How might animals detect the magnetic field?

Analogous to vision, touch, or hearing, it is predicted that the magnetic sense relies on a specialized population of cells that are associated directly, or indirectly, with the central nervous system [12] We expect that these cells express a suite of molecules that are necessary for the transduction of a magnetic stimulus into a cellular response. We anticipate that this involves the modulation of intracellular signaling cascades, which in turn mediates neurotransmitter release, the activation of secondary neurons, and the subsequent integration of information within specific circuits of the central nervous system [13]. But, how exactly would the transduction process work? Currently, there are 3 dominant hypotheses that are being tested. The first predicts that magnetic fields are detected by a mechanically sensitive magnetite-based magnetoreceptor. The second hypothesis relies on a light-sensitive, chemical-based mechanism, and the third idea is dependent on an anatomical structure that would enable electromagnetic induction. It should be emphasized that these concepts are not mutually exclusive, and that animals may have evolved multiple mechanisms to detect different components of the field.

A mechanically sensitive magnetite-based magnetoreceptor

In many sensory systems, the receptor proteins involved in transducing the signal are directly influenced by the incoming stimulus and undergo structural rearrangements upon activation. For instance, it is known that the mechanosensitive channel Piezo1 directly senses forces at the plasma membrane, resulting in cation influx in response to pressure [14,15]. Might magnetosensation rely on a similar mechanism? There are 2 issues that arise here. First, the Earth's magnetic field is weak, and secondly, proteins exhibit very low magnetic susceptibility because they are primarily composed of carbon, nitrogen, and oxygen. If a membrane protein is to undergo a conformation change in an Earth strength field, it would need to be coupled to a ferrimagnetic structure made of an iron oxide such as magnetite (Fe₃O₄) [16]. This conceptually simple idea is tenable because it is known that a number of species are able to form biogenic magnetite. The best example are magnetotactic bacteria, which generate a chain of intracellular magnetite crystals [5,17]. They employ this internal compass needle to guide their swimming along the incline of the magnetic field vector to deeper waters with favorable redox conditions. Such observations have led to the formulation of the magnetite-based hypothesis for magnetoreception, which predicts the existence of a mechanosensitive channel attached to magnetite crystals (Fig 2A) [18,19]. This has motivated many researches to search for magnetite in vertebrates, employing methods such as superconducting quantum interference device SQUID magnetometry, energy filtered transmission electron microscopy (EFTEM), magnetic purification techniques, and the histological stain Prussian Blue (which labels ferric iron [Fe³⁺]). To date, this search has not been marked by success. Macrophages have been mistaken for magnetoreceptors [20,21], iron contamination from the laboratory environment has led investigators astray [22–24], and the ability to locate nanometer size crystals in a whole organism has proved an extremely challenging task [25]. An alternative approach has been to employ bioinformatic methods to search for eukaryotic orthologues of the well characterized Mam proteins that are required for magnetite synthesis in magnetotactic bacteria. Although some domain structures are conserved (e.g., MamP has a PDZ domain), the available sequence data suggest there are no genuine homologues of the Mam proteins in eukaryotic magnetosensitive species [17,26,27]. One thing is clear, if indeed a magnetite-based magnetoreceptor exists, it's not going to be easy to find.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Three mechanisms proposed to underlie the magnetic sense.

