MAGNETIC RESONANCE IN MEDICINE
18,280-293 ( I99 I )
Evidence for a Contribution of Paramagnetic Ions to Water Proton Spin-Lattice Relaxation in Normal and Malignant Mouse Tissues * WILLIAMNEGENDANK,~ THOMASCORBETT,~ MICHAELCROWLEY,~ AND
CHRISTOPHER KELLOGG-~
t Hematology-Oncology Division, Department of Medicine, and $Medical Physics Program, Depariment of Radiology, P.O. Box 02188, Wayne State University, Detroit, Michigun 48201 Received January 26, 1990 revised June 12, 1990 Paramagnetic ions complexed to proteins may lose, retain, or enhance solvent paramagnetic relaxation (SPR) relative to free solution. We measured T , and T 2 of three mouse cancers, their normal counterparts, and six additional tissues. Long TI of cancers was not caused by necrosis or by different contents of water, fat, or blood. Dissociable (TCA-extractable) and nondissociable (ashed) Mn, Cu, and Fe were measured by AA. Cancers had less Mn, Cu, and Fe than did normal counterparts. All 12 tissues had inverse correlations between T Iand dissociable Mn and Cu. For Mn alone to account for reduced T I ,the extent to which SPR ofthe Mn-protein complexes would be enhanced is by factors of 0.6 to 13, below the maximum observed in Mn-enzymes. Different amounts of paramagnetic ion-protein complexes may account for part of the differences in TI of water protons in different tissues, and the longer TI of cancer cell water may be caused in part by reduced amounts Of such complexes. Q 1991 Academic Press, Inc. INTRODUCTION
The observation that the water proton in cancers has T , and T2 relaxation times longer than those in normal tissues ( 1) is, with some exceptions, consistent in vitro and in vivo in animals and humans (2). It is not specific, however, since long T I or T2occur in a variety of pathological states (2). Nevertheless, causes of long relaxation times in cancers may differ from causes of long relaxation times in other pathological states. Consequently, it is of interest to know why the water proton in cancers has a long T I and T 2 ,and whether or not they correlate with biological or biochemical aspects of cancer cells. To address these questions requires ability to study the normal cellular counterpart of a particular cancer, and to obtain relaxation times of the cancer cell and its normal counterpart in the presence of heterogeneous tissue elements. Thus a longer T I of a cancer compared to a normal tissue could be caused by more elements with long T I such as interstitial edema and necrosis, or fewer elements with short T I such as fat and fibrosis. In some cases the normal counterpart of the cancer is a minor element within a complex tissue. One approach to these problems, to study cells in culture or to compare culturable cells from benign and malignant tumors arising from the same * Presented at the Annual Meeting of the Society of Magnetic Resonance in Medicine, San Francisco, August, 1988. 0740-3194/9 I $3.00 Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.
280
PARAMAGNETIC CONTRIBUTION TO Ti
28 1
tissue ( 3 ) , has limited applicability because it is difficult to obtain or culture the normal or benign cellular counterpart of the malignant one. Literature describing water proton T I and T 2 in cancers and normal tissues was reviewed in detail by Bottomley et a/.( 2 , 5 ) .Factors contributing to T iand T2include water content relative to solids, long-range hydrodynamic or ordering effects by biopolymers on bulk cell water, and interaction of water with macromolecular surfaces. Water in cells exchanges rapidly between different environments and relaxation rates are averages weighted by the populations of water molecules in these environments (6 ). A small population of water on a macromolecular surface with a rapid relaxation rate can dominate observed TI and TZ.If the interaction is brief, the relaxation correlation time may be determined by magnetization transfer between water protons and macromolecular protons or nitrogens (cross relaxation ) or by paramagnetic centers. There are many potential sources of these mechanisms within cells and it is difficult to isolate them experimentally. Recent studies using saturation transfer in vivo suggest that some form of magnetization transfer is important ( 7 ) . It has been suggested that paramagnetic ions affect water proton T I of cells (8-10) but results of studies on this subject are inconclusive (11-19). Concentrations of freely dissolved Mn, Cu, and Fe are too low to be a significant factor ( 11, 17, 18).Moreover, determination of the extent of solvent paramagnetic relaxation (SPR) by Mn, Cu, and Fe