A method is presented to measure the absolute concentration of intracellular Na+([Na+]i) in vivo by using interleaved 23Na- and 31P-nuclear magnetic resonance (NMR) spectroscopy and TmDOTP5− as shift reagent and chemical marker of tissue extracellular space (ECS). The technique was used to determine [Na+]iand relative ECS in livers of control rats (21 ± 3 and 0.11 ± 0.02 mM, respectively) and in rats exposed to carbon tetrachloride (103 ± 29 and 0.23 ± 0.03 mM, respectively). The NMR measurements were confirmed independently on excised tissue samples by using atomic absorption spectroscopy. The results confirm that TmDOTP5− can be used as a combined cation shift reagent and ECS marker, thereby allowing quantitation of [Na+]iin vivo by NMR.
- nuclear magnetic resonance
- atomic absorption spectrometry
the transmembraneNa+ gradient is used to drive several vital cellular processes, including cell volume regulation, cotransport of other ions and metabolites across cellular membranes against a concentration gradient, and transmission of nerve impulses. Any noninvasive method that can detect and quantitate abnormal intracellular Na+ concentration ([Na+]i) in vivo would be useful in understanding the role of Na+ gradients in various pathophysiological states. Ion-selective microelectrodes, electron microprobe X-ray analysis, fluorescent dyes, and whole tissue analysis have all been used to measure tissue Na+ (8, 11, 21), but none of these methods offers the potential of providing quantitative data noninvasively in vivo. Nuclear magnetic resonance (NMR) spectroscopy with the aid of a hyperfine shift reagent (SR) offers such potential.
To quantify Na+ in the intracellular compartment by NMR, the intra- and extracellular Na+ signals must be clearly differentiated and the relative intracellular (rICS) and extracellular (rECS) spaces must be either known or determined. The hyperfine SR, thulium-1,4,7,10-tetraazacyclododecane-1,4,7,10-tetrakis(methylene phosphonate) (TmDOTP5−), satisfies both requirements. We have previously shown that TmDOTP5− produces baseline-resolved signals in vivo and is well tolerated by animals (2-4, 23-26, 28). In an accompanying paper (17), we have also shown that the distribution of TmDOTP5− in the extracellular space (ECS) is identical to a standard ECS marker, CoEDTA−. In the present work, we used a combination of31P- and23Na-NMR to evaluate relative extracellular tissue space and [Na+]iin livers from control rats and from rats exposed to the hepatotoxin, carbon tetrachloride (CCl4). Atomic absorption spectroscopy (AAS) analyses of excised tissues was used to confirm the NMR measurements.
Animal preparation and infusion protocol. Animal protocols were approved by the Animal Care and Research Advisory Committee of University of Texas Southwestern Medical Center. Male Sprague-Dawley rats (350–450 g) were provided with food and water ad libitum. Two groups were studied; one control group of healthy rats and one group of animals after oral ingestion of CCl4, a toxin known to disrupt the transmembrane Na+gradient (6, 7). The CCl4 group was given an oral dose of 2.5 ml/kg body weight of CCl4 mixed with 5 ml/kg of corn oil 24 h before each NMR experiment while the controls were given 7.5 ml/kg body weight of corn oil. The CCl4-treated animals failed to eat during that 24-h period and lost ∼3–4% body weight between the time of treatment and the NMR experiment.
In preparation for the NMR experiment, rats were initially anesthetized by intramuscular injection of a 0.5-ml mixture of ketamine (85 mg/ml) and xylazine (15 mg/ml). Both jugular veins and a carotid artery were cannulated through a midline neck incision. One jugular vein was used to maintain the anesthesia (2 mg/ml ketamine, 0.25 mg/ml xylazine in saline) at a rate of 2–3 ml/h; the other was used to infuse the SR. A catheter placed in the carotid artery and connected to a Gould transducer was used to measure pulse pressure and heart rate on a Coulbourn polygraph. A tracheotomy was performed, and the respiration rate was maintained at 90 breaths/min and a tidal volume of 2–3 cm3 by using a Harvard rodent ventilator model 683 (Harvard Apparatus, South Natick, MA). A laparotomy was performed, and the liver was exposed through a subcostal incision.
