Cell surface involvement in cancer metastasis: an NMR study

NMR spectroscopy is one of the few techniques which has the sensitivity to detect subtle changes to the surface chemistry of cells. It has previously been demonstrated that high resolution ¹H NMR methods can distinguish tumour cells with the capacity to metastasise and this information appears to arise from a type of proteolipid in or attached to the plasma membrane. Here we report that the ¹H NMR signal, which we have used to identify metastatic cells in rat tumours, is significantly reduced in intensity after cultured cells are treated with trypsin/EDTA. The long T₂ relaxation value (⪢ 350 ms) observed in metastatic cells is absent after enzyme treatment. 2D scalar correlated NMR (COSY) spectra of these treated cells show that a cross peak normally associated with malignancy and metastatic disease is markedly reduced. These findings indicate that the plasma membrane lipid particle which generates the high resolution spectrum is directly affected by trypsin/EDTA. Alterations to the cell surface properties were also demonstrated in vivo since reduced numbers of metastases were observed in animals injected with enzyme-treated cells. The correlation between the absence of a long T₂ relaxation value and the diminished numbers of metastases in animals suggests that the plasma membrane particle is involved in the metastatic process.     

Lipid domain in cancer cell plasma membrane shown by 1H NMR to be similar to a lipoprotein

Human blood lipoproteins have been characterised by 1H NMR methods and chemical analysis, and comparisons made with the properties of the triglyceride-rich plasma membrane domain found in cancer cells. By means of selective and non-selective T1 experiments, the lipids in HDL and LDL are shown to be in diffusive exchange. In contrast, the lipids of chylomicra and VLDL do not exhibit lipid diffusion, and therefore resemble the neutral lipids of cancer cell plasma membranes. 2D scalar correlated NMR (COSY) spectra of cancer cells or solid tumours are similar to those obtained from VLDL and LDL. The long T2 relaxation value observed for neutral lipid methylenes in metastatic cancer cells (greater than 300 ms) was not observed for any of the 4 lipoproteins studied. None of the lipoprotein classes gave a T2 longer than 250 ms.     

Adenosine triphosphate compartmentation in the rat heart: a 31P spin-lattice relaxation study

Longitudinal relaxation times (T1) of phosphorus compounds in the perfused rat heart and erythrocytes were measured using the 31P Driven-Equilibrium Single-Pulse Observation of T1 relaxation (DESPOT) method at 33 degrees C. Both creatine phosphate in the heart and the three phosphate groups of adenosine triphosphate (ATP) in erythrocytes showed single-exponential relaxation. The three phosphate groups of ATP in the heart, however, had two T1 components. The T1 values of the short and the long T1 components of the beta-phosphate of ATP were ca. 0.4 and 14 s, respectively. The fraction with the long T1 represented ca. 30% of the total ATP content. These results suggested that there were two major pools of intracellular ATP in the rat heart which could be determined by 31P NMR spectroscopy.     

23Na and flame photometric studies of the NMR visibility of sodium in rat muscle.

23Na nuclear magnetic resonance spectroscopy (NMR) is increasingly being used to study Na+ gradients and fluxes in biological tissues. However, the quantitative aspects of 23Na NMR applied to living systems remain controversial. This paper compares sodium concentrations determined by 23Na NMR in intact rat hindlimb (n = 8) and excised rat gastrocnemius muscle (n = 4) with those obtained by flame photometric methods. In both types of samples, 90% of the sodium measured by flame photometry was found to be NMR-visible. This is much higher than previously reported values. The NMR measurements for intact hindlimb correlated linearly with the flame photometric measurements, implying that one pool of sodium, predominantly extracellular, is 100% visible. From measurements on excised muscle, in which extracellular space is more clearly defined, the NMR visibility of intracellular Na+ was calculated to be 70%, assuming an extracellular space of 12% of the total tissue water volume and an extracellular NMR visibility of 100%. 23Na transverse relaxation measurements were carried out using a Hahn spin echo on both intact hindlimb (n = 1) and excised muscle (n = 2) samples. These showed relaxation curves that could each be described adequately using two relaxation times. The rapidly relaxing component showed a T2 value of 3-4 ms and the slowly relaxing component a T2 of 21-37 ms. A spin lattice relaxation (T1) measurement on intact hindlimb yielded a value of 51 ms. These relatively long relaxation times show that the quadrupolar relaxation effect of Na+ complexing to large macromolecules or being otherwise motionally restricted is relatively weak. This is consistent with the high NMR visibilities reported here.     

