| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
REVIEW |
a Department of Biochemistry, University of Cambridge, Cambridge CB2 1TP, England
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEPIDEMIOLOGYTUMOR DEVELOPMENTHOW MIGHT EMFS CAUSE...IN VITRO CELLULAR STUDIESCONCLUSIONSREFERENCES |
|---|
Key Words: ELF-EMFs • protein kinase C • tumor development • growth factors • cell proliferation • geomagnetic field
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEPIDEMIOLOGYTUMOR DEVELOPMENTHOW MIGHT EMFS CAUSE...IN VITRO CELLULAR STUDIESCONCLUSIONSREFERENCES |
|---|
In contrast to the fields used in clinical practice, those normally encountered by the population are of much lower magnitude and frequency, arising from power distribution systems in which the frequencies generated are 60 Hz [in the United States (U.S.)] or 50 Hz (in Europe and much of the world) and immediate harmonics. Electric and magnetic fields of 50 Hz measured beneath energized power transmission lines are generally in the ranges of 1—10 kVm-1 and 1–10 µT, respectively (but can reach 11 kVm-1 and 100 µT), and greatly exceed those present in most domestic and commercial environments. Background electric fields in homes are generally 1–10 Vm-1, rising to 20–200 Vm-1 near small appliances such as electric razors or hairdryers (2, 3). In immediate proximity to electric blankets or heated water beds, these fields can be as high as several kVm-1 and have been implicated in pregnancy complications such as spontaneous abortion or intrauterine development disorders (4). Background 50 Hz magnetic fields in typical homes are between 0.01 and 1 µT, with appliances generating fields of 0.1–100 µT (
Table 1;
, ). Electric fields arise wherever there is a voltage, regardless of any current in the conductor, and are strongly attenuated by buildings and other objects, including the human body. Magnetic fields, however, are directly proportional to the current flowing in the conductor and are very weakly attenuated by the objects they encounter, although they decrease rapidly in magnitude with distance from the source. Thus, the magnetic fields emitted by domestic appliances are generally undetectable at a distance of 1 m.
|
|
|
The magnetic fields generated by electrical equipment are many times higher than those occurring naturally, and their prevalence is a consequence of technological developments in the second half of the 20th century. Since 1940, U.S. per capita power generation has increased by a factor of 10 and per capita consumption by a factor of 20 (5). Therefore, there has also been a dramatic increase in the exposure of the general population to EMFs over that period of time. Although the increase in exposure is less than that for per capita generation and consumption because of progressive improvements in, for example, underground wiring technology, the collective exposure dose has nonetheless increased by perhaps four- to fivefold since 1940.
EMFs of the magnitude to which we are now regularly exposed have been implicated as a contributory factor in the incidence of childhood cancers, particularly leukemia and brain cancer. Cancers afflict 1 in 600 children below the age of 15 and account for 16% of all United Kingdom (U.K.) childhood fatalities. Brain cancers account for about one-quarter of the childhood cancers and leukemias for one-third, of which the most common is acute lymphoblastic leukemia (ALL). The prognosis for childhood cancers has improved over the last 20 years (the death rate has been halved and the 5-year survival rate for ALL has risen from <10% to >70%) largely as a result of the introduction of combination chemotherapy, but the reported incidence of childhood leukemia has also increased, by between 10 and 20%, since 1945. This increase does not necessarily imply an increase in exposure of the population to agents that cause or promote cancer, but could instead arise from improvements in diagnosis and/or widespread changes in childhood disease patterns due to increasing control of infectious diseases. Nevertheless, the incidence of cancer in children and the emotive nature of serious childhood disease mean that reports linking environmental agents to childhood cancer attract considerable public attention; this has been particularly true recently for EMFs, with a number of legal claims against electricity generators and distributors receiving widespread media attention. These claims cite a wide range of epidemiological and biological studies as evidence for a contributory role of EMFs in cancer. This review is intended to survey and review the most significant evidence and to summarize the main issues that EMF research must address.
|
EPIDEMIOLOGY |
|---|
|
TOPABSTRACTINTRODUCTIONEPIDEMIOLOGYTUMOR DEVELOPMENTHOW MIGHT EMFS CAUSE...IN VITRO CELLULAR STUDIESCONCLUSIONSREFERENCES |
|---|
Table 2),
some of which also concluded that residential exposure to EMFs carried an increased risk of cancers, particularly leukemias, although other studies concluded that there was no significant risk. However, many of these studies were based on small numbers of high field-exposed leukemia cases and many relied on the arbitrary wiring configuration method or distance from transmission lines and distribution centers to estimate exposure. These drawbacks, together with the marginal nature of the positive correlations found, mean that no definite conclusions regarding direct exposure to high levels of EMFs can be drawn from these studies. Furthermore, the indirect measures of EMF exposure used may also correlate with other factors such as social status or chemical pollution. These unconsidered factors may have contributed to the cancer rates reported and also to the conflicting conclusions arrived at by various EMF studies. It is notable that Wertheimer and Leeper (6) commented that indirect effects of EMFs, for example, that change the distribution of particles emitting ionizing radiation could cause cancer, which suggests that factors besides EMF strength could also affect the capability for EMF exposure to cause cancer. This suggestion has recently been extended to include the possibility that aerosol particles containing radon daughter nuclei (218Po and 214Po) undergo enhanced deposition in the lungs under the influence of strong electric fields (7).
The Nordic studies
Three recent reports, often collectively referred to as the Nordic studies (8–10; Table 2), are carefully executed surveys that represent considerable advances over previous reports in terms of exposure measurement and data analysis.
The Swedish case control study (8) was confined to children living near power lines and concluded that there was an increased risk of childhood leukemia at exposure levels of >0.2 µT (odds ratio, O.R., 2.7) or >0.3 µT (O.R. 3.8). The study further concluded that for lymphoma, central nervous system tumors, and all childhood cancers combined, there was no significantly increased risk at any of the exposure levels. Note that the total number of leukemia cases in the high exposure (>0.2 µT) category was only seven. Furthermore, although these authors gave careful consideration to the problems associated with such a survey, there are inconsistencies in the results. For example, leukemia data were analyzed for one-family homes and apartments: for the latter there was no increase in risk, even though spot measurements indicated that fields were significantly higher than in one-family homes.
The Danish case control study (9) concluded there was no significant association of calculated exposures 
0.2 µT with childhood cancers (O.R. 1.5), but that fields 0.4 µT showed an association with major childhood cancers. This deduction was based on a total of six patients. For leukemia, the case exposure ratio was 3/829, the control exposure ratio 1/1659 (O.R. 6.0); for central nervous system tumors, the corresponding ratios were 2/623 and 1/1864 (O.R. 6.0), for malignant lymphoma 1/247 and 1/1247 (O.R. 5.0).