Animals were positioned supine on a recirculating water heating pad in a specially constructed cradle and maintained at 37°C. A 2.3-cm-diameter surface coil dual tuned to 53 MHz for23Na and 81 MHz for31P was placed directly over the liver. A single layer of Saran wrap was placed between the coil and the liver to avoid wetting of the coil from body fluids. A small glass bulb (∼100 μl) containing 138.75 mM Na+ and 20 mM TmDOTP5− at pH 5.0 was placed in a holder in the center of the surface coil and used as an external concentration standard. This solution provided external reference signals for both the23Na- and31P-NMR spectra. The23Na signal [shifted by ∼10 parts/million (ppm)] was used as the reference for tissue23Na. The31P signal from the four equivalent phosphonate groups of TmDOTP5− [shifted to about −315 ppm, adjusted by lowering the pH of the standard (29)] was used as a reference for the in vivo tissue TmDOTP5− signal at about −330 ppm.
A stock solution of 80 mM Na4HTmDOTP (Magnetic Resonance Solutions, Dallas, TX) was prepared in deionized water. The SR was initially infused at a rate of 2 ml/h for 6 min. The rate was incrementally increased by 2 ml/h to a maximum of 8 ml/h (6-min duration at each rate) and maintained at this rate for 20–30 min. After a chemical shift separation of ∼5–7 ppm was achieved between the intra- and extracellular Na+ resonances, the infusion rate was reduced to 3 ml/h and periodically adjusted to maintain a constant shift.
Magnetic resonance spectroscopy data collection. NMR data were collected on a 4.7-T, 40-cm CSI Omega Spectrometer (Bruker Instruments, Fremont, CA) by using the dual-tuned surface coil. The magnet was shimmed on the23Na signal. A Na+ line width of 35–50 Hz was typical after shimming. Both23Na and31P spectra were collected in an interleaved fashion during TmDOTP5− infusion. For both nuclei, the nominal 90 excitation pulse at the coil center was 35–40 us. For 23Na, 2,048 complex data points were collected over a sweep width of 10,000 Hz with the preamplifier filter turned off. For31P, 2,048 complex data points were collected over a sweep width of 20,000 Hz with the preamplifier filter on. Switching of parameters and spectrometer frequency was automatically executed by using a script. Cyclops phase cycling was used for both nuclei. The recycle times for23Na and31P were 210 and 100 ms, respectively (the T1 of the phosphonate resonance of TmDOTP5− is ∼15–25 ms). The data were initially collected in blocks of 2 acquisitions for23Na and 16 acquisitions for31P. Thirty-two blocks were summed, yielding a total of 6423Na acquisitions and 51231P acquisitions. Interleaved23Na and31P spectra were collected throughout infusion of SR and for another 20 min after a steady-state hyperfine shift was achieved in the extracellular Na+ resonance. The free induction decays were Fourier transformed after baseline correction and multiplication by a single exponential corresponding to a 10-Hz line broadening for 23Na and a 40-Hz line broadening for 31P. The resonance areas were fit to Lorentzian lines by using a nonlinear Marquart-Levenberg optimization algorithm (18). Peak area ratios between the tissue and the corresponding reference bulb signals were then calculated, and tissue concentrations were determined from the 23Na and31P reference calibration curves as described below. The results from five consecutive steady-state spectra were averaged.
Sample preparation and analysis by AAS. At the conclusion of each in vivo NMR experiment, ∼0.2 ml of blood was withdrawn from the carotid artery, and that portion of liver positioned directly beneath the surface-coil (approximate size of 1 lobe) was removed. The livers were immediately freeze-clamped by using aluminum tongs precooled in liquid nitrogen. The frozen tissue was weighed in a nitrogen atmosphere to exclude water, dried overnight at 60°C to constant weight, and reweighed to establish the relative dry weight (rDW).