complexed to proteins is confounded by their different forms: SPR of the free ion may be lost on complexation (e.g., Fe in hemoglobin (20, 2 3 ) ) ,retained (e.g., Fe in transfemn ( 2 4 ) ) ,or enhanced (e.g., Mn in pyruvate carboxylase ( 2 5 ) ) .Recently, however, Ling (20, 22) found correlations between TCA-extractable ions and T I of frog and mouse tissues and murine ascites tumors, strongly suggesting an effect of paramagnetic ions on water proton T I . We studied factors affecting T I and T , in transplanted mouse tumors for which normal counterparts are available. We assessed elements of tissue heterogeneity like water content, fat content, and vascular space, and used small tumors to avoid spontaneous necrosis or hemorrhage. We used these models and six additional tissues to determine whether or not paramagnetic ions might contribute to water proton NMR relaxation times. MATERIALS AND METHODS
Mice were obtained from Jackson Laboratories or bred from Jackson stock. Tumors were maintained in the mouse strain of origin and transplanted into the PIhybrid or the strain of origin. Individual body weights of mice were over 17 g and within 5 g of each other. After experience with several tumors, we selected three for this study: ( 1 ) Hepatoma 129 was implanted subcutaneously in the flank of C3H mice and studied after five passages by which time its growth characteristics were uniform. It is a moderately well-differentiated cancer with cytological features resembling normal hepatocytes. Since most cells in liver are hepatocytes, the liver provides a reasonably welldefined normal counterpart of the hepatoma. Hepatoma 129 was also transplanted in the peritoneal ascites form, which provides cell suspensions used to study the relation between bulk water content and water proton T I .( 2 ) AJSR lymphoma, a T-cell lymphocytic lymphoma, was implanted intravenously ( 5 X lo5 cells) in AKR mice. ( 3 ) L 1210 leukemia, a B-cell lymphocytic leukemia, was implanted intravenously (5 X 10 cells) in DBA mice. In AKR lymphoma and L1210 leukemia we chose the spleen as
282
NEGENDANK ET AL.
the normal counterpart since it is the largest organ that contains primarily normal lymphocytes and it is convenient to use the infiltrated spleens to determine properties of the lymphoma and leukemia cells. Serial histologic studies were performed to determine optimal durations of tumor growth so that necrosis and hemorrhage were avoided. Tumors excised from mice sacrificed at different times after implantation were fixed in formalin, sectioned, and stained with hematoxylin/eosin. From these studies we chose 10- 1 1 days' growth for hepatoma 129 (i.e., approximately 700 mg), and 4 days for AKR lymphoma and L12 10 leukemia (spleen weights increased from approximately 75 to approximately 250 mg). Mice were sacrificed by cervical dislocation. The entire liver or spleen and the entire hepatoma 129 tumor were removed and placed on filter paper moistened with Hank's solution on a teflon cutting board using stainless steel instruments. Extraneous fat, connective tissue, and blood vessels were removed. Any surface blood was removed by a single dip in Hank's medium. Drying was avoided by letting the tissue touch only moistened filter paper and by working very quickly. For NMR studies tissues were cut into small pieces (not minced) and packed gently into tared 0.4-ml polyethylene microtubes to eliminate air spaces. The tube was capped to prevent drying. All preparations were at room temperature. NMR studies were performed within 1-2 h of tissue removal, so it is safe to assume that no significant changes in TI or T, occurred (26, 27). In general, NMR studies were performed first, then the samples were processed for water contents, paramagnetic ion contents, or both. For measurements of water proton T I and T2the 0.4-ml microtube containing the tissue was placed into proper position in a 7.5-mm NMR tube. TI was determined by inversion recovery (8 intervals, 10 averages) and T2by the CPMG sequence ( 10 echoes at 10 intervals), in a Bruker PC20 spectrometer at 20 MHz with the probehead thermostated to 37°C. TI and T , were calculated by fits to single exponential functions. In none of our tissues did the data reveal multiexponentiality (28, 2 9 ) . Instrument reproducibility and variance were monitored by a standard, 500 pM MnC12,with T I of 260 ms at 20 MHz reproducible within +I%. Reproducibility of a tissue sample was < k5%. Following the NMR experiment the 0.4-ml microtube containing the tissue was weighed and the specimen dried to constant weight (within 3 days) at 105°C. Water content