Proton magnetic relaxation time of normal and tumor cells

A proton NMR analysis of an in vitro culture of cells in heavy water has been made. The relaxation times of L-strain cells 929, He-La, transformed and normal embryonic human cells, C3H mice and isolated Yoshida sarcoma tumour cells, as well as of Yoshida sarcoma tumour tissue were determined. It turned out that spin-lattice (T1) and spin-spin (T2) relaxation times are characteristic of every cell and fairly different from those of corresponding tissues, which may be used for NMR identification of cells (NMR cytology). Furthermore, it has incontestably been proved that there is an ordered water fraction of cells, which is very slowly exchanged with surrounding heavy water.     

Selection of highly metastatic rat MTLn2 mammary adenocarcinoma cell variants using in vitro growth response to transferrin

We previously found that the proliferative response to transferrin and the expression of transferrin receptors (TfR) on the cell surface of various rat 13762NF mammary adenocarcinoma cell sublines correlated with their spontaneous metastatic capability. To further assess the involvement of transferrin and TfR in metastasis, transferrin-responsive cells were selected from the poorly-metastatic, low-transfferin responsive 13762NF MTLn2 subline. When maintained in low serum (0.3%) conditions, MTLn2 cells failed to survive. However, if like medium was supplemented with 0.5 μmg/ml rat transferrin, some colonies emerged, presumably due to their ability to proliferate in response to the added transferrin. The surviving cells were expanded and exposed to ten or 20 similar cycles of transferrin growth selection to obtain the sublines MTLn2-Tf10 and MTLn2-Tf20, respectively. The MTLn2-Tf20 cells proliferated in response to transferrin at a rate similar to that of the high metastatic 13762NF sublines. Using immunofluorescent staining, Scatchard analysis, and affinity isolation of TfR, we discovered that the MTLn2-Tf20 cells had 5 to 6 times more TfR than did the parental MTLn2 line. When injected into the mammary fat pads of rats, the MTLn2-Tf20 line metastasized to the axillary lymph node in seven out of ten animals and to the lungs in six out of ten (median number = 13). No metastases were seen in the MTLn2 parental line. The MTLn2-Tf10 cells showed intermediate properties compared with the MTLn2 and MTLn2-Tf20 cells. The results indicate that variant cells with a high response to transferrin may be more metastatic than the bulk cells in a poorly metastatic population. The selection of cells with high levels of TfR and a higher proliferative response to transferrin results in sublines with greater potentials for spontaneous metastasis. J. Cell. Physiol. 174:48–57, 1998. © 1998 Wiley-Liss, Inc.     

NMR spin-lattice relaxation times of intracellular Na-23 on rat livers and related lipid peroxidation following CCl4 intoxication

Liver tissues were isolated from rats acutely intoxicated with carbon tetrachloride, and Na-23 NMR signals were analyzed to investigate the T1 relaxation times of intracellular sodium ions under pathological conditions in presence of the paramagnetic shift reagent (dysprosium tripolyphosphate). We studied the significant increase of T1 found in CCl4 treated rats with respect to controls, which was elsewhere demonstrated as being independent of cell necrosis. Evidence is given that neither fat accumulation nor proliferative processes affect the observed T1 lengthening. When T1 relaxation times were measured in the liver of vitamin E treated rats subsequently intoxicated with carbon tetrachloride, a significative shortening of T1 with respect to CCl4-intoxicated rats was observed. These results were discussed in terms of the antioxidant action exerted by vitamin E, taking into account that peroxidation of microsomal lipids is the key factor in the process of carbon tetrachloride induced liver injury. Furthermore, the observed T1 changes were discussed in terms of the interactions of Na+ with cell membranes and/or the occurrence of viscosity changes.     