Blood and tissue samples from both control rats and from rats pretreated with CCl4 were prepared for AAS analysis by using standard procedures (17, 20). Briefly, liver samples were digested in 2 ml of an acid mixture containing 5:1:1 concentrated nitric, sulfuric, and perchloric acids, respectively. The blood samples were centrifuged to remove red blood cells (RBC) from plasma. A Varian SpectrAA-5 atomic absorption spectrometer was used to analyze the liver and plasma samples for Tm and Na at 372.2 and 330.2 nm, respectively. Tm was measured by using a nitrous oxide-acetylene flame, whereas Na was measured by using an air-acetylene flame.
Calibration of 23Na and31P signals.
Phantom experiments were performed to calibrate the23Na and31P signals from the reference bulb before the in vivo experiments. Nine sealed plastic bags filled with 20–70 mM NaCl were used to calibrate the23Na signal, and 13 bags filled with 0.3–3.0 mM TmDOTP5− and 60 mM NaCl were used to calibrate the 31P signal. The 31P signal was also calibrated with another set of six bags containing 0.5–3.0 mM TmDOTP5− and 154 mM NaCl to determine the effect of coil loading. The bags were positioned under the dual-tuned surface coil, and23Na or31P spectra were collected and processed by using the same parameters as in the in vivo experiments. The T1 values of all31P (from TmDOTP5−) and23Na signals from the phantoms and from liver in vivo were at least a factor of three to four times shorter than the repetition time used in each NMR experiment. The calibration curves of Fig. 1,A andB, were established by plotting the ratio of the peak areas from either Na+ or TmDOTP5− in the bags, relative to the reference standard in the bulb vs. the known concentration of Na+ or TmDOTP5− in the bag. A least squares linear regression of the data yielded calibration slopes and intercepts of 0.116 ± 0.002 and 0.12 ± 0.07, respectively, for23Na (r 2 = 0.998) and 1.44 ± 0.06 and −0.22 ± 0.09, respectively, for31P (r 2 = 0.976). The31P calibration curve obtained from the phantoms containing 150 mM NaCl was not significantly different from the curve obtained from the 60 mM NaCl phantoms.
Calibration of 23Na chemical shift for measuring extracellular [TmDOTP5−] in vivo.
To determine the relationship between23Na hyperfine shift and plasma [TmDOTP5−], a calibration curve was obtained in parallel bench experiments using control rats. Rats were anesthetized and infused with the SR as described above. At various times during infusion of the SR, blood samples (∼0.2 ml) were withdrawn, and23Na-NMR spectra were recorded to determine the hyperfine shift of extracellular plasma Na+ by using the RBC intracellular Na+ as an internal chemical shift standard. We assumed that any bulk magnetic susceptibility differences between extracellular plasma Na+and intracellular Na+ introduced by TmDOTP5− were small and comparable in 23Na spectra of in vivo liver and isolated whole blood (1, 16). Total plasma [Tm3+] was determined by AAS after centrifugation and removal of the RBC. The calibration curve in Fig. 1 C was constructed by plotting the hyperfine shift of extracellular Na+ in spectra of RBC vs. plasma [Tm3+] as determined by AAS. Least squares linear regression of these data gave a calibration slope of 1.06 ± 0.07 and an intercept of −0.4 ± 0.2 (r 2 = 0.922). These data were used to estimate extracellular [TmDOTP5−] from the in vivo 23Na spectra. This assumes, of course, that extracellular [TmDOTP5−] equals plasma [Tm3+]. The calibration curve has a small, nonzero intercept because of differential binding of Ca2+ to TmDOTP5− at lower vs. higher concentrations of SR (19).