was expressed as percentage of wet weight. The dried tissue pellet was extracted in three changes of 20 ml each of petroleum ether, then redried and reweighed to determine fat content ( 3 0 ) . Paramagnetic ions were extracted using solutions prepared from deionized water. All materials (test tubes, pipettes, instruments, etc.) were determined by sham extractions not to release Mn, Cu, or Fe. Dissociable paramagnetic ions were extracted with 10%TCA. The rationale for this approach is that Fe not extracted by 10% TCA, such as Fe in heme proteins (hemoglobin, myoglobin, and cytochromes) does not cause solvent proton relaxation (20, 23), while Fe and Mn dissociated by 10% TCA, such as Fe in transfenin ( 2 4 )and Mn in pyruvate carboxylase ( 2 5 ) ,do. It is not known if TCA-extractability and SPR are correlated in all macromolecular complexes of Fe, Mn, and Cu. However, in the tissues studied here, most Mn and, except for liver, most Cu, are TCA-extractable. The tissue was placed in 10%TCA at three times its volume, homogenized, and heated
PARAMAGNETIC CONTRIBUTION TO T ,
283
for 20 min in a boiling water bath. Suspensions were separated in 3- or 5-ml plastic tubes and aliquots of the supernate were analyzed by atomic absorption ( AA). Total tissue Mn, Cu, and Fe were determined by ashing. Fresh tissues were placed in tared quartz crucibles, dried at 105"C, and ashed in a muffle furnace at 600°C for 4-5 days (or until the ash is white). The ash was dissolved in 10%TCA for measurement of Mn, Cu, and Fe by AA. TCA extracts were flamed in a Varian Spectr-AA-40 atomic absorption spectrophotometer with a microinjection apparatus permitting samples of 100 pl. Standards were prepared from certified stocks diluted in 10% TCA. Replicate measurements were averaged to reduce effects of sporadic errors. Concentrations were expressed as pmol/kg wet tissue weight. Corrections were not made for the fraction of vascular space. Some of the Fe and a small amount of Cu extracted by TCA are from blood (Cu from red cells and plasma and Fe from plasma transfenin). Additional Fe is obtained from hemoglobin by ashing. Vascular spaces of tissues used in our experiments were previously determined ( 3 1 ) , using I9F NMR of tissues from mice whose blood was replaced by a perfluorocarbon. Corrections were not made for the fraction of interstitial space, which was not determined. Therefore, reported concentrations of Mn, and probably Cu, are lower than their intracellular concentrations. To estimate the extent to which the higher water content of hepatoma 129 contributes to its T I ,we grew the hepatoma in ascites form. Ten to thirteen days after inoculation of lo6 cells into the peritoneum, the ascitic suspension was harvested and the cells separated from fluid at 200 g. Aliquots of the cells were incubated in Hank's medium in which tonicity was varied by adding different amounts of sucrose. After 30 min cells were pelleted at 200 g, the supernatant removed, the cells packed at 1500 g in 0.4-ml microtubes, and relaxation times and water contents determined. Results are also reported from human peripheral blood lymphocytes obtained by leucopheresis of normal volunteers and purified by Hypaque-ficoll fractionation as described previously (32). RESULTS
Hepatoma 129 is moderately well-differentiated and bears a resemblance to normal hepatic parenchyma (Fig. 1 ) . Normal liver has large vascular channels (Fig. 1a) while hepatoma does not (Fig. 1b). Hepatoma taken on the 12th day of growth (Fig. 1b) does not have areas of necrosis, while the one taken on the 15th day (Fig. lc) does. Normal mouse spleen contains predominantly small lymphocytes and has large vascular channels. Malignant lymphocytes of the AKR lymphoma and the L12 10 leukemia displaced most of the normal splenic tissue by the 4th day of growth, at which time the splenic weight increased 300-350%. From these studies we determined maximal durations of growth to avoid necrosis or spontaneous hemorrhage into the transplanted tumor. Consequently, the NMR relaxation times are not affected by necrosis, which can cause an increased amount of interstitial water and either an increase (33, 34) or a decrease (35, 36) in TI, or hemorrhage, which can have different influences on T i and T2depending on the age of the extravasated or clotted blood (37). Properties of the tissues, along with water proton TI and T 2 ,are shown in Table 1. All three cancers have relaxation times longer than those of their normal counterparts. The longer T I of the cancers is not caused by less fat, which if present in fractions
284
NEGENDANK ET AL.