New Insights on Human Skeletal Muscle Tissue Compartments Revealed by In Vivo T2 NMR Relaxometry

The spin-spin (T2) relaxation of ¹H-NMR signals in human skeletal muscle has been previously hypothesized to reveal information about myowater compartmentation. Although experimental support has been provided, no consensus has yet emerged concerning the attribution of specific anatomical compartments to the observed T2 components. Potential application of a noninvasive tool that might offer such information urges the quest for a definitive answer to this question. The purpose of this work was to obtain new information that might help elucidate the mechanism of T2 distribution in muscle. To do so, in vivo T2 relaxation data was acquired from the soleus of eight healthy volunteers using a localized Carr-Purcell-Meiboom-Gill technique. Each acquisition contained 1000 echoes with an interecho spacing of 1 ms. Data were acquired from each subject under different vascular filling preparations expected to change exclusively the extracellular water fraction. Two exponential components were systematically observed: an intermediate component (T2 ∼ 32 ms) and a long component (100 < T2 < 210 ms). The relative fraction and T2 value characterizing the long component systematically increased after progressive augmentation of extracellular water volume. Characteristic relaxation behavior for each vascular filling condition was analyzed with a two-site exchange model and a three-site two-exchange model. We show that a two-site exchange model can only predict the observations for small exchange rates, much more representative of transendothelial than transcytolemmal exchange regimes. The three-site two-exchange model representing the intracellular, interstitial, and vascular spaces was capable of precisely predicting the observations for realistic transcytolemmal and transendothelial exchange rates. The estimated intrinsic relative fractions of each of these compartments corroborate with estimations from previous works and strongly suggest that the T2 relaxation from water within the intracellular and interstitial spaces is described by the intermediate component, whereas the long component represents water within the vascular space.     

31P NMR relaxation measurements of the phosphate backbone of a double stranded hexadeoxynucleotide in solution: determination of the chemical shift anisotropy

The five phosphates of the deoxynucleotide d(CpGpTpApCpG)₂ have been assigned by two-dimensional heteronuclear NMR spectroscopy. The chemical shift anisotropy and correlation time of each phosphate group has been determined from measurements of the spin-lattice, spin-spin relaxation rate constants and the ³¹P-{¹H} nuclear Overhauser enhancement (NOE) at three magnetic field strengths (4.7 T, 9.4 T, and 11.75 T) and two temperatures (288 K and 298 K). As expected, the relaxation data require two mechanisms to account for the observed rate constants, i.e. dipole-dipole and chemical shift anisotropy. At 9.4 T and 11.75 T, the latter mechanism dominates the relaxation, leading to insignificant NOE intensities. The correlation time, chemical shift anisotropy and effective P-H distance were obtained from least-squares fitting to the data. Comparison of the fitted value for the correlation time with that obtained from ¹H measurements shows that the molecule behaves essentially as rigid rotor on the nanosecond timescale. Large amplitude motions observed in long segments of DNA are due to bending motions that do not contribute significantly to relaxation in short oligonucleotides.Abbreviations CSA chemical shift anisotropy - NOE nuclear Overhauser enhancement Offprint requests to: A. N. Lane     

Role of 'cell type' in proton relaxation in normal and neoplastic lymph nodes characterized by pulsed NMR spectroscopy

Pulsed NMR studies have been undertaken on malignant lymphomas. It has been observed that water proton spin-lattice relaxation times of lymph node tissues show increase in lymphnodes of Hodgkin's and Non-Hodgkin's lymphoma as compared to those in normal subjects. The T1 values of normal lymphnodes showed a range of 200-300 msec, while the metastatic lymphnodes showed a range of 400-600 msec at 20 MHz. These studies have brought out the importance of histopathological significance and the role of 'cell type' and biomolecules as a factor influencing water proton relaxation times. Further the relevance of the present in vitro studies to Magnetic Resonance Imaging of ex vivo images of normal and metastatic lymphnodes has become evident from some recent studies reported in normal and afflicted lymphnodes.     

Relationship of biochemical and morphological changes in rat thyroid and proton spin-relaxation of the tissue water.

The effect was studied of biochemical and morphological changes induced by antithyroid substances (PTU, C10(-4)) on proton spin-relaxation properties of rat thyroid gland. It was found that thyroid stimulated by PTU (0.05%) or C10(-4) (1.0%) exhibit marked morphological changes (hyperplasia and epithelial hypertrophy) with alteration of the soluble iodoprotein pattern (content and composition.). Both relaxation times spin-lattice (T1) and spin-spin (T2) were increasing with the lenght of treatment with antithyroid drugs. Reversibility of the process was noted in accordance with biochemical and morphological data. The relaxation rate (formula: see text) for thyroid tissue water was in positive correlation with the suluble protein concentration and particularly with the TG content in the gland. There was no difference in relaxation times between normal thyroid and gland of rats treated chronically with excess iodide. The observed difference in T1 between normal glands and glands of PTU,-C10(-4)--treated rats was comparable with that found in cases of human thyroid cancer. This finding is of importance when the diagnostic potential of NMR in the detection of malignancy is considered. In conclusion, a strong correlation was found between microstructural and biochemical changes of the thyroid gland and proton magnetic relaxation of tissue water. The striking difference between the proton spin-relaxation times in normal and in goiter thyroid glands of rats suggests that pulsed NMR spectroscopy could be a method for evaluation of some disturbances in thyroid gland.     

