TY - JOUR
AU - Köckenberger,, Walter
AB - Abstract Micro‐imaging based on nuclear magnetic resonance offers the possibility to map metabolites in plant tissues non‐invasively. Major metabolites such as sucrose and amino acids can be observed with high spatial resolution. Stable isotope tracers, such as 13 C‐labelled metabolites can be used to measure the in vivo conversion rates in a metabolic network. This review summarizes the different nuclear magnetic resonance micro‐imaging techniques that are available to obtain spatially resolved information on metabolites in plants. A short general introduction into NMR imaging techniques is provided. Particular emphasis is given to the difficulties encountered when NMR micro‐imaging is applied to plant systems. Metabolite pools in plants, non‐invasive imaging, nuclear magnetic resonance microscopy. Introduction The rapid development of molecular biological techniques over the last decade now allows plant biologists to investigate metabolism in plants in detail by changing, selectively, the expression of an enzyme in a cell or, by site‐directed mutagenesis, its kinetics and allosteric properties. Whilst these experiments have yielded information on the complexity of metabolic regulation, they have also highlighted that metabolism is often compartmentalized within the tissue. Studies using β‐glucuronidase gene fusions, indirect immuno‐localization or, more recently, the use of constructs with genes encoding for the green fluorescent protein ( Fricker and Oparka, 1999 ) have demonstrated the existence of spatial patterns of key proteins at both the cellular and tissue ( Oparka et al ., 1999 ; Schulze et al ., 1999 ; Stadler et al ., 1999 ) level. There is also strong evidence that the distribution of a range of metabolites is not homogeneous within the tissue ( Borisjuk et al ., 1998 ; Koroleva et al ., 1997) and that these patterns are important for the regulation of gene expression ( Koch et al ., 2000 ; Farrar et al ., 2000 ; Roitsch, 2000; Jang and Sheen, 1997 ; Koch, 1996 ; Fu et al ., 1995 ). However, the investigation of metabolite compartmentation in plants is difficult, since most of the existing techniques are destructive and require either tissue sections (e.g. micro‐autoradiography, tissue printing, single photon bioluminescence, and single‐photon imaging) or the introduction of chemical probes. Therefore, it is frequently impossible to observe changes in the in vivo metabolite concentration or to study fluxes in intact plant systems. Variations in the metabolite concentration such as the diurnal changes of carbohydrate pools during photosynthesis are, however, responsible for the complex control of the metabolic network within plants. Recently it has been demonstrated that techniques based on NMR micro‐imaging provide a versatile tool for monitoring water movement in plant research ( Scheenen et al ., 2000 a , b ; Rokitta et al ., 1999 ; Köckenberger et al ., 1997 ). However, NMR imaging techniques can also be used to monitor the temporal changes of metabolic pools. They provide spatially resolved and chemically selective information via a non‐invasive measurement, thus making possible the observation of temporal changes and fluxes through successive experiments. This review summarizes the fundamental principles of spectroscopic NMR micro‐imaging and the difficulties encountered when this technique is applied to plant systems. In addition, potential applications for the investigation of metabolic compartmentation in plants are discussed. Fundamental principles of NMR A few important principles are summarized here to provide a suitable background for the discussion of NMR micro‐imaging experiments on plant systems. A detailed introduction to nuclear magnetic resonance and its micro‐imaging variant can be found in various textbooks ( de Graaf, 1998 ; Kimmich, 1997 ; Callaghan, 1991 ; Slichter, 1990 ). A number of nuclei, such as 1 H, 13 C, 32 P, 17 O, 15 N, and 19 F possess a magnetic moment and angular momentum (nuclear spin). The magnetic dipole axes of the nuclei are usually randomly ordered. However, when exposed to a magnetic field, this field interacts with the magnetic moments of the nuclei. A new thermal equilibrium is achieved in which the population of nuclei with the magnetic dipole axes aligned parallel with the external field is slightly bigger than the population of the nuclei with anti‐parallel aligned axes. A weak magnetization of the sample, which can be represented by a vector M o (Fig. 1a ), results from this unequal population distribution. The sample magnetization can be measured through the induction of a voltage signal in a coil surrounding the sample after its manipulation by an appropriate combination of radio‐frequency pulses. One of the most fundamental principles underlying nuclear magnetic resonance is the proportionality between the resonance frequency (the Larmor frequency) and the magnetic field that acts on the nuclei. Both, the identification of the chemical nature of a compound by its resonance lines and the acquisition of spatially resolved images of a sample are based on this principle. Since the main magnetic field is shielded by surrounding electron clouds the magnetic field acting on a nucleus depends on the chemical nature of the group and the chemical environment in which it is bound. For instance, the hydrogen nuclei bound to carbons at different positions in a sucrose molecule can be identified by their different resonance frequencies due to differences in the local shielding of the magnetic field. In magnetic resonance imaging, magnetic field gradients are superimposed on to the main magnetic field to give the local field a spatial dependence. The resonance frequency thus becomes a measure of the location within the sample. It is instructive to consider a number of magnetization vectors, each corresponding to a different location in the sample (Fig. 2 ). Initially, all vectors represent magnetization with identical resonance frequency. During the application of a magnetic field gradient their resonance frequency depends on the position within the sample and therefore they lose their initial phase coherence due to the differences in their resonance frequency. After the magnetic field gradient is switched off, these frequencies again become identical. The spatial position, however, remains encoded in the phase shift, which each magnetization vector has accumulated during the gradient application. In a conventional imaging experiment with phase‐space‐encoding, the amplitude of the gradient pulse is incremented and the NMR signal is acquired for each repetition (Fig. 2 ). Another way to look at the phase shift of each magnetization vector is to think of the magnetization vectors being rotated with position‐dependent frequency through the application of the field gradient. Fourier analysis of the signal can yield the different frequencies. The spectrum resulting from this analysis represents a projection of the sample perpendicular to the direction of the magnetic field gradient (Fig. 2 ). In an imaging experiment the space is encoded by the application of magnetic field