F I G . 1. Light microscopic sections (X30) of ( a ) normal C3H liver, ( b ) subcutaneous hepatoma 129 grown for 12 days, and ( c ) subcutaneous hepatoma 129 grown for 15 days. A band of necrosis is evident in (c).
greater than 2-370 can contribute a short T I component (38). The longer relaxation times in the cancers are not caused by more vascular space, which contains blood with a relatively long T I ,1 141 +- 7 ms (SD, n = 4 C3H mice); on the contrary, there is less vascular space in two, and no change in one of the cancers. The contribution of blood in vascular space to observed T Iand T2is relatively small. Thus, when blood with T I = 1141 and T2 = 245 ms is replaced by a perfluorocarbon substitute (by exchange transfusion performed as described in Ref. ( 3 1 ) )with T I = 2671 and T2 = 16 17 ms, T Iand T2of the excised liver and hepatoma were not significantly affected (eg., liver T I 350 It 20 vs 307 k 31 ms, SD n = 3, respectively; hepatoma TI 715 k 22 and 708 f 36 ms, respectively). This result is anticipated: since relaxation processes add as rates ( 1 / T I), liver T I would be expected to change by 5% and hepatoma by only 2%. AKR lymphoma and L1210 leukemia do not have higher water contents than spleen (Table 1 ), so their longer T Iis not caused by more water. Hepatoma 129 does have a higher water content than liver. To estimate the extent to which higher water content of hepatoma 129 contributes to T I ,we grew it in peritoneal ascites form, harvested the cells, and incubated them in media of different tonicities. The relation between 1 / T , and water content (Fig. 2) is linear over this range and extrapolates close to the values for medium containing no cells. Values for subcutaneously trans-
PARAMAGNETIC CONTRIBUTION TO Ti
FIG.I-Continued
285
286
NEGENDANK ET AL. TABLE 1 Properties of the Transplantable Mouse Tumor Models and Significance p of Unpaired &Test with n = 4 - 8
Water Mouse strain
Tumor model
Fat
Vascular space
T2
T,
(76 Wet weight)
(ms)
C3H
Liver Hepatoma 129 P
70.7 i 0.3 82.5 f 0.6 0.50
26.2 f 8.6 5.0 1.8 OSO
I,
*
18,280-293 ( I99 I )
Evidence for a Contribution of Paramagnetic Ions to Water Proton Spin-Lattice Relaxation in Normal and Malignant Mouse Tissues * WILLIAMNEGENDANK,~ THOMASCORBETT,~ MICHAELCROWLEY,~ AND
CHRISTOPHER KELLOGG-~
t Hematology-Oncology Division, Department of Medicine, and $Medical Physics Program, Depariment of Radiology, P.O. Box 02188, Wayne State University, Detroit, Michigun 48201 Received January 26, 1990 revised June 12, 1990 Paramagnetic ions complexed to proteins may lose, retain, or enhance solvent paramagnetic relaxation (SPR) relative to free solution. We measured T , and T 2 of three mouse cancers, their normal counterparts, and six additional tissues. Long TI of cancers was not caused by necrosis or by different contents of water, fat, or blood. Dissociable (TCA-extractable) and nondissociable (ashed) Mn, Cu, and Fe were measured by AA. Cancers had less Mn, Cu, and Fe than did normal counterparts. All 12 tissues had inverse correlations between T Iand dissociable Mn and Cu. For Mn alone to account for reduced T I ,the extent to which SPR ofthe Mn-protein complexes would be enhanced is by factors of 0.6 to 13, below the maximum observed in Mn-enzymes. Different amounts of paramagnetic ion-protein complexes may account for part of the differences in TI of water protons in different tissues, and the longer TI of cancer cell water may be caused in part by reduced amounts Of such complexes. Q 1991 Academic Press, Inc. INTRODUCTION
The observation that the water proton in cancers has T , and T2 relaxation times longer than those in normal tissues ( 1) is, with some exceptions, consistent in vitro and in vivo in animals and humans (2). It is not specific, however, since long T I or T2occur in a variety of pathological states (2). Nevertheless, causes of long relaxation times in cancers may differ from causes of long relaxation times in other pathological states. Consequently, it is of interest to know why the water proton in cancers has a long T I and T 2 ,and whether or not they correlate with biological or biochemical aspects of cancer cells. To address these questions requires ability to study the normal cellular counterpart of a particular cancer, and to obtain relaxation times of the cancer cell and its normal counterpart in the presence of heterogeneous tissue elements. Thus a longer T I of a cancer compared to a normal tissue could be caused by more elements with long T I such as interstitial edema and necrosis, or fewer elements with short T I such as fat and fibrosis. In some cases the normal counterpart of the cancer is a minor element within a complex tissue. One approach to these problems, to study cells in culture or to compare culturable cells from benign and malignant tumors arising from the same * Presented at the Annual Meeting of the Society of Magnetic Resonance in Medicine, San Francisco, August, 1988. 0740-3194/9 I $3.00 Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.