Calculations. The rECS was determined by dividing the concentration of TmDOTP5− in tissue ([TmDOTP5−]tissue) by its concentration in the ECS (22) ([TmDOTP5−]extwas assumed equal to plasma [TmDOTP5−]). Equation 1[TmDOTP5−]tissuewas determined using the relation Equation 2where and are the areas under the31P resonance from tissue TmDOTP5− and reference bulb TmDOTP5−, respectively. [TmDOTP5−]extwas determined from the chemical shift separation (in ppm) between the extracellular and intracellular Na+ resonances in the in vivo23Na spectrum and the calibration curve shown in Fig. 1 C (16). The rICS was determined from the rECS after subtracting the contribution from dry tissue Equation 3The [Na+]iin cellular water was calculated from the peak area of the resonance in23Na spectra and the rICS by using the following equation Equation 4where and are the areas under the intracellular and reference peaks in the23Na spectra, respectively.
rECS and [Na+]iwere independently determined by AAS by using methods outlined in the accompanying article (17). All results are reported as means ± SE.
Representative in vivo 23Na and31P spectra of rat liver before and after infusion of TmDOTP5− are shown in Fig.2. Before infusion of SR, the intra- and extracellular Na+ resonances were coincident (set to 0 ppm). The23Na signal at 10 ppm originates from Na+ in the reference bulb. The signal at −315 ppm in the31P-NMR spectrum acquired before infusion of SR was from the TmDOTP5− in the same reference bulb. After infusion of SR, the extracellular Na+ signal shifted downfield ∼5 ppm, leaving the intracellular Na+signal at 0 ppm. The chemical shift separation between intra- and extracellular 23Na resonances and the calibration curve shown in Fig. 1 Cprovided a direct measurement of [TmDOTP5−]ext. The 31P spectrum showed two resolved resonances, one at −315 ppm from the reference bulb and a second near −330 ppm from TmDOTP5− in the tissue. The ratio of these resonance areas and the calibration curve shown in Fig.1 B were used to evaluate [TmDOTP5−]tissue.
The quantitative data obtained by NMR and AAS analyses from four control animals are summarized in Table1. The average [TmDOTP5−]extas detected by NMR was 5.3 ± 0.5 mM, whereas the average of Na+ tissue concentration ([Na+]tissue) and [TmDOTP5−]tissuewere 34 ± 1 and 0.6 ± 0.1 mM, respectively. These values were not significantly different from those obtained independently by AAS analysis of the same tissues. The values of rECS, rICS, and [Na+]i, determined from this raw data are summarized in Table2. Again, the values of rECS and [Na+]idetermined by NMR vs. AAS analysis were not significantly different. This indicates that TmDOTP5−can be used to quantitatively measure [Na+]iin intact animals by using a combination of23Na- and31P-NMR spectroscopy. In addition, the close agreement between the two methods indicates that liver is essentially fully visible by23Na-NMR.
Figure 3 compares typical23Na and31P spectra of liver from a control and a CCl4-treated rat after infusion of TmDOTP5−. Clearly, the intracellular Na+resonance was much larger in the spectrum of the CCl4-treated rat. The average / peak area ratio was about threefold higher in animals treated with CCl4 compared with controls. The extracellular 23Na signals in the two spectra shown had similar intensities, but the hyperfine shift was slightly larger in the spectrum from the control animal, consistent with a slightly higher [TmDOTP5−]extin this particular control animal experiment. A higher tissue TmDOTP5− was also evidenced by the larger tissue SR resonance in the31P spectrum of the control animal compared with that of CCl4-treated animal.