In the Finnish cohort study (10), "no statistically significant increases in all cancers and in leukemia or lymphoma were found in children at any exposure level." A statistically significant excess of nervous system tumors was detected in boys exposed to 0.2 µT (five cases), but no such tumors occurred in girls similarly exposed. This result probably arose from an anomaly in the study's design, since the excess of brain tumors can be attributed to one boy with three tumors counted as three cases; the relative risk is no longer significant if he is counted as only one case. The conclusion was that "residential magnetic fields of transmission power lines do not constitute a major public health problem regarding childhood cancer. The small numbers do not allow further conclusions about the risk of cancer in stronger magnetic fields".
The data from these three studies have been pooled (although the studies do not allow a full meta-analysis) in a subsequent report concluding that, taken together, the studies "support the hypothesis that exposure to magnetic fields of the type generated by transmission lines has some etiolgical role in the development of leukemia in children" (11). Therefore, these recent studies together provide evidence for an association between childhood cancers and exposure to EMFs. However, the inconsistencies within individual studies and between the broadly similar studies, along with the small magnitude of the positive associations reported, indicate the difficulties of acquiring convincing epidemiological data in this field. The lack of evidence for an association between brain tumors and EMF exposure is confirmed by some recent studies (12, 13).
The National Cancer Institute Study
The most comprehensive study to date of residential exposure to EMFs and childhood ALL involved 629 children with leukemia, 619 controls, and the direct measurement of magnetic fields at multiple locations in the current and former homes of the subjects (14). Wire code classifications were also assigned to 408 matched pairs of children; all magnetic field measurements and wire code estimations were carried out blind with respect to the health status of the resident subject. This study revealed no association between a risk for ALL and electromagnetic field strength within the homes of the children, nor was there any increase in risk with increasing wire code. Moreover, there was no association between risk of ALL and magnetic field strength or wire code in the residences occupied by the mothers while pregnant in those cases analyzed. These data contrast with earlier studies (6), but are consistent with others (15–17) that have shown no significant increase in risk of ALL for children exposed to residential levels of magnetic fields that exceed 0.2 µT. The study by Linet et al. (14) carries considerable weight not only because of the large number of subjects and the care taken to minimize bias in data acquisition, but also because magnetic field measurements were generally made within 2 years of diagnosis of leukemia, a much shorter interval than in most earlier surveys.
Adult studies
Numerous investigations of the incidence of adult cancers have used approaches similar to those of the childhood surveys summarized above ( Table 2). None have provided convincing evidence that exposure to above-ambient levels of EMFs in the home influences the development of leukemias, lymphomas, or solid tumors (18). However, several studies have considered whether employment in occupations with high EMF exposure may be associated with an elevated incidence of brain cancers and leukemias. As with the many residential studies, the initial occupational surveys were based on crude classifications. Fields were not measured in individual cases, the evidence for increased exposure being based either on a classification of the occupation or, later, on a measurement to determine an average exposure for a given occupation. This feature of the study, coupled with the marginal effects in some studies, the lack of correlations in the remainder, and the possibility of a publication bias in favor of reports showing increases, limits the overall conclusions that can be drawn (19, 20).
Five large and more detailed recent surveys show broad agreement and indicate a marginally increased risk of brain cancer and/or leukemia. Of these, a joint France/Canada study of hydroelectric power workers (21, 22) revealed a significant increase in acute myeloid leukemia (AML) and chronic nonlymphocytic leukemia with greater than median EMF exposure (relative risk estimate for leukemia of 1.5 and for AML of 2.7), although there was no correlation between risk and length of exposure and the survey showed no consistency between the results obtained from different utility companies. The risk of brain cancer was increased in individuals who experienced the highest 10% of the exposure range. Similarly, a study of French electricity workers showed an increased risk of brain cancers for those exposed to the highest 10% of electric fields (OR 3.08 with 95% confidence interval of 1.08–8.74), but no corresponding increase for leukemia (23). Further studies of hydroelectric power workers in Ontario revealed a significant increase in leukemia with high exposure to electric and magnetic fields, with a dose response for exposure to electric, but not magnetic, fields (24). An American study of five major utilities (25) showed a slight increase in brain cancer mortality with cumulative exposure (relative risk factor of 1.94 per µT year of exposure in the 2–10 years before diagnosis), but no increase for leukemia. A Swedish survey (26) estimated exposures of individuals with leukemia or brain cancer (posthumously for many of the cases); this, too, showed an increased risk for chronic lymphocytic leukemia and a slight, but not statistically significant, increase for brain cancer. Although many increases in these studies were not statistically significant, nevertheless every set of data reveals an increase in the number of individuals contracting brain cancer relative to the general population, and three of the studies show statistically significant increases for leukemia. A similar small increase in brain cancers and leukemias alone of 20 different cancers is revealed in a British study of correlations between occupations and cancer and between mortality and occupation. This study is biased slightly toward recording occupations in electrical industries. Conversely, a cohort study of British electricity workers showed no significant increased risk of mortality with exposure to high magnetic fields (27).
Analysis of these studies
It is clear that the epidemiological studies discussed above do not provide a clear answer to the question of whether EMF exposure influences the development of cancer. This is due in part to two shortcomings shared by all of the studies: 1) small sample number, and 2) no specific known EMF exposure parameter with which correlations should be made. The problem of small sample numbers in such surveys can be countered in part by undertaking more comprehensive studies, such as the U.K. child cancer study currently in progress. The second problem is more difficult to address because it comprises two components. Regardless of whether exposure is assessed by using a wiring configuration, measurement of distance from source, direct measurement of fields (either as continuous exposure or in bursts), or by classification of occupation, a large subjective element can arise. Second, until the mechanism of putative EMF interactions with organisms is understood, these figures must all be considered as mere surrogates for exposure, and their relationship to any critical parameter of EMF will vary from case to case. This may provide an explanation for apparent anomalies in epidemiological data such as those in the studies by Feychting and Ahlbom (8) and Pool (28), in which estimated risk varied with the exposure classification used. Thus, Feychting and Ahlbom (8) obtained an odds ratio of 1.7 for estimated magnetic fields of 0.2 µT and AML, but the risk estimate was close to unity when spot measurements of magnetic field were used. Finally, the complexity of genetic and epigenetic factors that promote cancers (see below) mitigates against clear epidemiological correlations in all but the most statistically overwhelming cases (e.g., smoking, alcohol, hepatitis B virus infection).
|
TUMOR DEVELOPMENT |
|---|
|
TOPABSTRACTINTRODUCTIONEPIDEMIOLOGYTUMOR DEVELOPMENTHOW MIGHT EMFS CAUSE...IN VITRO CELLULAR STUDIESCONCLUSIONSREFERENCES |
|---|
Fig. 1).
|
Changes detected in DNA in cancer range from single base changes to gross chromosomal abnormalities (e.g., translocations and deletions). Tumor initiators such as chemical carcinogens, radiation, UV light, or free radicals cause direct damage to DNA (base alterations, cleavage of DNA strands), which must avoid detection and correct repair by the cell to become a sustained mutation. Components of the cellular DNA repair machinery may also undergo mutation, providing an additional mechanism of genetic alteration leading to cancer. Homozygous mutation in a family of DNA mismatch repair genes leads to a mutator or replication error phenotype (RER+) characterized by a high level of genetic instability. This is most readily detected in regions of repeated short DNA sequences (microsatellite repeats), which are susceptible to mutation due to their highly repetitive structure. RER+ cells show shortening or lengthening of microsatellite repeats (microsatellite instability) when compared with normal tissue. Widespread microsatellite instability is a characteristic feature of hereditary nonpolyposis colorectal carcinoma (HNPCC), suggesting the trait is inherited in HNPCC families, but somatic loss of DNA mismatch repair occurs in a large variety of types of tumors, including chronic lymphocytic leukemia. Cells of the RER+ phenotype will be particularly vulnerable to the effects of free radicals or other DNA damaging agents, since their capacity to repair DNA damage is diminished.