New Insights on Human Skeletal Muscle Tissue Compartments Revealed by In?Vivo T2 NMR Relaxometry

The spin-spin (T2) relaxation of ¹H-NMR signals in human skeletal muscle has been previously hypothesized to reveal information about myowater compartmentation. Although experimental support has been provided, no consensus has yet emerged concerning the attribution of specific anatomical compartments to the observed T2 components. Potential application of a noninvasive tool that might offer such information urges the quest for a definitive answer to this question. The purpose of this work was to obtain new information that might help elucidate the mechanism of T2 distribution in muscle. To do so, in vivo T2 relaxation data was acquired from the soleus of eight healthy volunteers using a localized Carr-Purcell-Meiboom-Gill technique. Each acquisition contained 1000 echoes with an interecho spacing of 1 ms. Data were acquired from each subject under different vascular filling preparations expected to change exclusively the extracellular water fraction. Two exponential components were systematically observed: an intermediate component (T2 ∼ 32 ms) and a long component (100 < T2 < 210 ms). The relative fraction and T2 value characterizing the long component systematically increased after progressive augmentation of extracellular water volume. Characteristic relaxation behavior for each vascular filling condition was analyzed with a two-site exchange model and a three-site two-exchange model. We show that a two-site exchange model can only predict the observations for small exchange rates, much more representative of transendothelial than transcytolemmal exchange regimes. The three-site two-exchange model representing the intracellular, interstitial, and vascular spaces was capable of precisely predicting the observations for realistic transcytolemmal and transendothelial exchange rates. The estimated intrinsic relative fractions of each of these compartments corroborate with estimations from previous works and strongly suggest that the T2 relaxation from water within the intracellular and interstitial spaces is described by the intermediate component, whereas the long component represents water within the vascular space.     

Virtual and solution conformations of oligosaccharides

The possibility that observed nuclear Overhauser enhancements and bulk longitudinal relaxation times, parameters measured by 1H NMR and often employed in determining the preferred solution conformation of biologically important molecules, are the result of averaging over many conformational states is quantitatively evaluated. Of particular interest was to ascertain whether certain 1H NMR determined conformations are "virtual" in nature; i.e., the fraction of the population of molecules actually found at any time within the subset of conformational space defined as the "solution conformation" is vanishingly small. A statistical mechanics approach was utilized to calculate an ensemble average relaxation matrix from which (NOE)'s and (T1)'s are calculated. Model glycosidic linkages in four oligosaccharides were studied. The solution conformation at any glycosidic linkage is properly represented by a normalized, Boltzmann distribution of conformers generated from an appropriate potential energy surface. The nature of the resultant population distributions is such that 50% of the molecular population is found within 1% of available microstates, while 99% of the molecular population occupies about 10% of the ensemble microstates, a number roughly equal to that sterically allowed. From this analysis we conclude that in many cases quantitative interpretation of NMR relaxation data, which attempts to define a single set of allowable torsion angle values consistent with the observed data, will lead to solution conformations that are either virtual or reflect torsion angle values possessed by a minority of the molecular population. On the other hand, calculation of ensemble average NMR relaxation data yields values in agreement with experimental results. Observed values of NMR relaxation data are the result of the complex interdependence of the population distribution and NOE (or T1) surfaces in conformational space. In conformational analyses, NMR data can therefore be used to test different population distributions calculated from empirical potential energy functions.     