gradients in three dimensions, x, y, z. Fig. 1. Open in new tabDownload slide Schematic representation of the nuclear resonance phenomenon: The sample magnetization M o arises from the uneven distribution of the nuclear spin ensemble between two different states with their axes either aligned parallel or anti‐parallel to the axis of the main magnetic field B o . The magnetic dipoles precess (here indicated as vectors) around the main field axis with the Larmor frequency ω. After the application of a π/2 pulse the original distribution is shifted away from the thermodynamic equilibrium and a phase coherence is established between the precessing dipoles. The result is a sample magnetization M ⊥ , precessing in the transverse plane with a frequency ω. The spin ensemble returns to the original distribution between the two states through T 1 relaxation. The loss of phase coherence is described by T 2 relaxation. Both processes occur simultaneously but are depicted separately in the diagram. Fig. 1. Open in new tabDownload slide Schematic representation of the nuclear resonance phenomenon: The sample magnetization M o arises from the uneven distribution of the nuclear spin ensemble between two different states with their axes either aligned parallel or anti‐parallel to the axis of the main magnetic field B o . The magnetic dipoles precess (here indicated as vectors) around the main field axis with the Larmor frequency ω. After the application of a π/2 pulse the original distribution is shifted away from the thermodynamic equilibrium and a phase coherence is established between the precessing dipoles. The result is a sample magnetization M ⊥ , precessing in the transverse plane with a frequency ω. The spin ensemble returns to the original distribution between the two states through T 1 relaxation. The loss of phase coherence is described by T 2 relaxation. Both processes occur simultaneously but are depicted separately in the diagram. Fig. 2. Open in new tabDownload slide Schematic representation of spatial encoding by an incremented pulsed magnetic field gradient g. Between excitation by a π/2 pulse and acquisition of the echo a magnetic field gradient is switched and incremented after each acquisition. The Larmor frequency ω of the magnetization vector is spatially dependent during the application of the magnetic field gradient. The position is encoded in the signal by the different phase shifts after the application of the magnetic field gradient g. These phase shifts are shown schematically for spin ensembles at four different positions in the sample. Fourier transformation of the signal along the repetition axis resolves the spatial dependent frequency components of the signal and yields a one‐dimensional projection of the sample perpendicular to the direction of the magnetic field gradient. Fig. 2. Open in new tabDownload slide Schematic representation of spatial encoding by an incremented pulsed magnetic field gradient g. Between excitation by a π/2 pulse and acquisition of the echo a magnetic field gradient is switched and incremented after each acquisition. The Larmor frequency ω of the magnetization vector is spatially dependent during the application of the magnetic field gradient. The position is encoded in the signal by the different phase shifts after the application of the magnetic field gradient g. These phase shifts are shown schematically for spin ensembles at four different positions in the sample. Fourier transformation of the signal along the repetition axis resolves the spatial dependent frequency components of the signal and yields a one‐dimensional projection of the sample perpendicular to the direction of the magnetic field gradient. T 2 * relaxation and echo formation Another important phenomenon can be understood if the same picture as above is used, in which the sample magnetization is divided in a number of magnetization vectors representing subunits of the sample. Initially, after the application of a π/2 pulse these magnetization vectors are synchronized and in phase (Fig. 3b ). However, the precession frequencies differ within a sample in which the homogeneity of the main magnetic field is perturbed by local differences of the magnetic susceptibility. Therefore, the synchronization of the precession deteriorates continuously. While the local magnetization vectors fan out in the transverse plane the average magnetization of the sample decreases with a time constant T 2 * (Fig. 3c ). It is possible to reverse this process by the application of a π pulse (Fig. 3d ). This pulse rotates the magnetization vectors by 180° about a common axis with the effect that the phase coherence is restored after another evolution delay that equals the delay between excitation pulse and refocusing pulse. The detectable sample magnetization recovers during the formation of an echo and the amplitude of the echo is attenuated only by transverse relaxation (Fig. 3f ). Fig. 3. Open in new tabDownload slide Schematic depiction of an echo after the application of a π‐pulse. (a) Rotation of the magnetization vector into the transverse plane by the π/2 pulse. After the creation of transverse magnetization through a π/2 pulse the phase coherence between the magnetization vectors in an inhomogeneous magnetic field decreases continuously (b, c). A π pulse at time τ rotates the magnetization vectors by 180° (d). This results in a reversal of the loss of phase coherence (e) and an echo is created after a time 2τ (f). Fig. 3. Open in new tabDownload slide Schematic depiction of an echo after the application of a π‐pulse. (a) Rotation of the magnetization vector into the transverse plane by the π/2 pulse. After the creation of transverse magnetization through a π/2 pulse the phase coherence between the magnetization vectors in an inhomogeneous magnetic field decreases continuously (b, c). A π pulse at time τ rotates the magnetization vectors by 180° (d). This results in a reversal of the loss of phase coherence (e) and an echo is created after a time 2τ (f). T 1 , T 2 relaxation and diffusion In micro‐imaging applications it is important to consider diffusion of the molecules and relaxation processes during the time period between excitation by the π/2 pulse and acquisition. After the thermal equilibrium is perturbed by the initial π/2 pulse, the nuclei start to return to the initial state by relaxation (Fig. 1 ). The initial population distribution between the different states can be restored only by an exchange of energy between the excited nuclei and their environment. This is a first order process with a time constant T 1 for an identical ensemble of nuclei. In the classical mechanical treatment longitudinal relaxation represents the recovery of the bulk magnetization vector in the direction of the main magnetic field. The phase coherence can only be restored through a π pulse if the local magnetic field does not change in the evolution period. However, there are continuous fluctuations of the local magnetic fields due to rotational and translational molecular motion, which cause an unrecoverable loss of the phase coherence (Fig. 1 ). The resulting decay of the