Ferromagnetism in Two Mouse Tumours

Our previous studies revealed biological effects of non-ionizing radiation on tumour cells (Batkin & Tabrah, 1977). In a trial study to attempt to enhance this effect, we added paramagnetic spin-labelled compounds to the cells in vitro. Cell changes were observed after the addition of the paramagnetic compounds alone (Batkin, Tabrah & Misconi, 1980) and after further exposure to altered magnetic fields. Reports of ferritin-bearing tumours (Gropp, Havemann & Lehmann, 1978; Hann et al. 1979; Kew et al. 1978) and of magnetite in varied biological specimens led us to analyse selected tumours in our laboratory for the presence of ferromagnetic material.

Cells of Lewis lung tumour and YC-8 lymphoma were maintained by repeated subcutaneous transplant in female C57BL/6 and BALB/c mice. Experimental humours were excised from stock animals using non-magnetic glass microtome knives on the 13th day following transplant. Control samples of muscle and connective tissue from both strains of mice were also taken for comparison. In addition, three human carcinomas (gastric, colon and renal) were obtained after routine surgery in Honolulu. All tissue samples for magnetic analysis were first thoroughly cleaned using ultrasonic treatment in glass-distilled water. Using glass knives to avoid possible contamination from metal instruments, each tumour was subdivided into two or more subfractions of appropriate size for the SQUID magnetometer. All samples were frozen prior to and during the magnetic analysis.

Magnetic measurements on these tissue samples were performed on a super-conducting rock magnetometer of the type described by Goree & Fuller (1976), which was housed in a magnetically shielded mu-metal room. To reduce the chance of unwanted ferromagnetic contamination from dust particles during the measuring procedure, a temporary clean-lab enviroment was made in the magnetometer area using sheets of polyethylene plastic. A remanent magnetization was induced in each sample prior to measurement by briefly exposing it to the field of a small CoSm magnet (maximum intensity 0·3 Tesla, T). Although these magnets have an inhomogeneous field, most of the volume of each sample was exposed to intensities in excess of 0·1 T, which is above the coercivity of all but the most highly elongate magnetite crystals. Tissue samples which displayed a measurable ferromagnetic remanence were then progressively demagnetized in an alternating magnetic field, using standard paleomagnetic techniques (see McElhinny, 1973, for detailed discussions of these methods). Ice-cube control samples made from glass-distilled water were magnetized and measured along with the experimental tissues as a control for laboratory generated contamination ; no magnetic remanence was found in them. Estimates (± 10%) for the number of tumour cells in the experimental tissues were then made from histological sections, using a micrometer grid. In both the YC-8 lymphoma and Lewis lung tumour samples, between 5 and 10% of the volume was composed of blood vessels and connective tissue which was similar in appearance to that of the control samples.

Results from the magnetic analysis are displayed in Table 1 and Fig. 1. All samples of Lewis lung tumour and YC-8 lymphoma gave signals from 100 to 1000 times the background noise on the superconducting magnetometer, whereas samples from the normal mice and the human stomach, colon, and renal carcinomas were not notably ferromagnetic. Per cell, Lewis lung tumour is on the average nearly twice as magnetic as YC-8 lymphoma. Measurements on subdivided samples indicate that the magnetic material has a nearly uniform distribution. The progressive demagnetization experiments (Fig. 1) indicate that there are additional differences in the magnetic particles between the two cell types. Most of the ferromagnetic crystals in Lewis lung tumour have coercivities from 20 to 30 milli-Tesla (mT), whereas those in lymphoma cells are below 15 mT. Normally, the magnetic moment of such a particle is parallel to its long dimension, but it can point in either of two directions. For single-domain crystals, the coercivity is essentially the external magnetic field strength necessary to flip the direction of its remanent moment between these two positions. Although the coercivity of a ferromagnetic crystal is a complex function of its composition, size and shape, it generally increases with both particle elongation and volume. A comparison of these results with fig. 50 of McElhinny (1973) suggests that the Lewis lung tumour crystals are slightly larger or more elongate than those in YC-8 lymphoma cells, and that cells within one strain have nearly identical magnetic particles. Magnetite crystals made by a pure strain of magnetotactic bacteria are also remarkably regular in size and shape (Balkwill, Maratea & Blakemore, 1980), although a great deal of variation exists between different types (Kirschvink & Gould, 1981). Similarly, the magnetic properties reported here are consistent with the interpretation that cells of each tumour type use the same genetic blueprint for constructing their magnetic crystals. This situation would be difficult or impossible to produce inorganically in nature.