280
PARAMAGNETIC CONTRIBUTION TO Ti
28 1
tissue ( 3 ) , has limited applicability because it is difficult to obtain or culture the normal or benign cellular counterpart of the malignant one. Literature describing water proton T I and T 2 in cancers and normal tissues was reviewed in detail by Bottomley et a/.( 2 , 5 ) .Factors contributing to T iand T2include water content relative to solids, long-range hydrodynamic or ordering effects by biopolymers on bulk cell water, and interaction of water with macromolecular surfaces. Water in cells exchanges rapidly between different environments and relaxation rates are averages weighted by the populations of water molecules in these environments (6 ). A small population of water on a macromolecular surface with a rapid relaxation rate can dominate observed TI and TZ.If the interaction is brief, the relaxation correlation time may be determined by magnetization transfer between water protons and macromolecular protons or nitrogens (cross relaxation ) or by paramagnetic centers. There are many potential sources of these mechanisms within cells and it is difficult to isolate them experimentally. Recent studies using saturation transfer in vivo suggest that some form of magnetization transfer is important ( 7 ) . It has been suggested that paramagnetic ions affect water proton T I of cells (8-10) but results of studies on this subject are inconclusive (11-19). Concentrations of freely dissolved Mn, Cu, and Fe are too low to be a significant factor ( 11, 17, 18).Moreover, determination of the extent of solvent paramagnetic relaxation (SPR) by Mn, Cu, and Fe complexed to proteins is confounded by their different forms: SPR of the free ion may be lost on complexation (e.g., Fe in hemoglobin (20, 2 3 ) ) ,retained (e.g., Fe in transfemn ( 2 4 ) ) ,or enhanced (e.g., Mn in pyruvate carboxylase ( 2 5 ) ) .Recently, however, Ling (20, 22) found correlations between TCA-extractable ions and T I of frog and mouse tissues and murine ascites tumors, strongly suggesting an effect of paramagnetic ions on water proton T I . We studied factors affecting T I and T , in transplanted mouse tumors for which normal counterparts are available. We assessed elements of tissue heterogeneity like water content, fat content, and vascular space, and used small tumors to avoid spontaneous necrosis or hemorrhage. We used these models and six additional tissues to determine whether or not paramagnetic ions might contribute to water proton NMR relaxation times. MATERIALS AND METHODS
Mice were obtained from Jackson Laboratories or bred from Jackson stock. Tumors were maintained in the mouse strain of origin and transplanted into the PIhybrid or the strain of origin. Individual body weights of mice were over 17 g and within 5 g of each other. After experience with several tumors, we selected three for this study: ( 1 ) Hepatoma 129 was implanted subcutaneously in the flank of C3H mice and studied after five passages by which time its growth characteristics were uniform. It is a moderately well-differentiated cancer with cytological features resembling normal hepatocytes. Since most cells in liver are hepatocytes, the liver provides a reasonably welldefined normal counterpart of the hepatoma. Hepatoma 129 was also transplanted in the peritoneal ascites form, which provides cell suspensions used to study the relation between bulk water content and water proton T I .( 2 ) AJSR lymphoma, a T-cell lymphocytic lymphoma, was implanted intravenously ( 5 X lo5 cells) in AKR mice. ( 3 ) L 1210 leukemia, a B-cell lymphocytic leukemia, was implanted intravenously (5 X 10 cells) in DBA mice. In AKR lymphoma and L1210 leukemia we chose the spleen as
282
NEGENDANK ET AL.