average sample magnetization vector in the transverse plane is again a first order process with a time constant T 2 . The mechanistic explanations of both longitudinal and transverse relaxation rates include a variety of physical processes, such as inter‐ and intramolecular dipole–dipole interactions between two nuclei and the effect of paramagnetic ions. The local variation of relaxation parameters can be used to create contrast in NMR micro‐images ( Xia, 1996 ; Callaghan, 1991 ). Typical values for proton T 1 and T 2 in plants at 11.75 T (500 MHz) are 1.5 s and 15 ms, respectively. T 2 *, which determines the linewidth in a spectroscopic experiment, is, however, mostly shorter and of the order of 5–10 ms at such high field strength. In young castor bean plants, this time constant is around 15 ms, therefore making this plant particularly appropriate for spectroscopic NMR micro‐imaging studies. The most important process contributing to the unrecoverable loss of phase coherence in plants is the diffusion of the molecules (e.g. the water molecules) in internal magnetic field gradients in the time period between excitation and signal acquisition (diffusive motion of water: 1–2 μm 2  ms −1 ). The precession frequency of the magnetic dipole changes continuously while the molecule is moving through areas with different local magnetic fields. The accumulated phase shift during the random motion cannot be completely reversed by the application of the π pulse. Therefore, diffusion in an inhomogeneous magnetic field increases the apparent transverse relaxation rate and gives rise to additional line broadening and attenuation of the NMR signal. The inhomogeneity of the magnetic field in plants stems mainly from the presence of small air‐filled intercellular spaces ( Callaghan et al ., 1994 ; Bowtell et al ., 1990 ; Connelly et al ., 1987 ). Air and water have different magnetic susceptibility when exposed to a magnetic field (Δχ=9×10 −6 ). The susceptibility difference can result in strong internal magnetic field gradients at the interface between water and air. Since these internal magnetic field gradients can be much stronger than the externally applied field gradients for spatial encoding, they can also lead to image distortions in addition to the increase in the T 2 relaxation rate ( Callaghan et al ., 1994 ; Bowtell et al ., 1990 ). The internal gradients depend linearly on the strength of the applied field. Therefore, susceptibility artefacts and shortening of the transverse relaxation time constant T 2 can become more pronounced with increased magnetic field strength. This dependency can counteract the expected signal increase when working with high field magnets. Signal‐to‐noise ratio The signal in an NMR experiment is a small electromotive force that is induced in the detection coil by the precession of the sample magnetization in the transverse plane. The created voltage across both ends of the coils is of the order of several μV. The voltage is amplified and digitized. However, from the coil to digitization there are a number of sources for experimental noise that impair the detection sensitivity ( Hoult and Richards, 1976 ). The limitation of the available signal‐to‐noise ratio is one of the major obstacles for gaining higher sensitivity for the detection of metabolites or increasing the spatial resolution. The signal‐to‐noise ratio is approximately proportional to the square of the applied magnetic field (B o7/4 ) and to the number of nuclei within the sample. In addition, it depends on the design of the coil, the temperatures of the sample and the coil and experimental parameters such as the detection bandwidth. Averaging the signal increases the signal‐to‐noise ratio but only by the square root of the number of averages. Therefore signal averaging leads rapidly to very long experimental times. The duration of the experiment, the achievable spatial resolution within this time and the concentration of a molecule that is still detectable in the experiment depend on each other in a triangular relation. For instance, if the experiment has to be made more sensitive to metabolites with lower concentrations either the spatial resolution can be decreased (that is, larger volume elements are investigated) or the experimental time can be increased. Similarly, if the spatial resolution should be increased, a longer experimental duration is necessary and/or one has to reduce on the detection sensitivity for metabolites. In fact, the experimental time is proportional to the sixth power of the inverse of the resolution. Therefore, if the resolution has to be improved by a factor of two in all three dimensions the experimental duration will increase by a factor of 64. Micro‐imaging and compartmentation Figure 4a shows a 1 H NMR micro‐image of an intact carrot taproot. The carrot was grown in a 30 mm diameter glass tubes filled with soil substrate and the glass tube was inserted into a micro‐imaging probe head. The contrast in the image is due to differences in the local relaxation parameters. The image has an in‐plane resolution of 117×117 μm. Xylem and xylem parenchyma can be recognized in the centre of the taproot surrounded by a ring of meristematic tissue with very high intensity. Located around the meristematic tissue are the phloem parenchyma and the phloem. There is a base of a side root on one side of the carrot taproot. Figure 4b shows a micro‐image of a Tradescantia stem with an in‐plane resolution of 28×28 μm. The high spatial resolution in this experiment was favoured by the comparatively long T 2 * and T 2 time constants in Tradescantia plants, which are about 13 ms and 23 ms, respectively. Another example of a 1 H NMR micro‐imaging experiment is provided by the image in Fig. 6a . The image was acquired from the hypocotyl of 6‐d‐old castor bean seedlings. The in‐plane resolution is 24×24 μm with 1 mm slice thickness. The eight vascular bundles on a ring of meristematic tissue are clearly recognizable. The parenchyma cells in Tradescantia stems and castor bean hypocotyls have a large diameter (>100 μm). In addition, the parenchyma of castor bean plants consists of columns of very large cells (100–200 μm diameter) which makes it possible to select a thick slice (1000 μm) and to average over several cell layers while preserving the information of the anatomical structure. Cells with small diameters (∼10 μm) can hardly be resolved in an NMR micro‐imaging experiment. Nevertheless, the unique ability of NMR imaging to create different contrasts by the appropriate choice of experimental parameters can be used to acquire images that show a high degree of anatomical detail. If relaxation time constants are different in the various compartments such as vacuoles, cytoplasm and apoplast, the contribution of water in the individual compartments to the total signal can be estimated by appropriate NMR experiments ( van Dusschoten et al ., 1996 ; Snaar and van As, 1992 a ) and the use of exchange models between the compartments (Belton and Ratcliffe, 1985). In principle, it is feasible to estimate membrane permeability coefficients of water in the intact tissue if some