparse sampling. In the sparse sample (also non-uniform sampling (NUS) experiments) not all of the data points are measured but a fraction of them, and the missing ones are reconstructed using mathematical methods [3]. Another set of experiments has been developed in which experiments are repeated very rapidly [4,5]. Normally, a relaxation delay after each scan secures that the perturbed spins have returned to equilibrium, but the length of this delay can be significantly reduced when selective pulses are applied. Thereby the overall experiment duration is cut down to one-tenth, often allowing complete 2D spectra to be recorded within seconds for isotope-labeled samples.

Sensitivity is the primary bottleneck of NMR, and some advances have been made to increase signal to noise ratio (S/N). One (unfortunately expensive) development are the so-called cryo-probes, that bring down the thermal noise resulting in a S/N that is often 4 times larger [6]. An experiment that could have lasted a week can now be done in just one night! Other methods to increase S/N is the usage of other NMR tubes, either with smaller diameter (3 mm, 1.7 mm or 1 mm) or susceptibility-matched glass inserts so that in mass-limited cases the concentration will be much higher within the NMR-active volume of the probe. The combination of all these advancements allows to record experiments when this was completely impossible before. A good example is presented by the INADEQUATE experiment in Figure 1, a carbon-carbon (natural abundance 13C) correlation experiment, that used to require at least 50 mg of a small molecule compound, and which now becomes accessible with 10 mg or even less (in this case with approx. 4 mg).

Typical Small-Molecule Applications

The vast majority of spectra recorded at UZH still present 1D proton or carbon spectra, often using automatic sample changers. Other nuclei such as 19F, 31P, 29Si, 11B or 195Pt (as part of inorganic complexes for catalysis or medicinal applications) are also measured quite frequently.

Chemists mostly know what the molecular formula of the expected reaction product looks like. However, when this is unclear, de-novo structure elucidation is necessary. This usually requires 1D proton and carbon spectra (also to check the purity) plus a standard set of 2D NMR spectra, mostly COSY, TOCSY, ROESY (or NOESY), [13C,1H]-HSQC and [13C,1H]-HMBC data [1]. The latter two experiments are the proton-carbon correlations along the one-bond coupling and the long-range couplings, respectively. The incorporation of 13C shifts into assignments is mandatory for highly substituted compounds, and facilitates structure determination in general, in particular when using automatic methods (vide infra).

Software has been developed to evaluate spectra in a more automatic fashion. Peak multiplets are automatically evaluated to yield scalar couplings and chemical shifts and used to uncover correlations in spectra. In particular when incorporating 2D data, programs can nowadays propose complete structures of small molecules with good reliability. Figure 2 displays the workflow used for semi-automatic structure identification in commercial Bruker software.

The mantra of NMR analysis was the requirement for pure compounds. Sometimes reactions are monitored in the NMR tube, and the spectra will inherently contain compound mixtures. A method to separate a compound mixture into its constituents is diffusion-weighted measurement [7]. In the experiment signal losses are proportional to the fraction of molecules that have diffused significantly far away from their original position and hence are stronger for small than for large molecules. Such experiments can be recorded in a pseudo-2D fashion (called DOSY). In favorable cases sub-spectra of the individual molecules can be obtained.

Solid-state NMR spectroscopy used to be a playground for physicists. Recent technical advances have also helped to make the experiments simpler to perform. Disciplines of chemistry interested in solid-sate NMR spectra are those from material sciences or researchers in (heterogeneous) catalysis.
NMR is one of the methods suitable for determining activation energies of exchange processes [1,8]. At UZH a lot of variable temperature NMR spectra are recorded. Observing the change in peak position and linewidths as a function of temperature (or the determination of the coalescence temperature) allows extracting the reaction rates and the activation energy for the exchange process.

Applications in Life Science and Biochemistry

The usefulness of NMR is increasingly being recognized in biology and biochemistry as well. The determination of a protein structure is a considerable task, and often such a project, depending on the size of the protein and how well-behaved it is, can take a few years. The time-consuming step is the assignment of chemical shifts using data from 3D triple-resonance experiments. Structures of proteins exceeding 40 kDa are usually tough projects and require producing a lot of specially labeled protein samples. One can resort to 15N labeling even for small proteins (say even less then 50 residues, if there is a route for recombinant production), 15N, 13C double labeling in the range of 50-180 residues, while deuteration is required for those that are even larger. Figure 3 shows an example from research at UZH: It was discovered that the designed Armadillo repeat protein of the YM3A type, in which Y and A denote the N- and C-terminal caps and M the three internal (identical) repeats, can be reconstituted from two complementary fragments. The spectrum of YM2 on the top left clearly resembles a molten-globule type protein, and adding the unlabeled MA fragment converts that spectrum into one of a properly folded protein. The structure of the complex was determined based on NOEs.