Although abnormally high rates of proliferation can give rise to cancers, tumor development is not merely a consequence of deregulated proliferation. An additional cause is the modulation of the normal processes that lead to cell loss (programmed cell death or apoptosis). Apoptosis occurs normally in the body as a mechanism for clearance of redundant cells in development (e.g., in selection of appropriate reactive cells of the immune response), for the removal of cells at the end of their `active life' (e.g., granulocytic cells after wound healing), and also as a protective mechanism, removing damaged cells. The tumor suppressor gene product p53 is essential for many apoptotic responses (including response to some DNA damaging agents or to hypoxia). Loss of function of p53 may contribute to tumor development either by permitting damaged cells to continue to proliferate or by preventing differentiation or death of stem cells, both possible precursors of tumor development. These and other patterns of cell behavior have been described mathematically, the model accommodating observed features of cancer development that include the formation of early premalignant lesions that can regress spontaneously (e.g., in cervical intraepithelial neoplasia) and the occurrence of long-lag phases of growth that persist until mutations arise that eventually cause exponential growth (29).
Control of proliferation
The growth of cells can be controlled by the interaction of growth factors with their receptors on the plasma membrane. Such interactions may, for example, cause quiescent, somatic cells to leave Go, traverse G1, and enter S phase, whereupon they are normally committed to at least one round of the cell cycle ( Fig. 1). Activated growth factors [e.g., platelet-derived growth factor (PDGF) and epidermal growth factor; Fig. 2
] generate one or more primary signals that promote a sequence of metabolic events, including ionic changes (Ca2+ signals, Na+ influx), activation of protein kinase C (PKC) isoforms, increases in protein phosphorylation, increased activity of ornithine decarboxylase (ODC), activation of glucose, and amino acid uptake and stimulation of glycolysis. In parallel with these events, the coordinated transcription of ~100 genes is activated (30, 31). The `immediate early response genes' activated within approximately 1 h include ornithine decarboxylase (ODC1), ETS1 and ETS2, the JUN family (JUN, JUNB, and JUND), and the FOS family (FOS, FRA1, FRA2, and FOSB). The proto-oncogenes MYC and MYB are also transcriptionally activated approximately 2 and 6 h after cell stimulation, respectively. Appropriate combinations of these responses drive cells through the cell cycle and, in principle, any component of these pathways possesses oncogenic potential ( Fig. 2).
|
Patterns of mutations in cancers
It is not known how many mutations are required for the genesis of a specific cancer, but it is generally supposed that more than five are involved; in all but a small proportion of cases (<1%), these mutations are acquired somatically. Malignant colorectal carcinomas have been particularly informative in this respect (32). The majority of these cancers develop from benign adenomas in a process in which DNA hypomethylation occurs together with the acquisition of mutations in KRAS2, APC, MCC, DCC, and P53. The mutation or deregulation of a single component of a signaling pathway is not sufficient for the immediate development of a tumor, and recognizable stages of tumor progression can often be defined. This is well illustrated by in vitro cellular studies that show cell transformation to be at least a two-stage process, one oncogene being needed for immortalization (proliferation of cell through endless rounds of cell division without senescence) and another for full transformation (loss of checks to growth such as contact inhibition, stimulation by growth factors). Thus, primary fibroblasts can often be immortalized by RAS. Full transformation, however, generally requires the expression of two complementing oncogenes, typically RAS and MYC.
Animal models of tumorigenesis
The notion of progressive accumulation of mutations is also suggested by experimental animal models of tumorigenesis. The chemical induction of tumors generally requires the sequential application of two types of agents: an `initiator' and a `promoter'. The initiator is usually a mutagen that has irreversible effects and must be administered before the promoter. These model systems indicate that the initiating event in tumorigenesis is mutation and that subsequent tumor progression may be mediated by either genetic or epigenetic mechanisms. The classical system for studying the chemical initiation of tumors in animals is the mouse skin carcinoma model (33). The application of a low dose of a carcinogen to mouse skin can cause mutations in oncogenes, most frequently in Hras-1 (34), but causes carcinoma only if followed by multiple treatments with a tumor promoter. Tumor promoters are not themselves mutagens; alone they have no carcinogenic properties, but can stimulate cell proliferation. It seems probable that tumor promoters act in carcinogenesis to stimulate growth of a population of cells carrying an initial mutation, thereby increasing both the number of cells that carry this mutation and the probability of subsequent mutations occurring as cells are driven through successive rounds of DNA replication. Any process that perturbs the normal cell growth control mechanisms, for example, by modulating the activity of a component of a signaling pathway, could promote tumor formation either by stimulating cell proliferation or by removing checks to growth. Tumor promoters such as TPA (12-O-tetradecanoylphorbol 13-acetate) act by direct activation of PKC or by inactivation of protein phosphatases, and are presumed to act by increasing phosphorylation of substrates involved in mitogenic signaling pathways. It is established that anomalous stimulation by, for example, heat shock or UV radiation can dramatically activate transcription of the early response genes FOS ( Fig. 3)
and JUN. Normal growth factors synthesized in an appropriate setting may also act as `promoters' in the early developmental stages of cancer. For example, gastric-releasing peptide (GRP or mammalian bombesin) functions as an autocrine growth factor in small cell lung cancer: these tumor cells produce large amounts of the peptide, which causes phosphatidylinositol-(4,5)-bisphosphate hydrolysis and an increase in the concentration of intracellular free calcium ([Ca2+]i), characteristic responses of cells entering the cell cycle (35). In this situation, the growth factor may act selectively to promote the proliferation of a clone of tumor cells.
|
|
HOW MIGHT EMFS CAUSE CANCER? |
|---|
|
TOPABSTRACTINTRODUCTIONEPIDEMIOLOGYTUMOR DEVELOPMENTHOW MIGHT EMFS CAUSE...IN VITRO CELLULAR STUDIESCONCLUSIONSREFERENCES |
|---|
EMFs as mutagenic agents
EMFs might act in the same manner as conventional carcinogens to cause mutations that either initiate the progression of a cell to acquisition of a fully tumorigenic complement of abnormalities or accelerate its progress along that pathway. Although the energy associated with environmental EMFs is too low to cause direct changes to the structure of DNA, EMFs might affect the production of agents such as free radicals, which themselves can react with DNA, or of other agents that cause chromosomal damage, instigating translocation by inducing DNA breaks or by formation of unnatural DNA structures. Interfering with the mechanisms of DNA repair or chromosomal replication and segregation can also cause DNA damage or increase the probability that a particular DNA sequence will be lost from the genome.