Table 2 summarizes the results of NMR and AAS measurements from control and CCl4-treated rats. There was an excellent agreement between values determined by NMR and AAS in all cases. There was a significant decrease in rDW of livers from animals exposed to CCl4 compared with controls (0.27 ± 0.01 and 0.33 ± 0.01, respectively,P < 0.05), and rECS as measured by either NMR or AAS was significantly higher in the CCl4 population (0.23 ± 0.03) than in controls (0.11 ± 0.02; P< 0.05). Both NMR and AAS detected significantly higher [Na+]iin the CCl4-treated rats compared with the control rats (103 ± 29 and 21 ± 3 mM, respectively, by NMR; P < 0.05 vs. 119 ± 21 and 23 ± 2 mM, respectively, by AAS; P< 0.05). Extracellular Na+concentration ([Na+]e), however, was not significantly different between the two groups, as measured by either NMR (160 ± 17 and 174 ± 6 mM, respectively,P < 0.05) or AAS (164 ± 3 and 163 ± 4 mM, respectively, P < 0.05). In summary, livers from animals exposed to CCl4 24 h before observation were characterized by a lower rDW and a larger rECS, both characteristic of tissue edema. Liver [Na+]iwas remarkably higher in the CCl4-treated animals as well, indicating damage at the cellular level.
Absolute quantitation of [Na+]iby NMR requires separation of the intra- and extracellular23Na resonances and a determination of rECS. In this study, we have used TmDOTP5− in a combined role, both as SR and ECS marker, thereby allowing quantitation of intracellular Na+ in vivo with the addition of only one exogenous agent. In the accompanying paper (17), we have validated the use of TmDOTP5− as an ECS marker by showing that both TmDOTP5−and the pharmacologically inert CoEDTA− give identical values for rECS. Because the four phosphonate groups in TmDOTP5− can be easily detected by 31P-NMR (24) and the extracellular Na+ hyperfine shift is proportional to [TmDOTP5−]ext, rECS can be determined directly from a single in vivo31P and23Na spectrum. As TmDOTP5− can also be detected in the tissue by 1H-NMR (29), it should also be possible to use combined23Na- and1H-NMR spectroscopy to evaluate [Na+]iexactly as demonstrated here.
Liver [Na+]i, as determined here by NMR, agreed closely with values determined by AAS (Table 2) and with previously reported values determined by using a variety of methods (5, 12, 14, 27). Rats treated with CCl4, a classic hepatotoxin that causes cellular damage by generating free radicals in the liver (13), had substantially higher [Na+]ias anticipated. Free radicals are known to attack the cell membranes, leading to a nonspecific increase in permeability that utimately results in an increase in [Na+]i(6, 7). Our data indicated that [Na+]iin liver had increased by about fivefold 24 h after exposure to CCl4, in agreement with a previous report by Macknight (15). The good agreement between NMR- and AAS-determined [Na+]ialso indicates that in the liver is fully visible by 23Na-NMR, both in control animals and in animals with a severely damaged liver. This is in complete agreement with a recent report indicating that in perfused liver is 100% visible by 23Na-NMR (10).
The quantitative NMR method reported here may be a more favorable technique for measuring [Na+]ithan destructive chemical methods, such as AAS, in many situations. Obviously, repetitive measurements in the same animal cannot be performed with AAS, whereas the NMR method could potentially be done sequentially during periods of stress that may alter ion homeostasis or during a pharmaceutical intervention. In addition, the NMR method is better suited for studies where blood flow to the tissue of interest may become restricted. The AAS method requires the assumption that chemical markers in plasma are always in equilibrium with chemical markers in the interstitial space (22). This assumption may not be valid under many experimental conditions, such as during ischemia when flow is restricted. The NMR method, however, measures TmDOTP5− and Na+ directly in tissue, so it is not necessary for the SR in plasma to be in equilibrium with interstitial SR once an appropriate calibration curve has been established. Thus the NMR method may ultimately prove more versatile compared with techniques that use plasma samples for determining rECS.