simplifications are introduced ( Snaar and van As, 1992 b ). This approach has provided insight into water exchange processes in maize and millet plants during mild water stress ( van der Weerd et al ., 1998 ). Fig. 4. Open in new tabDownload slide (a) 1 H NMR image of a carrot tap root. The carrot was grown in a glass tube filled with soil substrate. The in‐plane resolution is 117×117 μm with 1.5 mm slice thickness. The total acquisition time was 25 min. The meristematic tissue shows high intensity in the image (arrow). The xylem and xylem parenchyma is inside the ring of meristematic tissue. The tissue around the meristematic tissue consists of phloem and phloem parenchyma. The base of a side root can be recognized on one side of the carrot. (b) 1 H image of a Tradescantia stem. The in‐plane resolution is 27×27 μm with a 1 mm thick slice. The disperse vascular bundles appear as black regions in the parenchyma cells. Fig. 4. Open in new tabDownload slide (a) 1 H NMR image of a carrot tap root. The carrot was grown in a glass tube filled with soil substrate. The in‐plane resolution is 117×117 μm with 1.5 mm slice thickness. The total acquisition time was 25 min. The meristematic tissue shows high intensity in the image (arrow). The xylem and xylem parenchyma is inside the ring of meristematic tissue. The tissue around the meristematic tissue consists of phloem and phloem parenchyma. The base of a side root can be recognized on one side of the carrot. (b) 1 H image of a Tradescantia stem. The in‐plane resolution is 27×27 μm with a 1 mm thick slice. The disperse vascular bundles appear as black regions in the parenchyma cells. Fig. 6. Open in new tabDownload slide Set of correlation peak images of the hypocotyl of a 6 d castor bean seedling and corresponding high‐resolution 1 H‐image as a reference. (a) 1 H‐image, in‐plane resolution is 24×24 μm with 1 mm slice thickness, (b) correlation peak images showing the distribution of glutamine, (c) valine, (d) arginine. In‐plane resolution of the correlation peak images is 375×375 μm, slice thickness 4 mm. Total acquisition time for this experiment was 4.5 h. (Modified from Metzler et al . Plant histochemistry by correlation peak imaging. Proceedings of the National Academy of Sciences, USA92, 11912–11915. Copyright 1995. With permission of National Academy of Sciences, USA.) Fig. 6. Open in new tabDownload slide Set of correlation peak images of the hypocotyl of a 6 d castor bean seedling and corresponding high‐resolution 1 H‐image as a reference. (a) 1 H‐image, in‐plane resolution is 24×24 μm with 1 mm slice thickness, (b) correlation peak images showing the distribution of glutamine, (c) valine, (d) arginine. In‐plane resolution of the correlation peak images is 375×375 μm, slice thickness 4 mm. Total acquisition time for this experiment was 4.5 h. (Modified from Metzler et al . Plant histochemistry by correlation peak imaging. Proceedings of the National Academy of Sciences, USA92, 11912–11915. Copyright 1995. With permission of National Academy of Sciences, USA.) Imaging of metabolites Metabolites can be identified in NMR spectra by the characteristic chemical shift of their resonance lines. The chemical shift effect scales with the strength of the static magnetic field. Therefore, a large magnetic field gives a dispersion of the resonance lines over a wide range. The in vivo identification of the chemical nature of the metabolites is frequently made difficult by the broad linewidths of the resonance lines which is caused by the short T 2 * relaxation in plant tissue. In addition, the resonance line of the water protons in the plant tissue can dominate the in vivo1 H NMR spectrum. Since the gain of an NMR spectrometer is adjusted according to the resonance with the highest intensity, a large difference between the intensity of the water resonance and the metabolite resonance lines makes the detection of the metabolites difficult. Therefore, strategies for water suppression can improve the detection sensitivity for low concentrated metabolites in plant tissue ( Xia and Jelinski, 1995 ). Since the concentrations of major metabolites such as sucrose and various amino acids in plants are in comparison to water a factor of 10 3 –10 5 lower, experiments for the in vivo detection of metabolites require longer experimental times and can not provide information with the same high spatial resolution as conventional NMR micro‐imaging using the water protons. Typically, the in‐plane resolution is several hundreds of micrometres with observed slices of several millimetres thickness. Thick slices are, however, frequently favoured by the axial homogeneity of the plant tissue such as in shoots. With such parameters and favourable plants such as castor bean seedlings it is possible to detect metabolites with concentrations as low as a few mM. Chemical shift imaging One possible experiment to map metabolites in plants is selective excitation of only the resonance frequency of the desired metabolite in a chemical shift imaging experiment. This is especially practicable if a characteristic resonance of the metabolite of interest is well separated from all other resonance lines within the sample ( Pope, 1992 ). However, several problems can make the application of this technique to plant tissue difficult. First, the resonance frequency can be shifted due to internal susceptibility gradients, therefore making a homogeneous excitation across the sample difficult ( Pope et al ., 1992 ). Second, it is difficult to judge how much the signal is contaminated by signal arising from other metabolites since only the total intensity is recorded for each volume element. The advantage of the technique lies in the fact that it is fast in comparison to the techniques discussed in the next sections. Xia and Jelinski ( Tse et al ., 1996 ; Xia and Jelinski, 1995 ) have applied this technique to the detection of sucrose in pea seeds with a sensitivity of 100 mM. Pope and co‐workers used chemical shift imaging for the mapping of sugars and lipids in grapes and fennel seeds ( Pope et al ., 1991 , 1993). Spectroscopic imaging In spectroscopic imaging a complete spectrum is recorded for each volume element ( Brown et al ., 1982 ). While chemical shift imaging requires the repetition of only n separate excitations to obtain an image, spectroscopic imaging needs n × n separate repetitions to acquire a complete data set. This technique yields a three‐dimensional data set with two spatial and one spectral dimension. It is more time‐consuming in comparison to chemical shift imaging. However, the longer acquisition time is compensated by the information that can be extracted from the spatially resolved spectra. In principle, it is possible to monitor more than one metabolite simultaneously using this technique, provided that the resonance lines of the other metabolites are well separated with a sufficient signal‐to‐noise ratio. In addition, the signal overlap for each resonance line can be examined in the spatially resolved spectra. This technique is therefore less sensitive to artefacts introduced by susceptibility variations within the plant tissue. In order to improve