Neuroanatomy of Magnetoreception: The Superior Colliculus Involved in Magnetic Orientation in a Mammal

See allHide authors and affiliations

Science  12 Oct 2001:
Vol. 294, Issue 5541, pp. 366-368
DOI: 10.1126/science.1063351


The neural substrate subserving magnetic orientation is largely unknown in vertebrates and unstudied in mammals. We combined a behavioral test for magnetic compass orientation in mole rats and immunocytochemical visualization of the transcription factor c-Fos as a marker of neuronal activity. We found that the superior colliculus of the Zambian mole rat (Cryptomys anselli) contains neurons that are responsive to magnetic stimuli. These neurons are directionally selective and organized within a discrete sublayer. Our results constitute evidence for the involvement of a specific mammalian brain structure in magnetoreception.

Behavioral studies have provided abundant evidence for magnetic compass orientation among vertebrates, but its sensory and neural basis remains enigmatic (1, 2). A few electrophysiological studies have addressed the involvement of a specific brain structure in the processing of magnetic information (3–9). This method, however, has a particular drawback: It does not allow systematic screening of neuronal activities in the central nervous system. Therefore, well-aimed electrophysiological studies cannot be conducted in the absence of a known receptor site. Here, we investigated magnetoreception by combining two established methodological approaches: a behavioral test designed to assess magnetic compass orientation in mole rats (10,11) and immunocytochemical visualization of the transcriptional regulatory protein c-Fos as a marker of neuronal activity, a neuroanatomical technique used extensively in sensory research (12–14).

We detected the evoked expression of c-Fos in order to map neuronal activities that had been entrained either by active orientation via the magnetic compass or by changes in the ambient magnetic field. Experimental animals built nests in an unfamiliar arena [i.e., performed a magnetically based spatial orientation task (15)] under different test conditions (16). Controls (used also to assess basal levels of c-Fos expression) were of two types: (i) untreated animals freely moving within a familiar home area, and (ii) animals resting or sleeping in a shielded magnetic field. We focused on neuronal activities in the superior colliculus (SC), a prominent subcortical sensorimotor integrator that plays an important role in orientation to diverse stimuli (17–19). The unique intrinsic circuitry of the SC (20) may serve to integrate magnetic information with multimodal sensory and motor information. Magnetic stimuli thus may directly elicit orientation responses via initiation of activity in the premotor efferent collicular pathways.

The SC in all of the experimental and control animals displayed a symmetrical bilateral distribution of c-Fos immunoreactivity (21) (Fig. 1, A to F). The density and distribution pattern of immunoreactive cells, however, differed markedly between animals subjected to different experimental conditions [Figs. 1 and 2; Web figs. 1 to 3 (22)]. In control animals resting or sleeping in the shielded magnetic field, basal expression of c-Fos was very low and only a few randomly scattered immunoreactive neurons were observed [Fig. 1F; Web fig. 1F (22)]. In control animals active in their home cages, c-Fos expression always remained moderate; labeled cells were more numerous than in the SC of resting animals, but were still randomly scattered and devoid of any obvious pattern of alignment [Fig. 1E; Web fig. 1E (22)]. In contrast, the SC of experimental animals invariably showed large numbers of heavily immunostained cellular nuclei, arranged in tangentially oriented bands aligned with collicular layers or sublayers [Fig. 1, A to D; Web fig. 1, A to D (22)].

Figure 1

Characteristic distribution patterns of c-Fos–immunoreactive neurons in the SC of Zambian mole rats subjected to different experimental conditions. (A to D) Nesting in unfamiliar circular arena: (A) natural (constant) magnetic field; (B and C) experimental magnetic field, the horizontal component of which was manipulated every 5 min (B) and every second (C); (D) shielded magnetic field. (E) Movement within home cage, natural magnetic field. (F) Inactivity, shielded magnetic field. Each dot represents a single labeled neuronal nucleus.

Figure 2

Mean numbers of c-Fos–immunoreactive (IR) neurons (±SEM) in the InGi (30). Density counts are from the rostral (a), middle (b), and caudal (c) parts of the SC; see Fig. 1 for coding of experimental conditions A to E. Significant differences (P < 0.05) are indicated by stars. Solid stars indicate comparison with control group E; open stars indicate comparison with adjacent experimental group on the right or with another experimental group when so indicated. Upper right inset shows position of individual counting frames; lower left inset shows approximate levels of counting.