A wide variety of experimental approaches have been used to detect mutagenic effects of EMFs, including assays for point mutations or analysis of gross DNA alterations such as DNA breaks, translocations, the formation of micronuclei, or the presence of cells with an inappropriate chromosome complement ( Table 3).
In bacterial studies, strains of Salmonella exposed to 15 or 30 mT magnetic fields at 0.3 Hz, 1–10 µT sinusoidal magnetic fields at 50 Hz applied in 2 µs pulses or to a wide range of radio frequency and microwave fields showed no increase in mutational frequency (36). Fields (0.2 mT) at 60 Hz have been reported to be comutagenic in Salmonella, causing a 14% increase in azide-induced mutations after 48 h exposure (37). Although EMF exposure increased the number of revertant colonies in every experiment described (a total of 15 experiments), the number of revertant colonies per `sector' of Petri dish (separate regions of the dishes in which colonies were counted) varied considerably in every experiment, and the resultant errors were significant at the 90%, but not the 95%, confidence level (P=0.078; paired Student's t test). Human lymphocytes exposed to sinusoidal 50 Hz, 60 Hz, or pulsed EMFs did not undergo detectable chromosomal alterations or inhibition of DNA repair mechanisms (36, 38–43). Primary human lymphocytes have been reported to show increased micronuclei formation when exposed to 75 or 150 µT fields at 32 Hz, but only when a parallel 42 µT d.c. field was applied concurrently to establish the cyclotron and paramagnetic resonance condition for calcium (44). The statistical analysis applied led to the conclusion that there was no increase in micronuclei formation in the absence of a d.c. field or at 50 Hz. However, as several of the individual experiments in the absence of a static field showed percentage increases comparable to or exceeding those detected when the 42 µT d.c. field was applied, these data appear to be equivocal. Nordenson et al. (45) reported increases in chromosome defects (gap and break frequencies) in human amniotic cells exposed to a continuous or intermittent 30 µT sinusoidal field at 50 Hz, but attempts to reproduce this work have been unsuccessful (46). The same field exposure conditions were used to investigate the fidelity of chromosome segregation in Chinese hamster cells and no effect of field exposure was seen (47). Hintenlang (48) reported that exposure to ionizing radiation, followed by 60 Hz magnetic fields of up to 1.4 mT, enhanced the frequency of near-tetraploid chromosome complements in human peripheral blood lymphocytes, an effect not seen with ionizing radiation alone. Negative effects have been reported in whole-animal studies: 50 Hz electric fields of 20 kVm-1 failed to increase the frequency of dominant lethal mutations in male mice, and 60 Hz fields of as much as 50 kVm-1 with magnetic fields of up to 1 mT did not affect the frequency of sister chromatid exchange in mouse bone marrow (19, 20). Thus, there are some reports of positive EMF effects. However, the inadequate internal controls and the failure of replication attempts suggest that EMFs of these strengths are not mutagenic.
EMF effects on chemical reactions involving free radicals
Free radicals are generated as intermediates in metabolism and may attack lipids, proteins, and DNA. Thus, any elevation in free radical production could increase the rate of chemical damage to DNA as occurs, for example, as a consequence of sustained activation of the immune system in response to chronic infection. Magnetic fields of more than 1 mT can have measurable effects on the kinetics and yield of chemical reactions that use geminate radical pairs through their effect on the spin precession rates of unpaired electrons and consequent effects on the lifetime of radicals (49–52). The magnetic field can increase or decrease precession rates between singlet and triplet spin-correlated states. Hence, a geminate radical pair born in the singlet spin state may rapidly recombine; after precession to the triplet spin state, recombination is prohibited by the Pauli exclusion principle, resulting in a longer radical lifetime. The consequence of this may be, for example, increased enzyme product or release of radicals from the enzyme ( Fig. 4).
|
Harkins and Grissom (53) have demonstrated the effects of magnetic fields on the enzyme B12 ethanolamine ammonia lyase. Fields of greater than 50 mT caused a decrease in the ratio Vmax/Km of the enzyme that was maximal (20% reduction) at 100 mT. McLauchlan et al. (51) describe effects of magnetic fields on a variety of photochemical reactions: the reaction of pyrene with dicyanobenzene generates radical pairs that fluoresce when they recombine. If the reaction is performed in a fluctuating magnetic field of a few mT, the fluorescence generated by the reaction fluctuates with the applied magnetic field. In more complex photochemical reactions, radicals are generated in the triplet state (spins parallel) and must move to the singlet state (spins antiparallel) before reacting. This interchange can occur as a result of local magnetic fields from nearby magnetic nuclei through the hyperfine interaction. When the reaction is performed in a static magnetic field, these interconversions are affected. The magnetic field removes the degeneracies of the triplet state sublevels, and at higher magnetic fields can cause separation between triplet states greater than the hyperfine interaction, effectively preventing interchange of electrons and stopping up to two-thirds of radical pairs reacting ( Fig. 4). An opposite, smaller effect may occur at very low field strengths to increase reaction yields by a few percent (due to the Zeeman splitting that removes the constraints of spin angular momentum conservation of the system, allowing singlet radical pairs to interconvert to more triplet states).
Alternating magnetic fields superimposed on static magnetic fields can further affect reactions by providing quanta of energy equal to the gap between singlet and triplet states, allowing transition of radicals and hence increasing reaction probability. The effect requires both static magnetic fields and fields fluctuating at a resonance frequency (generally in the range between low audio and microwave frequencies), dependent on the magnetic fields applied. Magnetic fields of only a few mT can affect chemical reactions and some enzymes that use radical intermediates. The theoretical explanation of these effects is that the fields do not alter the nature of reaction products, merely the yield, although it is possible that radicals escaping reaction due to these effects may leave the vicinity of the reactions and react with other components of a biological system. These examples represent clear, reproducible effects of magnetic fields on biochemical systems with a firm theoretical basis and provide a plausible approach to investigating other reported biological effects of low-strength magnetic fields.