detection sensitivity it is frequently necessary to suppress the water resonance line. Rumpel and Pope were the first to apply this technique to plant systems ( Rumpel and Pope, 1992 ). They obtained images of the distribution of anethole and other chemical compounds in the fennel mericarp (Fig. 5 ). The sucrose distribution in castor bean seedlings was measured quantitatively by calibration of the spatially resolved spectra with spectra from a sucrose reference ( Metzler et al ., 1994 ). The relaxation times of the sucrose resonance lines were determined and the parameter of the spectroscopic imaging experiment adjusted to minimize the effect of the variations in relaxation rates across the hypocotyl on the quantification. The total sucrose concentration measured non‐invasively in the plant tissue by the NMR technique was in good agreement with the sucrose concentration found in extracts of hypocotyl sections. Further optimization of the experimental parameters of the spectroscopic imaging experiment made possible the observation of the dynamic variations of the sucrose concentration in the phloem ( Verscht et al ., 1998 ). The temporal resolution in these experiments was 70 min. Fig. 5. Open in new tabDownload slide Spectroscopic images of a dried fennel mericarp. In‐plane resolution is 30×30 μm, slice thickness 1 mm. (a) Image showing the distribution of protons in a methylene (chemical shift: δ=1.3 ppm) and a methyl group (δ=1.8 ppm). (b) Image of protons in the methoxy resonance of anethole (δ=3.8 ppm). (c) Image of olefinic protons in reserve oil (δ=5.4 ppm), (d) Image of aromatic/olefinic protons of anethole (δ=7.0 ppm). (Reprinted from Rumpel and Pope. The application of 3D chemical shift microscopy to non‐invasive histochemistry. Magnetic Resonance Imaging10, 187–194. Copyright 1992 Elsevier Science. With permission of the author and the publisher.) Fig. 5. Open in new tabDownload slide Spectroscopic images of a dried fennel mericarp. In‐plane resolution is 30×30 μm, slice thickness 1 mm. (a) Image showing the distribution of protons in a methylene (chemical shift: δ=1.3 ppm) and a methyl group (δ=1.8 ppm). (b) Image of protons in the methoxy resonance of anethole (δ=3.8 ppm). (c) Image of olefinic protons in reserve oil (δ=5.4 ppm), (d) Image of aromatic/olefinic protons of anethole (δ=7.0 ppm). (Reprinted from Rumpel and Pope. The application of 3D chemical shift microscopy to non‐invasive histochemistry. Magnetic Resonance Imaging10, 187–194. Copyright 1992 Elsevier Science. With permission of the author and the publisher.) Correlation peak imaging Frequently, metabolites cannot be identified in a conventional one‐dimensional spectrum because of the signal overlap of their broad resonance lines in in vivo experiments. For instance, the proton resonance lines of several amino acids are scattered closely together around 2.1 ppm and 1.5 ppm. One solution to improve the identification of chemical compounds is the use of two‐dimensional correlation spectroscopy. The two‐dimensional spectroscopic experiment can be combined with spatial encoding to give an experiment that provides information in four dimensions, two spatial and two spectral dimensions ( von Kienlin et al ., 2000 ; Ziegler et al ., 1996 ; Metzler et al ., 1995 ). This technique has been termed correlation peak imaging since the spatial distribution of a single cross peak representing a chemical group in a metabolite is measured. A further increase in experimental time is required since, in addition to the two space encoding gradients, an additional evolution delay is incremented in several steps. Cross peaks from the two‐dimensional spectra can be selected and maps of the distribution of the corresponding metabolites can be obtained by displaying the spatial distribution of their intensities. A variant has also been developed to acquire metabolite maps from plants with high radial symmetry. Here, the spatial encoding was modified to allow the observation of concentric annular volume elements rather than the conventional quadratic raster ( Meininger et al ., 1997 ). 13 C imaging The stable isotope 13 C can be detected in an NMR experiment, however, the relative detection sensitivity is in comparison to 1 H nuclei, a factor of 64 worse. In addition, the natural abundance of 13 C is only 1.1%. The low abundance can also be considered as an advantage, since it is possible to utilize position‐labelled 13 C compounds as tracers in an in vivo investigation of the flux through a metabolic network. A number of experiments in which position‐labelled precursors were supplied to excised plant tissue and perfused cells have demonstrated the versatility of this approach in studying primary and secondary plant metabolism ( Roberts, 2000 ; Ratcliffe, 1994 ). However, the low NMR sensitivity for 13 C nuclei has restricted the use of 13 C direct detection to spectroscopic investigations. The sensitivity problem can be overcome by using indirect detection techniques. In particular, it has been demonstrated that is it possible to detect 13 C nuclei in a plant with an enhancement of almost the theoretical factor 64 using a cyclic polarization transfer technique ( Heidenreich et al ., 1998 ). In this experiment, polarization is transferred from the protons in a chemical group onto the 13 C nuclei and back to the protons, while all residual proton resonance lines in the sample are suppressed ( Kunze and Kimmich, 1994 ). 13 C‐labelled fructose and glucose were supplied to the cotyledons of castor bean seedlings. These hexoses were taken up and sucrose was synthesized and exported into the hypocotyl. The arrival and the accumulation of the labelled sucrose were monitored non‐invasively using the cyclic polarization transfer techniques in connection with micro‐imaging (Fig. 7 ) ( Heidenreich et al ., 2001 ). Fig. 7. Open in new tabDownload slide Set of three indirect detected 13 C‐images of the hypocotyl of a 6‐d‐old castor bean seedling and a high resolution 1 H image as reference. (a) 1 H‐image, (b) indirect detected 13 C‐image 4.3 h after supply of 13 C‐labelled hexoses ( 13 C 1 ‐glucose and 13 C 1 ‐fructose) to the cotyledons, (c) after 9.5 h, (d) after 16.4 h. The resonance line in the lower right hand corner indicates the total indirect detected 13 C‐signal. Note the appearance of the 13 C‐signal in vascular bundles first (arrows). The spatial resolution of the 13 C‐images is 94×380 μm, with a slice thickness of 4 mm. The acquisition time for each indirect detected 13 C‐image was 1.5 h. Fig. 7. Open in new tabDownload slide Set of three indirect detected 13 C‐images of the hypocotyl of a 6‐d‐old castor bean seedling and a high resolution 1 H image as reference. (a) 1 H‐image, (b) indirect detected 13 C‐image 4.3 h after supply of 13 C‐labelled hexoses ( 13 C 1 ‐glucose and 13 C 1 ‐fructose) to the cotyledons, (c) after 9.5 h, (d) after 16.4 h. The resonance line in the lower right hand corner indicates the total indirect detected 13 C‐signal. Note the appearance of the 13 C‐signal in vascular bundles first (arrows). The spatial resolution of the 13 C‐images is 94×380 μm, with a slice thickness of 4 mm. The