the normal counterpart since it is the largest organ that contains primarily normal lymphocytes and it is convenient to use the infiltrated spleens to determine properties of the lymphoma and leukemia cells. Serial histologic studies were performed to determine optimal durations of tumor growth so that necrosis and hemorrhage were avoided. Tumors excised from mice sacrificed at different times after implantation were fixed in formalin, sectioned, and stained with hematoxylin/eosin. From these studies we chose 10- 1 1 days' growth for hepatoma 129 (i.e., approximately 700 mg), and 4 days for AKR lymphoma and L12 10 leukemia (spleen weights increased from approximately 75 to approximately 250 mg). Mice were sacrificed by cervical dislocation. The entire liver or spleen and the entire hepatoma 129 tumor were removed and placed on filter paper moistened with Hank's solution on a teflon cutting board using stainless steel instruments. Extraneous fat, connective tissue, and blood vessels were removed. Any surface blood was removed by a single dip in Hank's medium. Drying was avoided by letting the tissue touch only moistened filter paper and by working very quickly. For NMR studies tissues were cut into small pieces (not minced) and packed gently into tared 0.4-ml polyethylene microtubes to eliminate air spaces. The tube was capped to prevent drying. All preparations were at room temperature. NMR studies were performed within 1-2 h of tissue removal, so it is safe to assume that no significant changes in TI or T, occurred (26, 27). In general, NMR studies were performed first, then the samples were processed for water contents, paramagnetic ion contents, or both. For measurements of water proton T I and T2the 0.4-ml microtube containing the tissue was placed into proper position in a 7.5-mm NMR tube. TI was determined by inversion recovery (8 intervals, 10 averages) and T2by the CPMG sequence ( 10 echoes at 10 intervals), in a Bruker PC20 spectrometer at 20 MHz with the probehead thermostated to 37°C. TI and T , were calculated by fits to single exponential functions. In none of our tissues did the data reveal multiexponentiality (28, 2 9 ) . Instrument reproducibility and variance were monitored by a standard, 500 pM MnC12,with T I of 260 ms at 20 MHz reproducible within +I%. Reproducibility of a tissue sample was < k5%. Following the NMR experiment the 0.4-ml microtube containing the tissue was weighed and the specimen dried to constant weight (within 3 days) at 105°C. Water content was expressed as percentage of wet weight. The dried tissue pellet was extracted in three changes of 20 ml each of petroleum ether, then redried and reweighed to determine fat content ( 3 0 ) . Paramagnetic ions were extracted using solutions prepared from deionized water. All materials (test tubes, pipettes, instruments, etc.) were determined by sham extractions not to release Mn, Cu, or Fe. Dissociable paramagnetic ions were extracted with 10%TCA. The rationale for this approach is that Fe not extracted by 10% TCA, such as Fe in heme proteins (hemoglobin, myoglobin, and cytochromes) does not cause solvent proton relaxation (20, 23), while Fe and Mn dissociated by 10% TCA, such as Fe in transfenin ( 2 4 )and Mn in pyruvate carboxylase ( 2 5 ) ,do. It is not known if TCA-extractability and SPR are correlated in all macromolecular complexes of Fe, Mn, and Cu. However, in the tissues studied here, most Mn and, except for liver, most Cu, are TCA-extractable. The tissue was placed in 10%TCA at three times its volume, homogenized, and heated
PARAMAGNETIC CONTRIBUTION TO T ,
283
for 20 min in a boiling water bath. Suspensions were separated in 3- or 5-ml plastic tubes and aliquots of the supernate were analyzed by atomic absorption ( AA). Total tissue Mn, Cu, and Fe were determined by ashing. Fresh tissues were placed in tared quartz crucibles, dried at 105"C, and ashed in a muffle furnace at 600°C for 4-5 days (or until the ash is white). The ash was dissolved in 10%TCA for measurement of Mn, Cu, and Fe by AA. TCA extracts were flamed in a Varian Spectr-AA-40 atomic absorption spectrophotometer with a microinjection apparatus permitting samples of 100 pl. Standards were prepared from certified stocks diluted in 10% TCA. Replicate measurements were averaged to reduce effects of sporadic errors. Concentrations were expressed as pmol/kg wet tissue weight. Corrections were not made for the fraction of vascular space. Some of the Fe and a small amount of Cu extracted by TCA are from blood (Cu from red cells and plasma and Fe from plasma transfenin). Additional Fe is