These principles of magnetic field interactions have been investigated in a few intact cell systems, notably in neutrophils that generate reactive oxygen species, including the superoxide radical, from plasma membrane NADPH oxidase as a defense against bacteria. Superoxide can be further converted by superoxide dismutase to peroxide or by myeloperoxidase to hypochlorite. The generation of these reactive oxygen species can be monitored in cells in real time, using fluorescent dyes for oxidants. In such experiments, Roy et al. (54) have reported effects of 0.1 mT fields at 60 Hz on the phorbol ester-induced respiratory burst in primed rat neutrophil preparations. In all sets of experiments (five total), cells in magnetic fields showed up to 10% greater fluorescence than unexposed cells, although the effect was consistently smaller than the standard deviation of the measurements (standard deviations were as much as ±20%); there was also considerable variation in fluorescence between experiments, probably arising from differences in cell preparations on different days. Other groups have examined the effect of magnetic fields on nitric oxide production by murine macrophages (55) and observed no effect of exposure for 14 h to static fields of 1, 10, 50, and 100 mT or 1 Hz sinusoidal fields of 1.6 mT on nitrite levels.
Currently available data therefore do not provide any consistent evidence that ELF-EMFs affect cellular systems via free radical mechanisms. However, the data reviewed above indicate that complex effects can occur in defined chemical systems exposed to EMFs.
A free radical basis for magnetic field effects would have some important implications for investigations and epidemiological studies. The processes affected occur very rapidly, and so at the level of simple effects are independent of frequency; in many cases, the geomagnetic field exposure would far outweigh the alternating field. However, as described above, more complex effects can occur in vitro with specific combinations of static and alternating magnetic fields, and these combinations vary with the free radical species involved. These factors, together with the probable magnitude of any ELF-EMF-induced biological effect that may occur, suggest that it will be difficult to define quantitatively the required magnetic field variables.
EMFs as tumor promoters: modulation of function in signaling pathways
The foregoing discussion indicates that despite a possible mechanism whereby EMFs affect free radical formation, there is no firm evidence that EMFs cause damage to DNA; many experiments show no effect of EMF on DNA damage. Hence, the more plausible route by which EMFs might contribute to carcinogenesis is by acting as a promoter, increasing the probability that previously initiated cells will proliferate and progress to neoplasticity. This hypothesis has been tested in well-established animal model systems for tumor promotion, but with equivocal results ( Table 3). Rannug and co-workers (56) used two different chemically induced tumor systems in rats and mice to examine the long-term effects of field exposure on tumor development. Mice were treated with subcarcinogenic amounts of the tumor initiator DMBA and exposed to 50 Hz sinusoidal fields of 50 or 500 µT for 20 h/day for 2 years. EMFs had no effect on any aspect of carcinogenesis when the fields were applied continuously, but some effects were seen if the fields were applied as intermittent signals (15 s on/15 s off, again for 20 h/day, 103 wk; ref 57). Mice exposed to intermittent signals showed an increased skin tumor incidence of around twofold over nonexposed animals, and this was further increased in mice exposed at the higher field strength. However, neither increase was statistically significant. No measurable effect on endothelial hyperplasia or tumor load was seen. As a comparison, TPA, a phorbol ester and strong promoting agent, increased tumor incidence by almost 20-fold (from 5% in DMBA treated mice to 98%) and also had extensive effects on endothelial hyperplasia, time of appearance of first tumor, and tumor load per mouse.
In later studies in which Stuchly and co-workers (58, 59) used the mouse skin model, mice were treated with a single dose of DMBA and subsequent weekly treatments of the tumor promoter TPA. Half of these mice were also exposed to 2 mT fields at 60 Hz (6 h/day, 5 day/wk for 23 wk). The only aspect of carcinogenesis that appeared to be field dependent was the initial rate of tumor development, although the differences observed were not statistically significant at all time points. No effect of EMF exposure was seen at higher doses of TPA. Other known promoters with TPA—for example, prostaglandin—affected the rate and incidence of tumor development and the tumor load in individual mice. Considering the variability of the data (statistically significant differences at some time points only) and the small, rather anomalous effects, it is reasonable to conclude that no robust copromoting effect of EMFs was demonstrated. Magnetic fields (30 mT) at 50 Hz increased the number of mammary tumors induced by DMBA in rats by approximately 30% in some experiments reported by Mevissen et al. (60), although identical experiments reported in the same paper showed a decrease of similar magnitude in EMF-exposed rats relative to controls; the authors acknowledge that the small sample size and large variability limited the conclusions that could be drawn. These factors, coupled with the size and nature of any effects observed, apply equally well to all of the studies discussed here, and we conclude from the whole-animal studies that no unequivocal effects of EMFs on tumor promotion have been demonstrated.
|
IN VITRO CELLULAR STUDIES |
|---|
|
TOPABSTRACTINTRODUCTIONEPIDEMIOLOGYTUMOR DEVELOPMENTHOW MIGHT EMFS CAUSE...IN VITRO CELLULAR STUDIESCONCLUSIONSREFERENCES |
|---|
Table 3); major examples of these are summarized below.
Cell proliferation
Mitogen-stimulated lymphocyte growth
Numerous studies have determined the effects of EMF exposure on cell proliferation by counting the number of cells, estimating cell cycle phase progression, or measuring incorporation of radioactively labeled thymidine into DNA. EMF exposure (50 Hz sawtooth signals of 2.5 mT peak field for at least 6 h) increased thymidine incorporation by 20% in primary human lymphocytes stimulated with phytohemoglutanin (PHA), a T cell mitogen; it caused cells to enter the cell cycle and proliferate, but had no effect on unstimulated cells (61). Larger effects were seen with lymphocytes prepared from older donors (DNA synthesis increased by at least 60%), and the most dramatic effects were seen in B cells from patients with chronic lymphocytic leukemia, for which the mitotic index increased by threefold and thymidine incorporation by 100%. Similar but smaller effects (8% increase in cell cycle progression, assessed by proportion of cells undergoing mitosis) were seen by Rosenthal and Obe (43) under similar exposure conditions (50 Hz sinusoidal signal at 5 mT root mean square, or rms) although other independent laboratories (62–65) have reported inhibition of lymphocyte proliferation by EMF exposure [3 Hz square wave signal of 6 mT peak field, Conti et al. (62, 63); 3 Hz sawtooth signal of 4.5 mT, Mooney et al. (64)]. Interchange of exposure apparatus between laboratories suggests that the effect (inhibition or stimulation of proliferation) is dependent on the exposure apparatus and signal used (38). These results are therefore suggestive of small comitogenic effects of EMF on stimulated primary cell cultures, strengthening the case for EMFs as tumor promoters, even though there are major discrepancies concerning the nature of effects (stimulatory or inhibitory). Cadossi (61) further stresses the importance of the cells' origin, pointing out that samples from aged or diseased donors or suboptimally stimulated cells are more responsive to EMFs than other cells. Clearly these results do not yet constitute a well-defined and reproducible magnetic field effect. Conti et al. (62) have also measured the effect of 3 Hz signals on calcium fluxes and report that EMF exposure abolishes increased 45Ca2+ uptake by human lymphocytes stimulated with PHA or TPA. For PHA stimulation, the effects reported are small (PHA stimulation of 45Ca2+ uptake is at most 50% above unstimulated cells), and in only one of three experiments did the effect appear significant. The effect of EMF on TPA stimulation of calcium uptake was more consistent. These results may be compared with other calcium flux studies, reviewed below.