SARA: a software environment for the analysis of relaxation data acquired with accordion spectroscopy

We present SARA (Software for Accordion Relaxation Analysis), an interactive and user-friendly MATLAB software environment designed for analyzing relaxation data obtained with accordion spectroscopy. Accordion spectroscopy can be used to measure nuclear magnetic resonance (NMR) relaxation rates in a fraction of the time required by traditional methods, yet data analysis can be intimidating and no unified software packages are available to assist investigators. Hence, the technique has not achieved widespread use within the NMR community. SARA offers users a selection of analysis protocols spanning those presented in the literature thus far, with modifications permitting a more general application to crowded spectra such as those of proteins. We discuss the advantages and limitations of each fitting method and suggest a protocol combining the strengths of each procedure to achieve optimal results. In the end, SARA provides an environment for facile extraction of relaxation rates and should promote routine application of accordion relaxation spectroscopy.     

Synthesis, characterization, and relaxivity of two linear Gd(DTPA)-polymer conjugates

Two linear polyamide conjugates of Gd(DTPA)2- were synthesized and characterized by high-resolution nuclear magnetic resonance (NMR) spectroscopy and size exclusion chromatography (SEC). DTPA was copolymerized with two different diamines, 1,6-hexanediamine and trans-1,4-cyclohexanediamine, yielding the polymers DTPA-HMD and DTPA-CHD, with low polydispersity. Their molecular flexibility in solution was studied using 13C spin-lattice relaxation time measurements, indicating that the cyclohexanediamine linking moiety of the DTPA-HMD polymer is more rigid than that of DTPA-CHD. The influence of the flexibility of the linking functionalities on the relaxivity of the Gd3+-DTPA-polymer conjugates was studied by water nuclear magnetic relaxation dispersion (NMRD). The relaxivity of the Gd(DTPA-CHD) polymer was only slightly higher than that of the Gd(DTPA-HMD) polymer, and only two times higher than the usual values for small Gd-DTPA-like chelates. The low relaxivities obtained for both polymers, much lower than expected from the polymer apparent molecular weights, result from their substantial residual flexibility, and also from a too long, nonoptimal, value of the inner-sphere water exchange rate. These polymeric compounds are also cleared very quickly from the blood of rats, indicating that they are of limited value as blood pool contrast agents for MRA.     

Oxygen-17 relaxation times in the blood sera of rats with various cancers. Can a systemic effect be detected?

17O NMR relaxation times of water in the serum of rats with various cancers were measured. No systemic effect could be detected at 4.7 and 8.4 T. The serum T1(17O) value was 7.6 +/- 0.5 ms at 37 degrees C independent of the magnetic field. T2(17O) was approximately half T1(17O). The 17O relaxation times could be determined at a faster rate than the 1H relaxation times.     

Polymer mobility in cell walls of cucumber hypocotyls

Cell walls were prepared from the growing region of cucumber (Cucumis sativus) hypocotyls and examined by solid-state 13C NMR spectroscopy, in both enzymically active and inactivated states. The rigidity of individual polymer segments within the hydrated cell walls was assessed from the proton magnetic relaxation parameter, T2, and from the kinetics of cross-polarisation from 1H to 13C. The microfibrils, including most of the xyloglucan in the cell wall, as well as cellulose, behaved as very rigid solids. A minor xyloglucan fraction, which may correspond to cross-links between microfibrils, shared a lower level of rigidity with some of the pectic galacturonan. Other pectins, including most of the galactan side-chain residues of rhamnogalacturonan I, were much more mobile and behaved in a manner intermediate between the solid and liquid states. The only difference observed between the enzymically active and inactive cell walls, was the loss of a highly mobile, methyl-esterified galacturonan fraction, as the result of pectinesterase activity.     

N-linked oligosaccharide changes with oncogenic transformation require sialylation of multiantennae