One of the difficulties associated with the use of a surface coil for obtaining quantitative NMR data is the uncertainty of the volume of the tissue being interrogated. To avoid detecting23Na signals from tissue beneath or surrounding the liver, we purposely used a small-diameter radio-frequency coil so that the entire sensitive volume of the coil was within the liver. Care was also taken to make sure that the entire active volume of the coil was also contained within the calibration standards to avoid variations in filling of the sensitive volume. A second consideration is that the coil may couple strongly to samples with high ionic content, and thus its sensitivity could vary from sample to sample (9). To check this, we used two different sets of TmDOTP5− phantoms containing different levels of matrix Na+. Because the two sets of phantoms, one containing 60 mM NaCl and the other containing 150 mM NaCl, yielded identical calibration slopes, any changes in coil loading or sensitivity were considered insignificant. In addition, we found no change in the tune, match, and quality factor of the coil using calibration standards of various Na+ concentrations or with the in vivo livers. The calibration curves of Fig. 1,A andB, have nonzero intercepts due to the low signal-to-noise ratios in the spectra of the lower concentration standards. Figure 1 C has a nonzero intercept due to partial overlap of the plasma Na+ and intracellular RBC Na+ resonances in the spectra collected during early infusion of TmDOTP5−. These curves could have been fit by using a nonlinear function, but all experimental data tended to fall in the central region of the calibration curves, thereby minimizing any small error in the uncertainty of intercepts of these plots. A fitting of these data to various higher order equations did not alter the quantitative results significantly.
The data used to generate the calibration curve of Figure1 C were obtained by periodically withdrawing blood from animals during infusion of SR. An alternative way to obtain such data might be to incrementally add SR to an isolated blood sample and evaluate the 23Na hyperfine shift after each addition. However, this approach would likely yield inaccurate results because TmDOTP5− is known to form strong ion-pair interactions with Ca2+ and Mg2+ in addition to Na+ (19). We have previously shown that total plasma Ca2+concentration increases during infusion of TmDOTP5− in vivo, whereas ionized Ca2+ concentration remains relatively constant (3). Because total Ca2+ concentration and Mg2+ concentration are fixed in an isolated blood sample, the levels of uncomplexed Ca2+ and Mg2+ would gradually decrease on addition of SR, ultimately producing larger hyperfine shifts in the extracellular Na+ resonance of blood than that measured in vivo for an equivalent amount of infused TmDOTP5−. Thus it is necessary to perform the calibration experiments by using intact animals to obtain valid measurements of [TmDOTP5−]extfrom the hyperfine shift difference between intra- and extracellular23Na resonances. The excellent agreement between rECS, [Na+]i, and [Na+]evalues determined by NMR vs. AAS (Table 2) indicates that the calibration curves derived from control rats are also applicable to CCl4-treated animals, even though the extracellular ionic status in the later animals may differ somewhat. Although [Na+]ewas no different between the two groups, it is quite likely that plasma Ca2+ concentration was different because less SR was required to resolve the intra- and extracellular23Na resonances in the CCl4-treated group.
In conclusion, these experiments show that TmDOTP5− may be used as an effective 23Na SR and an ECS marker for evaluation of [Na+]iin vivo. [Na+]idetermined here by NMR compares favorably to values obtained by AAS and other methods (5, 12, 14, 27). Application of this technique to an established CCl4 hepatic injury model designed to increase [Na+]iyielded the expected results. We conclude that the NMR method described here may be generally useful for obtaining repeated measurements of rECS and [Na+]iin vivo, provided that TmDOTP5− ultimately proves to have low chronic toxicity and that its tendency to accumulate on bone (17) does not prove detrimental.
This work was supported in part by a grant from the Whitaker Foundation (to N. Bansal); by Grant AT-584 from the Robert A. Welch Foundation (to A. D. Sherry); by Grants HL-54574 (to N. Bansal), HL-34557 (to A. D. Sherry), and RR-02584 (to C. R. Malloy) from the National Institutes of Health; and by a Veterans Affairs Clinical Investigator Award to C. R. Malloy.
Address for reprint requests: N. Bansal, Dept. of Radiology, B1 Stellar-Chance Laboratories, Univ. of Pennsylvania Medical Center, 422 Curie Blvd., Philadelphia, PA 19104-6100 (E-mail:).
A portion of this work was presented at the Third Annual Meeting of the Society of Magnetic Resonance, August 19–25, 1995, Nice, France.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.
- Copyright © 1998 the American Physiological Society