Advances in spectroscopy have pushed the size-limit considerably. A major breakthrough was the development of the so-called TROSY-type pulse sequences [9], that allow to largely reduce the impact of the direct dipolar coupling for relaxation. To be most efficient all other protons should be removed by deuteration [10]. Lewis Kay has demonstrated that TROSY-type experiments can be used to assign the mega-Dalton assembly of the proteasome [11]. While this is certainly something for a highly specialized lab, the take home message here is that recording 15N-1H (or 13C-1H for methyl groups) correlation experiments are possible even for large proteins. Assigning them will be difficult, but sometimes this is not necessary, e.g. when screening for ligand-protein interactions. In fact, there are a number of biochemical problems that do not require extensive NMR expertise but just access to a spectrometer and a couple of hours of measuring time!

Even for the non-professional there is a lot to be discovered from simple protein NMR spectra: The chemical shift dispersion in 1D proton spectra will report on whether a protein is folded or not, and whether it is well-behaving (non-aggregating) at NMR concentrations. Signals from flexible residues are much more intense for large proteins. A typical question could be whether the expressed protein construct contains long flexible tails that hamper crystallization. If the protein can be produced in 15N-labeled form a proton-nitrogen correlation map will even allow to quickly identify flexible parts and their location. When the protein is produced in E. coli, producing the 15N-labeled form is usually easy, comparably cheap and worth the effort, because so much more information can be extracted from the 2D spectra.

Dynamic Studies

A fair amount of protein dynamic studies are also performed at UZH [12,14]. The backbone dynamics data are derived from 15N relaxation experiments. The data analysis is rather straightforward, and it adds valuable data to the function-dynamics topic. One of the experiments recorded is the 15N{1H}-NOE. Even in the absence of assignments this experiment will immediately reveal whether flexible tails or long flexible loops are present and thereby help to eliminate or truncate such flexible moieties.

Another question of interest often is whether the protein interacts with a small molecule, a typical question in drug-discovery [15,16]. In contrast to biochemical assays only two components exist in the NMR experiment, and if pH, salt content and temperature are tightly controlled, no false-positive will be seen. In the protein-observe methods the protein is usually 15N labeled. When adding the small molecule, peaks from residues in contact with the ligand will shift, indicating that the small molecule binds to the protein. If assignments already exist, the binding site can be rapidly identified. The advantage of the ligand-observe techniques is that no labeled protein is required. The saturation-transfer difference experiment (STD) [17] has become popular to detect binding on the ligand. The experiment is very simple to perform, and it works very well in presence of large receptor proteins.

Prof. Roland Sigel here at UZH determines interactions of metals with RNA as parts of the so-called riboswitches. The structural biology of RNA is much more versatile and interesting than DNA. Moreover, RNA can be labeled at comparably low cost. The determination of RNA structure has become routine nowadays [18], and procedures for structure determination are very similar to those used for proteins.

Another field from the life science area that has attracted some attention is metabolomics [19]. In metabolomics the presence of metabolites in biological fluids, e.g. in urine, is monitored. Urine from animals living under a certain diet can be compared to standard conditions. Metabolites can then be identified from reference spectra or data bases. Only soluble metabolites in significant concentrations can be detected.


This is a collection of a few applications that are frequently used in the Chemistry Department at UZH, and this overview is far from complete. Interested readers are therefore referred to the literature or to the book mentioned above [1].

[1] Jurt, S. & Zerbe, O. NMR for Chemists and Life Scientists (VCH-Wiley, Weinheim) (2013).
[2] Keeler J. et al.: Meth. Enzymol. 239, 145-207 (1994)
[3] Hyberts S. G. et al.: Top Curr. Chem. 316, 125-148´ (2012)
[4] Schanda P. et al.: J. Biomol. NMR 33, 199-211 (2005)
[5] Schanda P. and Brutscher B.: J. Am. Chem. Soc. 127, 8014-8015 (2005)
[6] Russell D. J. et al.: J. Nat. Prod. 63, 1047-1049 (2000)
[7] Johnson C. S.: Prog. NMR Spectrosc. 34, 203-256 (1999)
[8] Bain A. D.: Progr. NMR Spectrosc. 43, 63-103 (2003)
[9] Wider G.: Methods Enzymol. 394, 382-398 (2005)
[10] LeMaster D. M.: Methods Enzymol. 177, 23-43 (1989)

Please ask the authors for further references.

Simon Jurt, Oliver Zerbe

Department of Chemistry, University of Zurich,

Read an interview with Oliver Zerbe and Simon Jurt on their book "Applied NMR Spectroscopy for Chemists and Life Scientists"



University of Zürich
Winterthurerstr. 190 -272
8057 Zürich
Phone: 0041/44/6341111
Telefax: 0041/44/6344901

Register now!

The latest information directly via newsletter.

To prevent automated spam submissions leave this field empty.