Glycopeptides derived from NIH 3T3 fibroblasts and these cells transformed by transfection with human DNA containing oncogene H-ras were analyzed by 500-MHz 1H-NMR spectroscopy and binding to immobilized lectins. The cells were metabolically labeled with D-[3H]glucosamine or L-[3H]fucose and the glycopeptides included in Bio-Gel P-10 (Mr 5000-3500) were separated into neutral and charged fractions on DEAE-cellulose. The major portion (80%) of these [3H]fucose glycopeptides from the non-transformed NIH 3T3 fibroblasts were neutral or contained one or two charged residues, whereas 90% of the glycopeptides from the transformed cells contained two or more charged residues. The structure of the predominant neutral glycopeptide from the non-transformed NIH 3T3 cells was determined by 1H-NMR spectroscopy to be tetraantennary containing terminal Gal alpha 1----3. (formula; see text) This structure was verified by binding to the immobilized alpha-Gal-specific lectin, Griffonia simplicifolia I and leukoagglutinating phytohemagglutinin from Phaseolus vulgaris (L-PHA), which binds certain tri- or tetraantennary glycopeptides. In contrast, the structure derived by NMR spectroscopy of one of the predominant charged glycopeptides from the transformed cells was triantennary containing terminal NeuNAc alpha 2----3 in addition to Gal alpha 1----3. (formula; see text) In attempting to verify this structure by lectin-binding properties it was found that removal of NeuNAc alpha 2----3 reduced the affinity to L-PHA - agarose. The other major glycopeptides of the transformed cells which were more charged also cotained NeuNAc alpha 2----3 but no NeuNAc alpha 2----6 or Gal alpha 1----3. A tentative structure was proposed for the major glycopeptide of the first charged class from NIH 3T3 cells on the basis of lectin-binding properties and the NMR spectrum which showed, in addition to NeuNAc alpha 2----3, the presence of NeuNAc alpha 2----6 and Gal alpha 1----3. On the basis of the NMR spectrum and other results, it is concluded that the presence of tetraantennary oligosaccharides are not sufficient for the transformed oligosaccharide phenotype. Rather, the tri- or tetraantennae must be sialylated in alpha 2----3 linkage, on more than one antennae, when properties of transformation are expressed in NIH 3T3 cells. Prior to transformation the tetraantennary oligosaccharides of these cells are terminated in alpha-Gal residues, whereas after transformation alpha-Gal residues appear to be replaced by NeuNAc alpha 2----3.(ABSTRACT TRUNCATED AT 400 WORDS)     

Post mortem changes in skeletal muscle proton spin-lattice relaxation times

A previously published report indicated that there are early post mortem changes in the pulsed NMR proton spin-lattice relaxation time (T1) of skeletal muscle which must be taken into account in in vitro tissue analysis. We re-examined this subject. When T1 measurements were done by allowing the signal intensity to decay over two orders of magnitude from the original intensity, the T1 decay curves showed more than a single exponential relaxation time component as reported earlier. However, when T1 measurements were done by allowing signal intensity to decay to about one order of magnitude from the original intensity only a single exponential relaxation time component was found. We postulate that the latter method gives the average T1 value of the various macroscopic tissue components present and that this method gives a representative T1 value for the entire tissue. Using data obtained in this manner we could not find significant post mortem changes in the muscle T1 relaxation times during the first four hours but found change at a later time.     

Water proton magnetic resonance studies of normal and sickle erythrocytes. Temperature and volume dependence.

The temperature and cell volume dependence of the NMR water proton line-width, spin-lattice, and spin-spin relaxation times have been studied for normal and sickle erythrocytes as well as hemoglobin A and hemoglobin S solutions. Upon deoxygenation, the spin-spin relaxation time (T2) decreases by a factor of 2 for sickle cells and hemoglobin S solutions but remains relatively constant for normal cells and hemoglobin A solutions. The spin-lattice relaxation time (T1) shows no significant change upon deoxygenation for normal or sickle packed red cells. Studies of the change in the NMR linewidth, T1 and T2 as the cell hydration is changed indicate that these parameters are affected only slightly by a 10-20% cell dehydration. This result suggests that the reported 10% cell dehydration observed with sickling is not important in the altered NMR properties. Low temperature studies of the linewidth and T1 for oxy and deoxy hemoglobin A and hemoglobin S solutions suggest that the "bound" water possesses similar properties for all four species. The low temperature linewidth ranges from about 250 Hz at -15 degrees C to 500 Hz at -36 degrees C and analysis of the NMR curves yield hydration values near 0.4 g water/g hemoglobin for all four species. The low temperature T1 data go through a minimum at -35 degrees C for measurements at 44.4 MHz and -50 degrees C for measurements at 17.1 MHz and are similar for oxy and deoxy hemoglobin A and hemoglobin S. These similarities in the low temperature NMR data for oxy and deoxy hemoglobin A and hemoglobin S suggest a hydrophobically driven sickling mechanism. The room temperature and low temperature relaxation time data for normal and sickle cells are interpreted in terms of a three-state model for intracellular water. In the context of this model the relaxation time data imply that type III, or irrotationally bound water, is altered during the sickling process.