acquisition time for each indirect detected 13 C‐image was 1.5 h. Filtering Another approach to selecting only the resonance arising from a chemical group of interest is the application of multiple quantum filtering. Here, a particular coherence is selected with an appropriate pulse sequence and all unwanted coherences, including the water proton resonance are effectively suppressed. In medical NMR imaging such filters have been successfully used to detect the resonance of lactate and glucose in muscles or brain, respectively ( de Graaf et al ., 2000 ; Lei and Peeling, 1999 ; Keltner et al ., 1998 ; van Dijk et al ., 1992 ). However, coherence selection by double quantum filtering usually results in the loss of up to 75% of the initial magnetization of the chemical group of interest. This disadvantage limits the use of this technique to the detection of metabolites with a sufficiently high concentration and could, therefore, counterbalance the excellent performance of this technique in suppressing the unwanted resonance of water and other molecules. Recently, the distribution of carbohydrates was measured in sugar cane ( Saccharum officinatum L.) by spectroscopic imaging in conjunction with multiple quantum filtering ( Wolf et al ., 2000 ). This work provides a good example of the suppression of unwanted resonance line by this technique. Other nuclei Apart from 1 H and 13 C nuclei, micro‐imaging experiments on plants have been performed with sodium and deuterium. In the 23 Na‐NMR experiment, NaCl was supplied to the roots of castor bean seedlings and the uptake and accumulation of the sodium ion was observed ( Olt et al ., 2000 ). In this experiment, the favourable NMR properties of the castor bean seedling made a high spatial (156×156 μm) and temporal resolution (85 min) possible. Interesting accumulation patterns were discovered in the time‐course study of NaCl uptake from a NaCl solution supplied to the roots with a concentration as low as 200 mM. Deuterium has been used as a tracer in water uptake studies of plants ( Link and Seelig, 1990 ). However, a rather large concentration of deuterated water (up to 50%) was used. Katz and co‐workers have described the reduction of plant growth with the percentage of deuterated water supplied to the roots ( Katz and Crespi, 1970 ). 20% deuterated water showed a marked effect on the growth of mentha plants and, therefore, deuterium can only be used as a low‐concentration tracer in NMR experiments. One reason for this observation could be the inhibition of plasma membrane bound ATPases by deuterated water ( Sacchi and Cocucci, 1992 ). Other nuclei such as 15 N can, in principle, be used for NMR imaging experiments. However, the relative NMR sensitivity of 15 N is a factor 1000 worse than in 1 H NMR. Therefore, similar enhancement schemes as described for the 13 C indirect detection have to be used to allow the acquisition of NMR data with spatial encoding. However, protons bound to the nitrogen atom in an amino group show a rapid exchange at physiological pH values. The polarization transfer efficiency in an indirect detection experiment is severely impaired by this fast exchange ( Fox et al ., 1992 ). A nucleus with a high relative NMR sensitivity is 19 F. However, the natural abundance of fluorine in plants is very low. Nevertheless, xenobiotic compounds could be introduced into plants to study transport or flux through a particular metabolic pathway ( Ratcliffe and Roscher, 1998 ). An interesting aspect of 19 F NMR is the high sensitivity of the chemical shift of some fluorinated molecules to the pH value of their environment. Therefore, such compounds may be useful to monitor internal pH changes in plants non‐invasively. Quantification Quantification of metabolites in a spectroscopic NMR micro‐imaging experiment is possible if reference capillaries with solutions of the metabolites of interest are attached to the plant. The localized spectra of the plant tissue can then be calibrated during post‐processing using the resonance lines from the compounds in the reference capillary. However, since the NMR signal is very sensitive to properties of the immediate environment of the metabolite, spatial variation in signal intensity can also reflect the spatial dependence of these properties rather than variations of the metabolite concentration. This is a problem particularly in heterogeneous systems such as plants where relaxation parameters can vary strongly between different types of tissue. Ideally, the effect of these variations must be minimized by the proper choice of the experimental parameters or these variations have to be taken into account in any attempt to estimate the local in vivo metabolite concentration. In the extreme case, a fraction of a metabolite pools could be invisible in an NMR experiment due to rapid relaxation of the detectable magnetization. This indicates the location of the metabolite in different compartments with different relaxation properties. Such NMR invisible pools of metabolites have been described in medical application of spectroscopic imaging ( Schneider et al ., 2000 ). A related problem arises if indirect detection techniques or filters are used. Here, the coherences decay with different relaxation rates than in conventional 1 H imaging. Since these relaxation rates can also be spatially dependent the effect of differential relaxation has to be considered if quantification is required. The spatial variations of physical and chemical parameters that are accessible through NMR micro‐imaging experiments might shed light on interesting aspects of compartmentation in the physiology of plants. For instance viscosity, chemical exchange and diffusion all have an effect on relaxation rates. These parameters can be investigated with the appropriate micro‐imaging experiments. Quantification by comparison of the localized spectra with spectra of a reference solution is also biased by differences in the liquid fraction in the volume elements. In a reference capillary the volume element consists of solution only while a volume element in plant tissue sometimes contains a considerable fraction of solid‐like material such as starch or membrane and cell walls. The actual metabolite concentration in the plant tissue is then underestimated due to this partial volume effect. NMR micro‐imaging as a tool in the investigation of plant cell metabolism The various NMR micro‐imaging techniques available for the investigation of cell metabolism can be divided into two categories. In the first category the techniques can be used to measure the distribution of metabolites within the intact plants. In this respect they compete with destructive techniques such as tissue printing, some staining techniques of optical micrographs, single cell sap sampling of tissue cross sections or single photon bioluminescence and single‐photon imaging. NMR micro‐imaging can be used to detect a number of metabolites directly in intact plants. Its non‐invasive nature is clearly an advantage when transient changes of metabolic pools are monitored. For instance, there is no other technique available to detect changes in the sucrose