In both the outer sublayer of the intermediate gray layer (InGo) and the deep gray layer (DpG), the density of immunopositive cells was increased significantly in all experimental groups, irrespective of experimental conditions [Web figs. 2 and 3 (22)]. Density differences between experimental groups were, however, much less pronounced and mostly statistically insignificant. The distribution pattern of labeled cells was common to all experimental groups (Fig. 1, A to D). Therefore, increased neuronal activity in those two layers is likely an unspecific novelty response to the unfamiliar environment (information input from nonmagnetic sensory and motor systems likely accounted for the activation).

In contrast, in the inner sublayer of the intermediate gray layer (InGi), both the density and distribution pattern of immunoreactive cells correlated significantly with physical properties of the magnetic field (Figs. 1 and 2). This sublayer is distinguished by rather irregularly distributed cells and by the frequent occurrence of large, dark multipolar neurons in Nissl- or Klüver-Barrera–stained sections [Web fig. 4 (22)]. In animals building their nests in the constant magnetic field, very strong but focal immunoreactivity was detected in the mediorostral part of the SC [Fig. 1A; Web fig. 1A (22)]. The extent of the labeling was about 300 to 600 μm × <1000 μm (rostrocaudal × mediolateral dimension). Tightly packed, darkly labeled neurons were distributed within the InGi in a patchy manner, forming two or three clusters. Animals building their nests in the periodically changing magnetic field exhibited comparatively weaker staining [Fig. 1, B and C; Web fig. 1, B and C (22)]. The area of labeled neurons, however, was significantly larger, spanning almost completely the InGi throughout the rostral half of the SC (about 1200 μm × 2000 μm, in rostrocaudal × mediolateral dimension). More widely spaced immunopositive cells formed five or six clusters that were less compact and consisted of a smaller number of labeled cells. Finally, in animals building their nests in the shielded magnetic field, only a few scattered immunoreactive cells were found in the InGi [Fig. 1D; Web fig. 1D (22)].

Neuronal activation within the InGi was related to the presence of a perceptible magnetic field of about the strength of Earth's field. Changes in the polarity of the magnetic field led to the activation of increasing numbers of collicular compartments, but activation within individual compartments was less pronounced, as indicated by both lower intensity of immunostaining and lower density of immunoreactive cells.

Because mole rats use a polarity compass for orientation (10), one would expect that the change of field polarity stimulates their magnetosensory system. We therefore expected increases in c-Fos expression proportional to the frequency of polarity changes. However, this was not the case: In trials with changing polarity, the intensity of expression decreased, whereas the area involved expanded. We propose the following explanation for this phenomenon: (i) The presence of a magnetic field of a given polarity, rather than a change of polarity, represents a relevant stimulus; (ii) neurons of individual collicular compartments respond only to magnetic fields with a distinct range of polarity; and (iii) compartments responding to different field polarities (or, under natural conditions, to different orientations of the animal toward the polarity) are distributed systematically within the InGi. Such an arrangement can account for spatial and temporal segregation of neuronal activities when magnetic field polarity is periodically manipulated, and thus it can account for periodicity of excitatory influence on neurons of individual compartments. It seems likely that, under such conditions, only smaller numbers of neurons per compartment could attain a level of activity above the threshold required for immunocytochemical detection. Our findings thus suggest that, as in the case of other sensory modalities (20), the magnetosensory input is also organized in a topographical map of external sensory space within the mole rat SC. Experiments with immobilized animals are needed to provide direct evidence for the existence of such a magnetotopic map.

Our data show that the SC of mole rats contains populations of neurons that are responsive to magnetic stimuli and that it is involved in the neural processing of magnetic information. As such, our work offers experimental evidence that a specific brain structure serves neural processes underpinning magnetic compass orientation in mammals. Our experiments also have methodological implications. Detection of immediate early gene expression may be useful for identifying neurons that have been activated by magnetic stimuli. Such a method could be used to screen for neuronal activities throughout the central nervous system.

  • * To whom correspondence should be addressed. E-mail: pgnemec{at}


View Abstract

Stay Connected to Science

Navigate This Article