Thermomagnetic, X-ray, Mössbauer, or electron diffraction analyses have identified magnetite as the source of the ferromagnetism in all organisms and tissues where it has been extracted, purified or otherwise examined to date (bacteria, Frankel et al. 1979; chitons, Lowenstam, 1962; honey bees, Gould et al. 1978; pigeons, Walcott et al. 1979; dolphins, Zoeger et al. 1981 ; tuna fish, Walker & Dizon, 1981 ; sea turtles, Perry, Bauer & Dizon, 1981). If magnetite is also the mineral responsible for the tumour remanence, the average volume present per cell would be 4·5 × 10⁻¹⁶ ml in Lewis lung tumour and 2·5 × 10⁻¹⁶ ml in YC-8 lymphoma, corresponding to cubes with dimensions of 77 and 61 nm respectively. Crystals of this size, if even slightly elongate in shape, will behave in general as single-domain ferromagnets, a property which is necessary for and consistent with the presence of a magnetic remanence in these tissues (Kirschvink & Gould, 1981). Although these data alone cannot determine the number of discrete crystals within each cell, individual particles less than about 35 nm in diameter would be superparamagnetic instead of single-domain and would not hold a stable magnetization. This puts an upper bound of about 11 and 6 crystals per Lewis lung tumour and YC-8 lymphoma cell, respectively.

Several problems arise, however, in the attempt to locate and identify these crystals using standard techniques on a transmission electron microscope (TEM). Firstly, there are only a few of these magnetic particles present per cell ; consequently their relative density in the tissue is quite low. A typical 100 nm thick sample mounted on a TEM grid 1 mm in diameter, for example, would contain a total of only 10–70 crystals. Given their small size, these would be difficult to locate even though they are naturally electron-dense. The second problem is that other, much more abundant, subcellular components like pigment granules can have a similar TEM appearance. Small-area electron diffraction analysis (Towe & Moench, 1981) or X-ray fluorescence must be performed on each and every candidate particle to establish its crystalline nature and iron composition. Finally, any acidic step in the sample preparation would dissolve the small magnetite crystals. These factors probably explain why these particles have not been recognized in previous TEM studies. A more realistic initial approach might be to perform subcellular fractionation and centrifugation experiments and then use the magnetic remanence to pinpoint organelles associated with the ferromagnetic material. The tumour growth effects noted below together with the observations by Walsh, Shulman & Heidenreich (1961) of ferromagnetic inclusions in nucleic acid samples suggest that the nucleus would be a good place for an initial look.

At this stage, the biological function of these magnetic crystals also remains a mystery. The simplest hypothesis would be that they constitute some form of iron storage. Many tumours, however, have high internal concentrations of the iron storage protein ferritin (Dorner et al. 1980) and would not need to use magnetite for this purpose. Some forms have so much ferritin that radioisotope-labelled antiferritil antibodies have been used to locate tumours clinically (Order, Klein & Leichner,1981). It is interesting to note that the core of the ferritin macromolecule is made of the mineral ferrihydrite, the precursor of magnetite in chiton teeth (Kirschvink & Lowen-stam, 1979). High ferritin levels might lead to magnetite precipitation as a by-product, but this theory does not explain the apparent crystal size uniformity nor the relatively small fraction of the cell’s total iron content that is present in the ferromagnetic form. Ferritin is paramagnetic and cannot produce the observed remanence.

Many mouse tumours are known to be induced by viral action, and one possibility to consider is the production of magnetite under viral control. Although a virus has apparently not been found in association with Lewis lung tumour, YC-8 lymphoma is associated with a 45 nm RNA Moloney-sarcoma virus (MSV) in T-lymphocytes. This virus is probably too small to contain single-domain magnetite particles itself, but it might cause them to form somewhere else in the cytoplasm. In turn, this would indicate that the magnetic material is not in the nucleus.

Strong static or weak oscillating magnetic fields have been reported to influence adversely the growth rate of in vivo tumours (Batkin & Tabrah, 1977; Barnothy, 1964; Kim, 1976; Weber & Cerilli, 1971 ; Winterberg, 1967). Previously, these effects have been interpreted as arising from unspecified paramagnetic interactions or from induced ionic currents or ion motion in the cell cytoplasm. The presence of ferromagnetic crystals, however, provides at least two additional interpretations. Firstly, direct magnetomechanical motion of the crystals produced by their strong (> > kT) interactions with the applied magnetic fields and field gradients may damage adjacent subcellular organelles and, thereby, inhibit cell growth. Secondly, the relatively high, metallic-type electrical conductivity of magnetite, roughly 10³ to 10⁴ better than cytoplasm, may yield locally induced current densities that are strong enough to damage neighbouring structures, depending on the frequency and intensity of the applied electromagnetic fields.

Answers to these questions clearly await completion of the micro-anatomical and mineralogical characterization of the ferromagnetic particles. The discovery of the presence of ferromagnetic material in the YC-8 lymphoma and Lewis lung tumour cells led to further studies of the direct effects of magnetic fields on these tumour cells without the addition of paramagnetic compounds (in preparation).

We thank C. E. Helsley and B. Keating at the Hawaii Institute of Geophysics for the use of their paleomagnetics laboratory, J. Szekerczes of the Cancer Center of Hawaii for technical help with the tumour studies, and Roy Yee of Kerns, Inc. of Honolulu for engineering advice and equipment. This work was supported directly and indirectly by NSF grants SP179-14845, EAR78-03204, and BNS78-24754 (J. L. K.); and by Cancer Center Support Grant CA-15655 and Leahi Foundation Grant (S. B. and F. L. T.). Contribution No. 3742 of the Division of Geological and Planetary Sciences of the California Institute of Technology.