concentration within the phloem of intact plants. In addition to the detection of temporal changes, the individual vascular bundles can be resolved within the parenchymatic tissue thus making studies of the responses of individual phloem strands possible ( Verscht et al ., 1998 ). Such experiments are important in, for example, the investigation of the effects that can be caused through the genetic modification of the expression of sucrose transporter proteins in the phloem tissue ( Bürkle et al ., 1998 ). NMR imaging could also be useful to map the distribution of metabolites that accumulate in special tissue structures such as oil canals of the fennel mericarp ( Rumpel and Pope, 1992 ). The cost of the non‐destructiveness and the direct detection is that the spatial resolution is generally inferior to the resolution achievable with other spectroscopic techniques. For instance, in correlation peak imaging experiments the patterns of amino acids in the hypocotyl of castor bean seedlings were detected for amino acids with an in‐plane resolution of 375×375 μm. However, this resolution was still sufficiently high to reveal novel information on the tissue‐specific accumulation patterns of the amino acids that would be difficult to obtain by any other technique ( Metzler et al ., 1995 ). NMR micro‐imaging experiments in the second category provide dynamic information in a completely different way. These techniques are used to detect NMR tracers such as the stable carbon isotope 13 C. In principle, it is possible to investigate the transport of these tracers, their accumulation rates and, provided the metabolite pools are large enough for their detection by NMR, the conversion of one metabolite into another by observing changes in the pattern of their 13 C resonance lines. Interesting applications for these techniques are experiments in which accumulation rates or fluxes can be detected in intact plants and compared between wild type and transgenic plants with alterations of primary metabolic pathways. Conclusions NMR micro‐imaging is probably one of the most versatile non‐invasive techniques to measure multiple chemical and physical parameters with high spatial resolution. The application to plants is challenging since plants are, in respect to the NMR experiments, very heterogeneous systems. Some plants such as castor bean plants appear to be more appropriate than other plant species. However, NMR micro‐imaging techniques have not yet been applied frequently to plants and a systematic investigation might provide a more detailed picture of the effect of compartmentation on the NMR micro‐image. Spatially resolved spectroscopy and spectroscopic imaging have the potential to help in the investigation of metabolite compartmentation in plant tissue of major plant metabolites such as hexoses, sucrose and amino acids and therefore could also be a useful tool in the investigation of sugar sensing in plant tissue. 1 To whom correspondence should be addressed. Walter.Kockenberger@nottingham.ac.uk The author would like to thank the Royal Society for continuing support. The author is also grateful to Dr R. Bocotell for carefully reading the manuscript. References Belton PS, Ratcliffe RG. 1985 . NMR and compartmentation in biological tissues. Progress in NMR Spectroscopy 17 , 241 –279. Borisjuk L, Walenta S, Weber H, Mueller‐Klieser W, Wobus U. 1998 . High‐resolution histographical mapping of glucose concentration in developing cotyledons of Vicia faba in relation to mitotic activity and storage processes: glucose as a possible developmental trigger. The Plant Journal 15 , 583 –591. Bowtell RW, Brown GD, Glover PM, McJury M, Mansfield P. 1990 . Resolution of cellular structures by NMR microscopy at 11.7 T. Philosophical Transactions of the Royal Society London A 333 , 457 –467. Brown TR, Kincaid BM, Ugurbil K. 1982 . NMR chemical shift imaging in three dimensions. Proceedings of the National Academy of Sciences, USA 79 , 3523 –3526. Bürkle L, Hibberd JM, Quick P, Kühn C, Hirner B, Frommer WB. 1998 . The H + ‐sucrose cotransporter NtSUT1 is essential from sugar export from tobacco leaves. Plant Physiology 118 , 59 –68. Callaghan PT. 1991 . Principles of nuclear magnetic resonance microscopy . Oxford University Press. Callaghan PT, Clarke CJ, Forde LC. 1994 . Use of static and dynamic NMR microscopy to investigate the origins of contrast in images of biological tissues. Biophysical Chemistry 50 , 225 –235. Connelly A, Lohman JAB, Loughman BC, Quiquampoix H, Ratcliffe RG. 1987 . High resolution imaging of plant tissue by NMR. Journal of Experimental Botany 38 , 1713 –1723. de Graaf RA. 1998 . In vivo NMR Sectroscopy . Chichester: J Wiley and Sons. de Graaf RA, Dijkhuizen RM, Biessels GJ, Braun KPJ, Nicolay K. 2000 . In vivo glucose detection by homonuclear spectral editing. Magnetic Resonance in Medicine 43 , 621 –626. Farrar J, Pollock C, Gallagher J. 2000 . Sucrose and the integration of metabolism in vascular plants. Plant Science 154 , 1 –11. Fox GG, Ratcliffe RG, Robinson SA, Slade AP, Stewart GR. 1992 . Detection of 15 N‐labelled ammonium. In vivo 15 N NMR versus mass spectrometry. Journal of Magnetic Resonance 96 , 146 –143. Fricker MD, Oparka KJ. 1999 . Imaging techniques in plant transport: meeting review. Journal of Experimental Botany 50 , 1089 –1100. Fu H, Kim SY, Park WD. 1995 . High‐level tuber expression and sucrose inducibility of a potato sus4 sucrose synthase gene require 5′ and 3′ flanking sequences and the leader intron. The Plant Cell 7 , 1387 –1394. Heidenreich M, Köckenberger W, Kimmich R, Chandrakumar N, Bowtell R. 1998 . Investigation of carbohydrate metabolism and transport in castor bean seedlings by cyclic J cross polarization imaging and spectroscopy. Journal of Magnetic Resonance 132 , 109 –124. Heidenreich M, Köckenberger W, Kimmich R, Chandrakumar N, Bowtell R. 2001 . Proton detected 13 C magnetic resonance microscopy: in vivo transport of sucrose in Ricinus communis. Manuscript in preparation. Hoult DI, Richards RE. 1976 . The signal‐to‐noise ratio of the nuclear magnetic resonance experiment. Journal of Magnetic Resonance 24 , 71 –85. Jang JC, Sheen JS. 1997 . Sugar sensing in higher plants. Trends in Plant Science 2 , 208 –214. Katz JJ, Crespi HL. 1970 . Isotope effects in biological systems. In: Collins CJ, Bowman NS, eds. Isotope effects in chemical reactions . ACS Monograph 167, Van Nostrand Reinhold. Keltner JR, Wald LL, Ledden PJ, Chen Y‐CI, Matthews RT, Küstermann EHGK, Baker JR, Rosen BR, Jenkins BG. 1998 . A localized double‐quantum filter for the in vivo detection of brain glucose. Magnetic Resonance in Medicine 39 , 651 –656. Kimmich R. 1997 . NMR tomography, diffusometry, relaxometry . Berlin, Heidelberg: Springer Verlag. Koch KE. 1996 . Carbohydrate modulated gene expression in plants. Annual Reviews in Plant Physiology and Plant Molecular Biology 47 , 509 –540. Koch KE, Ying Z, Wu Y, Avigne WT. 2000 . Multiple paths of sugar‐sensing and a sugar/oxygen overlap for genes of sucrose and ethanol metabolism. Journal of Experimental Botany 51 , Special Issue, 417 –427. Köckenberger W, Pope J, Xia Y, Jeffrey K, Komor E, Callaghan PT. 