Cain et al. (66) report a copromotional effect of 60 Hz 0.1 mT fields with TPA on cocultures of fibroblasts and transformed daughter cells. A twofold increase in the number of foci, area of foci per dish, and number of cells per focus was seen.
The effects of a variety of 50 Hz magnetic fields (vertical fields of 6 µT to 2 mT, or Ca2+ ion cyclotron resonance conditions, for as many as 21 days) on growth characteristics of the multipotential hemopoietic cell line, FDCP mix (A4), have been studied by Reipert et al. (67, 68; B. M. Reipert, D. Allan, and T. M. Dexter, unpublished results), who report no effect on growth rate, cell cycle state, clonogenic efficiency, or apoptosis.
Effects of melatonin and tamoxifen on the growth of breast tumor cell lines
Harland and Liburdy (69) have reported that continuous exposure to 60 Hz, 1.2 µT EMFs can block the cytostatic action of melatonin and tamoxifen on the human breast cancer cell line MCF-7. The effects were not seen at lower field strengths (0.2, 0.6 µT) and no effect of field exposure without melatonin/tamoxifen treatment was seen. These experiments used a well-defined cell system with a reliable assay (cell number) and simple exposure conditions.
Neurite outgrowth
Blackman et al. (70) have reported that EMF exposure of rat pheochromoctytoma cells (PC12 cells) treated with nerve growth factor can either increase or decrease neurite outgrowth, depending on the subset of PC12 cells used. In view of this variability, it may be concluded that this system is unsuitable for studies of EMF effects.
Intracellular responses
Calcium flux and intracellular free calcium ([Ca2+]i) measurements
Early studies used 45Ca2+ to measure fluxes, notably in brain tissue (71, 72). The effects detected were small (generally 10% over control samples; maximal effects of 30%) and varied significantly between laboratories. For example, Blackman et al. (73) observe that "although the results of exposure of isolated chicken brains to ELF fields as published by Bawin and colleagues and by us agreed both in the frequency dependence and generally in the intensity dependence of the results, there was one major difference. Blackman et al. (72) reported an enhancement in calcium ion efflux whereas Bawin and Adey (71) reported a reduction in efflux."
Transmembrane Ca2+ fluxes control the mobility of Amphora coffeaeformis (diatoms), and it has been reported that diatom mobility can be stimulated by specific combinations of d.c. and low-frequency EMF exposure in a manner consistent with the cyclotron resonance theory of EMF/Ca2+ interactions (74). Even though in principle these experiments are simple to perform, several attempts at replication have thus far failed outright or did not demonstrate consistent effects ( Table 4).
Furthermore, no effect of magnetic fields satisfying the resonance condition for calcium on Ca2+ transport was detected in the membranes of cultured cells (RINm5F-cloned insulin-secreting ß cells; ref 75) or in cytosolic calcium concentrations in a number of cultured cell lines (3T3 fibroblasts, L929, V-79, and rat osteosarcoma cell lines; ref 76). Effects of magnetic fields satisfying the resonance conditions for calcium on Ca2+ transport have been reported for primary human lymphocytes (77, 78), but these have also proved irreproducible (79, 80; Table 4). Other effects of EMF on calcium transport in lymphocytes appear more robust and are discussed below.
|
Increases in [Ca2+]i are essential for the initiation of growth in newly fertilized eggs and, as indicated above, the frequent association of increases in [Ca2+]i with stimulated proliferation of somatic mammalian cells implicates calcium-regulated proteins as crucial components of the growth signaling pathway. It has been postulated that EMFs may modulate cellular calcium regulatory mechanisms by, for example, affecting the function of transmembrane proteins or the affinity of calcium binding proteins such as calmodulin. Lindström et al. (81) have reported effects of 50 Hz, 100 µT magnetic field exposure on spontaneous [Ca2+]i oscillations in Jurkat cells. Using the fluorescent indicator fura 2, [Ca2+]i was measured every 2 s in individual cells. Calcium oscillations began 15–20 s after applying magnetic fields (0.1 mT, 50 Hz, generating electric fields of <50 µV/m), with [Ca2+]i rising from a baseline of 20–100 nM to 200–400 nM, similar to the effect of CD3 receptor complex activation. The largest response was seen with fields of 150 µT (81). EMF-induced [Ca2+]i oscillations were dependent on the presence of an intact CD45 intracellular phosphatase domain, which is also necessary for early events in T cell receptor-mediated signaling and coupling to inositol phosphate signaling pathways. Lindström et al. (82) have subsequently shown that inositol 1,4,5-triphosphate (IP3) levels are also increased in Jurkat cells after short exposures to 0.1 mT, 50 Hz fields. IP3 increased from 2 to 30–40 pmol/106 cells within a few minutes, similar to the response to anti-CD3 antibody, although with a different time course (EMF response maximal at 2 min: 6 min for anti-CD3); EMF response occurred in the presence of intracellular calcium chelators. In the unexposed control experiments, however, levels rose by eightfold to 17 pmol/106 cells. Thus, it is not clear that these experiments are rigorously controlled. The EMF response was not seen in all cells: although 85% of the Jurkat cells responded, only 10% of peripheral blood lymphocytes did so. An independent group has been unable to repeat these [Ca2+]i observations, although the data have been published only as an abstract (83). Walleczek et al. (86, 87) report that only 50% of Jurkat cells showed stable [Ca2+]i levels before exposure; of those `stable' cells, 14% showed deviations from basal [Ca2+]i during EMF exposure, and a similar proportion showed deviation during a sham exposure. Hence, the experimental system appears too unstable to quantify small [Ca2+]i changes.
Walleczek (84) and others have reported effects of short exposures (30 min–1 h) to higher EMFs (2–22 mT) on 45Ca2+ uptake by lymphocytes. Enhanced 45Ca2+ influx was detected only in cells treated with concanavalin A (Con A), was dependent on the magnitude of both the applied magnetic field and the induced electric field, and resulted from increased calcium influx through plasma membrane channels rather than release from intracellular stores (85). The response also depended on the sensitivity of each primary cell culture to the mitogen. Hence, 60 min exposure to 60 Hz sinusoidal fields of 6.5 mT enhanced the response of cells to Con A, increasing Ca2+ uptake over control cells by 3.9-fold, compared to only 2.5-fold in cells treated with Con A alone. When cells were exposed to a 3 Hz rectangular wave signal of 6.5 mT peak field with Con A treatment, EMF exposure inhibited Con A-induced Ca influx in Con A-responsive primary cultures, but introduced a `Con A'-like Ca influx in cultures that did not respond to Con A alone. These effects are similar to those of Conti et al. (62), described above. As the fields used are much higher and in many cases of a different waveform than those of central concern to this review, these results will not be discussed further; nevertheless, the results seen with 3 Hz fields are to some degree consistent between groups (86), are consistent with an effect at the cell plasma membrane, and many reports conclude that the magnitude and type of EMF effect depend on the degree of cell stimulation by other agents.