obtained from hemoglobin by ashing. Vascular spaces of tissues used in our experiments were previously determined ( 3 1 ) , using I9F NMR of tissues from mice whose blood was replaced by a perfluorocarbon. Corrections were not made for the fraction of interstitial space, which was not determined. Therefore, reported concentrations of Mn, and probably Cu, are lower than their intracellular concentrations. To estimate the extent to which the higher water content of hepatoma 129 contributes to its T I ,we grew the hepatoma in ascites form. Ten to thirteen days after inoculation of lo6 cells into the peritoneum, the ascitic suspension was harvested and the cells separated from fluid at 200 g. Aliquots of the cells were incubated in Hank's medium in which tonicity was varied by adding different amounts of sucrose. After 30 min cells were pelleted at 200 g, the supernatant removed, the cells packed at 1500 g in 0.4-ml microtubes, and relaxation times and water contents determined. Results are also reported from human peripheral blood lymphocytes obtained by leucopheresis of normal volunteers and purified by Hypaque-ficoll fractionation as described previously (32). RESULTS
Hepatoma 129 is moderately well-differentiated and bears a resemblance to normal hepatic parenchyma (Fig. 1 ) . Normal liver has large vascular channels (Fig. 1a) while hepatoma does not (Fig. 1b). Hepatoma taken on the 12th day of growth (Fig. 1b) does not have areas of necrosis, while the one taken on the 15th day (Fig. lc) does. Normal mouse spleen contains predominantly small lymphocytes and has large vascular channels. Malignant lymphocytes of the AKR lymphoma and the L12 10 leukemia displaced most of the normal splenic tissue by the 4th day of growth, at which time the splenic weight increased 300-350%. From these studies we determined maximal durations of growth to avoid necrosis or spontaneous hemorrhage into the transplanted tumor. Consequently, the NMR relaxation times are not affected by necrosis, which can cause an increased amount of interstitial water and either an increase (33, 34) or a decrease (35, 36) in TI, or hemorrhage, which can have different influences on T i and T2depending on the age of the extravasated or clotted blood (37). Properties of the tissues, along with water proton TI and T 2 ,are shown in Table 1. All three cancers have relaxation times longer than those of their normal counterparts. The longer T I of the cancers is not caused by less fat, which if present in fractions
284
NEGENDANK ET AL.
F I G . 1. Light microscopic sections (X30) of ( a ) normal C3H liver, ( b ) subcutaneous hepatoma 129 grown for 12 days, and ( c ) subcutaneous hepatoma 129 grown for 15 days. A band of necrosis is evident in (c).
greater than 2-370 can contribute a short T I component (38). The longer relaxation times in the cancers are not caused by more vascular space, which contains blood with a relatively long T I ,1 141 +- 7 ms (SD, n = 4 C3H mice); on the contrary, there is less vascular space in two, and no change in one of the cancers. The contribution of blood in vascular space to observed T Iand T2is relatively small. Thus, when blood with T I = 1141 and T2 = 245 ms is replaced by a perfluorocarbon substitute (by exchange transfusion performed as described in Ref. ( 3 1 ) )with T I = 2671 and T2 = 16 17 ms, T Iand T2of the excised liver and hepatoma were not significantly affected (eg., liver T I 350 It 20 vs 307 k 31 ms, SD n = 3, respectively; hepatoma TI 715 k 22 and 708 f 36 ms, respectively). This result is anticipated: since relaxation processes add as rates ( 1 / T I), liver T I would be expected to change by 5% and hepatoma by only 2%. AKR lymphoma and L1210 leukemia do not have higher water contents than spleen (Table 1 ), so their longer T Iis not caused by more water. Hepatoma 129 does have a higher water content than liver. To estimate the extent to which higher water content of hepatoma 129 contributes to T I ,we grew it in peritoneal ascites form, harvested the cells, and incubated them in media of different tonicities. The relation between 1 / T , and water content (Fig. 2) is linear over this range and extrapolates close to the values for medium containing no cells. Values for subcutaneously trans-
PARAMAGNETIC CONTRIBUTION TO Ti
FIG.I-Continued
285
286
NEGENDANK ET AL. TABLE 1 Properties of the Transplantable Mouse Tumor Models and Significance p of Unpaired &Test with n = 4 - 8
Water Mouse strain
Tumor model
Fat
Vascular space
T2
T,
(76 Wet weight)
(ms)
C3H
Liver Hepatoma 129 P
70.7 i 0.3 82.5 f 0.6 0.50
26.2 f 8.6 5.0 1.8 OSO
I,
*