1997 . A non‐invasive measurement of phloem and xylem water flow in castor bean seedlings by nuclear magnetic resonance micro‐imaging. Planta 201 , 53 –63. Koreleva OA, Farrar JF, Tomos AD, Pollock CJ. 1997 . Patterns of solute in individual mesophyll, bundle sheat and epidermal cells of barley leaves induced to accumulate carbohydrate. New Pythologist 136 , 97 –104. Kunze C, Kimmich R. 1994 . Proton‐detected 13 C imaging using cyclic J crosspolarization. Magnetic Resonance Imaging 12 , 805 –810. Lei H, Peeling J. 1999 . Simultaneous lactate editing and observation of other metabolites using a stimulated echo enhanced double‐quantum filter. Journal of Magnetic Resonance 137 , 215 –220. Link J, Seelig J. 1990 . Comparison of deuterium NMR imaging methods and application to plants. Journal of Magnetic Resonance 89 , 310 –330. Meininger M, Jakob PM, von Kienlin M, Koppler D, Bringmann G, Haase A. 1997 . Radial spectroscopic imaging. Journal of Magnetic Resonance 125 , 325 –331. Metzler A, Izquierdo M, Ziegler A, Köckenberger W, Komor E, von Kienlin M, Haase A, Decorps M. 1995 . Plant histochemistry by correlation peak imaging. Proceedings of the National Academy of Sciences, USA 92 , 11912 –11915. Metzler A, Köckenberger W, von Kienlin M, Komor E, Haase A. 1994 . Quantitative measurement of sucrose distribution in Ricinus communis seedlings by chemical shift microscopy. Journal of Magnetic Resonance B 105 , 249 –252. Olt S, Krotz E, Komor E, Rokitta M, Haase A. 2000 . N 23 and H 1  NMR micro‐imaging of intact plants. Journal of Magnetic Resonance 144 , 297 –304. Oparka KJ, Roberts AG, Boevink P, Santa Cruz S, Roberts L, Pradel KS, Imlau A, Kotlizky G, Sauer N, Epel B. 1999 . Simple, not branched, plasmodesmata allow the non‐specific trafficking of proteins in developing tobacco leaves. Cell 97 , 743 –754. Pope JM. 1992 . Application of chemical shift microscopy to non‐invasive histochemistry of plant materials. In: Blümich B, Kuhn W, eds. Magnetic resonance microscopy . Weinheim: VCH, 441 –457. Pope JM, Jonas D, Walker RR. 1993 . Applications of NMR micro‐imaging to the study of water, lipid and carbohydrate distribution in grape berries. Protoplasma 173 , 177 –186. Pope JM, Rumpel H, Kuhn W, Walker R, Leach D, Sarafis V. 1991 . Application of chemical‐shift‐selective NMR microscopy to the non‐invasive histochemistry of plant materials. Magnetic Resonance Imaging 9 , 357 –363. Pope JM, Walker RR, Kron T. 1992 . Artifacts in chemical shift selective imaging. Magnetic Resonance Imaging 10 , 695 –698. Ratcliffe RG, Roscher A. 1998 . Prospects for in vivo NMR methods in xenobiotic research in plants. Biodegradation 9 , 411 –422. Ratcliffe RG. 1994 . In vivo NMR studies of higher plants and algae. Advances in Botanical Research 20 , 44 –123. Roberts JKM. 2000 . NMR adventures in the metabolic labyrinth within plants. Trends in Plant Science 5 , 30 –34. Roitsch T. 2000 . Source‐sink regulations by sugar and stress. Current Opinion in Plant Biology 2 , 198 –206. Rokitta M, Peuke AD, Zimmermann U, Haase A. 1999 . Dynamic studies of phloem and xylem flow in fully differentiated plants using fast NMR micro‐imaging. Protoplasma 209 , 126 –131. Rumpel H, Pope JM. 1992 . The application of 3D chemical shift microscopy to non‐invasive histochemistry. Magnetic Resonance Imaging 10 , 187 –194. Sacchi GA, Cocucci M. 1992 . Effects of deuterium oxide on growth, proton extrusion, potassium influx and in vitro plasma membrane activities in maize root segments. Plant Physiology 100 , 1961 –1967. Scheenen TWJ, van Dusschoten D, de Jager PA, van As H. 2000 a . Quantification of water transport in plants with NMR imaging. Journal of Experimental Botany 51 , 1751 –1759. Scheenen TWJ, van Dusschoten D, de Jager PA, van As H. 2000 b . Microscopic displacement imaging with pulsed field gradient turbo spin‐echo NMR. Journal of Magnetic Resonance 142 , 207 –215. Schneider J, Fekete E, Weisser A, Neubauer S, von Kienlin M. 2000 . Reduced 1 H‐NMR visibility of creatine in isolated rat hearts. Magnetic Resonance in Medicine 43 , 497 –502. Schulze W, Frommer WB, Ward JM. 1999 . Transporters of ammonium, amino acids and peptides are expressed in pitchers of carnivorous plants Nephentes . The Plant Journal 17 , 637 –646. Slichter C. 1990 . Principle of magnetic resonance . Berlin, Heidelberg: Springer. Snaar JEM, van As H. 1992 a . A method for simultaneous measurement of NMR spin‐lattice and spin‐spin relaxation times in compartmentalized systems. Journal of Magnetic Resonance 99 , 139 –148. Snaar JEM, van As H. 1992 b . Probing water compartments and membrane permeability in plant cells by 1 H NMR relaxation measurement. Biophysical Journal 63 , 1654 –1658. Stadler R, Truernit E, Gahrtz M, Sauer N. 1999 . The AtSUC1 sucrose carrier may represent the osmotic driving force for anther dehiscence and pollen tube growth in Arabidopsis . The Plant Journal 19 , 269 –278. Tse TY, Spanswick RM, Jelinski LW. 1996 . Quantitative evaluation of NMR and MRI methods to measure sucrose concentration on plants. Protoplasma 194 , 54 –62. van der Weerd L, Ruttink T, Claessens M, Vergeldt F, van As H. 1998 . Osmotic stress effects on growth and cell water balance in maize and pearl millet. Journal of Experimental Botany 49 , Special Issue, 12. van Dijk, JE, Mehlkopf AF, Bovee WMMJ. 1992 . Comparison of double and zero quantum NMR editing techniques for in vivo use. NMR in Biomedicine 5 , 75 –86. Verscht J, Kalusche B, Köhler J, Köckenberger W, Metzler A, Haase A, Komor E. 1998 . The kinetics of sucrose concentration in the phloem of individual vascular bundles of the Ricinus communis seedling measured by nuclear magnetic resonance micro‐imaging. Planta 205 , 132 –139. van Dusschoten D, Moonen CTW, de Jager PA, van As H. 1996 . Unravelling diffusion constants in biological tissue by combining Carr‐Purcell‐Meiboom‐Gill Imaging and Pulsed Field Gradient NMR. Magnetic Resonance in Medicine 36 , 907 –913. von Kienlin M, Ziegler A, Le Fur Y, Rubin C, Decorps M, Remy C. 2000 . 2D‐spatial/2D‐spectral spectroscopic imaging of intracerebral gliomas in rat brain. Magnetic Resonance in Medicine 43 , 211 –219. Wolf K, van der Toorn A, Hartmann K, Schreiber L, Schwab W, Haase A, Bringmann G. 2000 . Metabolic monitoring in plants with double‐quantum filtered chemical shift imaging. Journal of Experimental Botany 51 , 2109 –2117. Xia Y, Jelinski LW. 1995 . Imaging low‐concentrated metabolites in the presence of a large background signal. Journal of Magnetic Resonance B 107 , 1 –9. Xia Y. 1996 . Contrast in NMR imaging and microscopy. Concepts in Magnetic Resonance 8 , 205 –225. Ziegler A, Metzler A, Köckenberger W, Izquirdo M, Komor E, Haase A, Decorps M, von Kienlin M. 1996 . Correlation peak imaging. Journal of Magnetic Resonance B 112 , 141 –150. © Society for Experimental Biology
TI - Nuclear magnetic resonance micro‐imaging in the investigation of plant cell metabolism
JF - Journal of Experimental Botany
DO - 10.1093/jxb/52.356.641
DA - 2001-04-01
UR - https://www.deepdyve.com/lp/oxford-university-press/nuclear-magnetic-resonance-micro-imaging-in-the-investigation-of-plant-O8k40ZJ0XR
SP - 641
VL - 52
IS - 356
DP - DeepDyve
ER -