In more recent work, Walleczek et al. (87) reported a 5% increase in calcium influx (measured by single cell fluorescence spectroscopy and Mn2+ influx) into human Jurkat T cells after 2 min exposure to 60 Hz, 2 mT EMFs. Again, the effect was critically dependent on the biological state of the cells, characterized by the initial (preexposure) Ca2+ influx rate.
Barbier et al. (88) have reported sustained increases in [Ca2+]i in rat pituitary cells at either 25°C or 37°C after exposure to 50 Hz sinusoidal fields of 50 µT (30 min or 3 h). [Ca2+]i rose from ~180 nM (range 80 to 314 nM) to ~350 nM (range 109 nM to 4267 nM). No effects were seen at 2, 10, or 250 µT. [Ca2+]i was assayed by indo-1 fluorescence; after exposure, individual cells were identified by double label immunochemistry. Only the lactotrophs (29% of pituitary cells) were responsive. In the absence of extracellular Ca2+, the [Ca2+]i increase was (on the basis of one experiment shown) greatly reduced but still significant (a two to threefold rise). The major component was thus ascribed to Ca2+ influx via voltage-dependent, dihydropyridine-sensitive channels (inhibited by PN 200–110). Double-wound coils were used, and in sham-exposed controls there was no change in [Ca2+]i. [Ca2+]i was measured before and after field exposure in these experiments, although continuous monitoring would have been preferable and feasible.
Ornithine decarboxylase
ODC activity is significantly stimulated (to high levels from almost undetectable levels in control cells) by treatment with phorbol ester and is elevated in all rapidly growing cells, including transformed cells and cells isolated from tumors. Stimulation of ODC activity has been detected in cultured human lymphoma cells (fivefold over controls) and mouse myeloma cells (1.7-fold over controls) after 1 h exposure to 1 Vm-1 electric fields at 60 Hz (89). Stimulation was transient, the maximum occurring 1–2 h after termination of exposure. However, no effect of 50 Hz, 30 µT fields on ODC activity was detected in human amniotic cells (46). More recently, Litovitz et al. (90, 91) reported a large increase (twofold and greater) in ODC activity in L929 cells exposed to 60 Hz magnetic fields of 10 µT for 4 h. Azadniv and co-workers (92) attempted to repeat these experiments and found no effect of field exposure on ODC activity, although the cells used could be shown to respond to the phorbol ester TPA, yielding a large increase in ODC activity. The researchers showed some variation in the measurement of ODC activity and, by using statistical methods, demonstrated that this was due to interexperimental variation and did not correlate with EMF exposure.
Uridine uptake and incorporation
Numerous experiments have indicated that EMFs stimulate both the uptake of uridine by cells and its incorporation into RNA. Greene et al. (93) reported a 50–60% increase in uridine incorporation in human leukemic (HL60) cells after exposure to 1 mT fields at 60 Hz for 1 h. However, Azadniv and Miller (94) repeated these experiments and found no effect of EMFs on uridine incorporation. Phillips et al. (95) also reported two- to threefold increases in uridine incorporation after treatment with pulsed fields of 3.5 mT peak strength, which were maximal after 2 h and remained elevated for 22 h. Goodman, Henderson and co-workers have reported the effects of a variety of fields (60 and 72 Hz sinusoidal fields and several pulsed waveforms) on the incorporation of tritiated uridine into nascent RNA on polytene chromosomes of Dipteran insect salivary glands. Fractionation of pulse-labeled total cellular RNA suggested that incorporation was into mRNA. These experiments could be taken to suggest that general RNA synthesis can be stimulated by EMF exposure, but the exquisite sensitivity of uridine transport to cell perturbation (96) indicates that the experiments require further verification.
Early response genes: FOS and JUN
Nuclear run-on assays used to measure gene transcription in T lymphoblastoid cells exposed to 100 µT 60 Hz fields for 15–120 min indicated that FOS, JUN, and MYC transcription may be modulated by EMF exposure (95). FOS transcription was induced maximally (2.5-fold) after 30 min of exposure and returned to control levels by 60 min. MYC was also increased, reaching a plateau of 1.5- to 2-fold control levels after 30 min. JUN transcription was affected in a cell density-dependent manner, being decreased by 70% after 30 min at 5 x 105 cells/ml, but increased by 2.2-fold after 60 min at 1 x 106 cells/ml. In both cases, expression returned to control levels after 24 h. PKC expression was also altered by EMF exposure, being maximally stimulated after 15 to 30 min. These data are in many ways consistent with the hypothesis that EMFs activate a mitogenic signaling pathway. FOS induction was rapid and transient, with approximately the same time course that follows mitogenic stimulation. MYC expression also followed a time course similar to that seen in mitogenically stimulated cells, though the magnitude of the effect (2- to 2.5-fold over control levels) was small compared with that seen in mitogenic activation. However, in contrast to conventional practice, in this study cells were exposed to the putative mitogen (EMFs) while in the logarithmic phase of growth in the presence of 10% serum. This will give relatively high levels of transcription against which any EMF-induced effects must be detected, although it is, of course, arguable that cells that are already partially activated may be ideally primed to respond to a further weak stimulus.
The effect of EMFs on early response gene expression (Fos, Jun, and JunB) in a murine hematopoietic progenitor cell line [FDCP mix(A4)] transfected with a Bcl-2 expression vector has been studied. BCL-2 synthesis maintains the cells as a quiescent but viable line in the absence of interleukin 3 (IL-3). Cells were stimulated by addition of IL-3 at a suboptimal concentration for the transcription of early response genes expression, and cells were exposed to 6 µT 50 Hz sinusoidal fields for 20 min to determine whether there were any additive or synergistic responses. No significant effects on Fos, Jun, or JunB transcription were detected (67).
MYC
The most extensive series of investigations of EMF effects on gene expression in human cells has been the joint studies of Goodman and Henderson (97–105). They have reported increases in MYC, ß-actin, and histone H2B transcript levels of up to threefold when measured by dot blot analysis in HL60 cells after exposure for 20 min to any one of five different EMF signals, including 1.1 and 1.5 mT sinusoidal magnetic fields at 60 and 72 Hz, respectively, and three pulsed fields of the form used clinically to treat non-union bone fractures (97). Exposure for 20 min to a 45 Hz sinusoidal magnetic field with an rms value of 0.8 mT increased MYC and histone H2B mRNA levels by sevenfold; a 5.7 µT, 60 Hz field caused a fourfold increase in mRNA levels, with transcription of ß-actin, hsp 70, SRC, and 
-tubulin being stimulated to the same extent as MYC and histone H2B (98, 99). At other field strengths (0.57, 57, and 570 µT), mRNA levels were increased by two- to threefold. Other waveforms at frequencies of 15 to 150 Hz stimulated gene expression by approximately twofold, although it is not possible from these experiments to obtain a clear picture of the relationship between field strength or frequency and stimulation of gene expression, since both signal parameters were changed between experiments. Increased transcription was detected after a 4-min field exposure, being maximal at 20 min and returning to control levels after 4 h exposure (100).
Closely similar patterns of gene expression stimulated by EMFs in six different types of eukaryotic cells have also been reported by Goodman and Henderson [<1 h exposure, 60 Hz, 5.7 µT–57 µT (98, 101–104), 1.1 mT (97, 105)]. The responses included activated gene expression in yeast (104), mouse myeloma cells (103), and three types of human cells (98, 102, 103), as well as changes in protein synthesis in Sciara (101, 105), with no reports of cell types that were insensitive to ELF-EMFs.
Several conclusions were drawn by Goodman and Henderson on the basis of these results. 1) A subset of genes including MYC, ß-actin, histone H2B, hsp 70, SRC, and -tubulin were EMF responsive, showing closely similar time courses and quantitative increases in transcription. Only one gene—ß2-microglobulin—was nonresponsive to EMFs. 2) Induction was rapid (detectable within 4 min) and transient; transcript levels were maximal after 20 min exposure and returned to control levels within 4 h of first exposure. 3) Responses were frequency and field strength dependent, the largest response consistently reported as occurring after 20 min exposure to a 5.7 µT sinusoidal magnetic field at 60 Hz. 4) Quantitatively similar patterns of EMF-stimulated gene expression occurred in at least six different types of eukaryotic cell.
In view of the well-established individuality of genes in terms of the temporal and quantitative nature of their responses to cell activating signals, it seems improbable that so many genes in such a diverse set of cell types would respond almost identically to stimulus by EMFs. Nevertheless, the central role played by MYC in controlling cell proliferation and apoptosis and the fact that MYC expression is anomalously regulated in the majority of human cancers imbued these findings with great significance; rigorous attempts have been made in two independent laboratories to reproduce the findings of Goodman and Henderson (106–109).
Each of these independent studies comprised two main components: 1) very close reproduction of the experimental conditions used by Goodman and Henderson, and 2) the design of systems and protocols to achieve maximum sensitivity in the identification of any EMF-induced effect on the transcription of MYC and other genes, together with maximum reproducibility of data.
One of the replicative studies (107) used exposure apparatus and signal generators built to the same specifications as those of Goodman and Henderson and closely followed their protocols for cell handling, field exposure, RNA preparation, and in some experiments, the exact methods for quantification of RNA transcripts. The other study (109) included experiments performed in Henderson's laboratory while using their exposure apparatus and cells.
Further experiments in both replicative studies refined several aspects of the original work of Goodman and Henderson by using more controlled experimental conditions and accurate methods to quantify gene transcripts, which were accurate to within 10%.
Neither replicative study (106–109) detected any effect of ELF-EMF exposure on gene expression for exposure periods of up to 1 h to 60 Hz fields of 0.1 to 100 µT. These findings directly contradict the multiplicity of reports generated by Goodman and Henderson (98–100). The following points have been raised to account for the irreproducibility of the findings by Goodman and Henderson, but none offer escape from the dichotomy. The issues have been analyzed previously (106–109) and are briefly summarized here.
1. HL60 cell phenotype
EMF responsiveness may be critically dependent on the phenotype of the subculture of cells used. The HL60 cells used in both replication studies were obtained from established cell culture collections. Both showed typical patterns of HL60 cell-surface antigen expression and both responded normally to the stimulation of differentiation by TPA. Furthermore, experiments by Saffer (108, 109) in New York, and subsequently in his own laboratories with cells obtained from Henderson (109), showed no EMF response. However, Goodman and Henderson (97–105) have reported responses to EMFs in six different types of eukaryotic cells as well as in HL60 cells, with no reports of cells that are insensitive to EMFs. It is therefore highly improbable that both replication studies selected unresponsive HL60 cell sublines, whereas Goodman and Henderson consistently selected responsive cells from a wide variety of sources for the diverse eukaryotic systems studied.
2. Field exposure
The electric fields and currents induced in cell suspensions by a given magnetic field vary considerably with the orientation and type of exposure system; hence replication studies may not reproduce critical parameters of the original EMF signal. This was not relevant in these two replication attempts. Saffer conducted experiments in Henderson's laboratory using their equipment, and experiments were performed by Lacy-Hulbert et al. (106, 107), who used exposure coils made to Goodman and Henderson's (98) specifications. Furthermore, Goodman and Henderson have reported that a wide variety of waveforms, field strengths, and exposure systems caused large changes in transcription.
3. Geomagnetic fields
Local magnetic fields may affect EMF responsiveness. In the first experiments reported by Goodman and Henderson, no magnetic shielding was used and background geomagnetic fields at the sample position were reported as 46 µT (Goodman laboratory; ref 97) and 8.2 µT (Henderson laboratory; ref 98). Later experiments used mu-metal shielding, and background geomagnetic fields were <0.1 µT (110, 111). The response of cells to ELF-EMFs therefore was not dependent on a large geomagnetic field component. In both replication studies, background magnetic fields were reduced by mu-metal shielding to <0.1 µT.
4. Magnetite
The presence of precipitates of the ferromagnetic mineral magnetite (Fe304) may contribute to in vitro EMF effects (112). It is possible that biological responses to EMFs are mediated by ferromagnetic minerals; magnetite crystals have been implicated in the mechanism by which migratory birds and bees detect the Earth's magnetic field (113). Hence, the absence of magnetite in any of the studies might prevent an EMF response. However, magnetite is omnipresent in laboratory culture reagents and plasticware, and no precautions against magnetite contamination were taken in any of the studies; therefore, whatever role it may play in EMF-induced effects, magnetite cannot explain the failure to reproduce the Goodman and Henderson data.
5. RNA extraction procedure
Selective extraction may arise from the use of different procedures. It has been suggested that detection of EMF-induced effects may depend on the extraction methods used (114). One replicative study tested both guanidine extraction (extracting total cellular RNA) and the SDS/phenol extraction method used by Goodman and Henderson (extracting cytoplasmic RNA), without significant difference. The other study used the cytoplasmic RNA preparation protocol described by Goodman and Henderson.
6. Pretreatment of cells
The immediate prehistory of the cells is critical. The Goodman and Henderson protocol assayed gene expression only 1 h after cell centrifugation and resuspension in fresh medium. However, as it is established that centrifugation and associated manipulations greatly perturb cell metabolism and transcription, the most probable effect of such a method would be to increase the variability of the results. The replicative studies repeated the 1 h preincubation protocol, but also used a 12 h preincubation in a controlled field environment, together with cell harvesting procedures that eliminated exposure to uncontrolled, ambient fields.
Consideration of these points indicates that it is improbable that the failure of two independent laboratories to reproduce the Goodman and Henderson findings arises from inadequacies in the replicative procedures. We conclude that the most convincing e
(PLC
(PKC
can excite electron transitions between levels, allowing transition to the singlet states even in high magnetic fields. The figure also shows the expected effect of increased static magnetic field on a chemical reaction (adapted from ref