(A) Image depicting a mechanically sensitive magnetite-based magnetoreceptor. Magnetite crystals (shown here as a chain) are attached to the plasma membrane via a cytoskeletal linker. This linear arrangement of single-domain crystals attempts to align with the Earth's geomagnetic field (like a compass needle), and thereby exerts a torque force on a mechanosensitive channel (shown in teal). This transiently activates the channel leading to cation influx (red arrow) and membrane depolarization. (B) Diagram showing a light-sensitive chemical-based magnetoreceptor. Blue light (shown with a blue arrow) induces the formation of long-lived radical pairs between Cry and the cofactor FAD. The spin state of these electrons interconverts between an antiparallel (↑↓) or parallel (↓↓) state, depending on the local magnetic environment. This, in turn, influences the biochemical or structural properties of Cry, resulting in the activation of an unknown signaling molecule (X) that modulates ion channel permeability. (C) Magnetoreception based on electromagnetic induction. This hypothesis relies on an accessory structure that transforms magnetic stimuli into an electrical information. Depicted is one semicircular canal of a vertebrate, filled with cation-rich endolymph, and sensory cells located on either side of the cupula (shown in orange). If the animal moves so that rotation occurs around an axis in the plane of a semicircular canal, there will be no displacement of the endolymph but electromagnetic induction could occur. Depending on the intensity and orientation of the external magnetic field, this will induce an electromotive force in the conductive endolymph. This results in the separation of charges within the circuit, inducing cation influx through highly sensitive voltage-gated ion channels (shown in teal). Cry, cryptochrome; FAD, flavin adenine dinucleotide.

https://doi.org/10.1371/journal.pbio.2003234.g002

A light-sensitive chemical-based compass

Alternatively, it is conceivable that the geomagnetic field could influence biochemical reactions within a receptor cell analogous to detection of light by visual pigments. Rhodopsin, the best characterized photoreceptive protein, is reliant on a chromophore (11-cis-retinal), which undergoes a structural change on the absorption of a photon. This alters the biochemical properties of the opsin, resulting in an activation of G-proteins, and alteration of downstream signaling cascades [28]. Behavioral data in European robins have demonstrated that magnetic orientation requires light in the blue/green spectrum, supporting a light-dependent hypothesis for magnetoreception [29]. This concept predicts that the geomagnetic field can influence the spin state of light-induced radical pairs and thereby modulate the outcome of biochemical reactions [30–32]. This is possible because electrons possess a quantal property known as spin angular momentum, and consequently radical pairs can be ordered in a parallel (↓↓) or antiparallel state (↑↓). It is known that external magnetic fields can influence the interconversion between these different spin states, which in turn could influence the reactivity of a particular molecule [33,34]. Consistent with the hypothesis, a number of behavioral studies in vertebrates have shown that low-intensity broadband electromagnetic fields (which influence electron spins) disrupt magnetic orientation [35]. It has been proposed that the cryptochromes (Crys), which have a conserved photolyase domain that binds the flavin adenine dinucletotide (FAD) cofactor, possess the molecular attributes to function as a magnetosensor. According to the favored model, a long-lived radical pair is formed when light induces the transfer of electrons to this FAD moiety along a triad of tryptophan residues in the Cry molecule (Fig 2B) [31,36].

Genetic studies in invertebrates have supported the role of Cry in magnetically sensitive behavior. For instance, Fedele and colleagues have shown that low frequency electromagnetic fields (300 μT, 50 Hz) shortened the circadian period and induced hyperactive behavior in Drosophila, which is abolished in the absence of Cry [37]. Although these results indicate that Cry is required for magnetic fields to influence certain behaviors in flies, it is questionable whether these animals possess a true magnetic sense analogous to migratory birds. Furthermore, given that Crys are expressed in all major organs in birds and are known to play a role in regulating circadian phenotypes in vertebrates, it is currently unclear how specificity would arise with respect to magnetic information and how it would be distinguished from a circadian input [38,39]. Moreover, we still don't know which cells are associated with a light-dependent chemical compass. Although it is widely assumed that the primary sensors reside in the retina, no report has convincingly shown that a particular cell type (e.g., cones) is sensitive to magnetic stimuli. While we have good reason to believe a radical-pair based magnetic sense exists, it is apparent that we still understand little with respect to the receptive cells, the molecules, and the downstream signaling cascade.

An anatomical structure that would enable electromagnetic induction

Finally, the detection of magnetic fields might rely on accessory structures that convert the magnetic field into another stimulus. For instance, terrestrial vertebrates do not detect gravity directly, but rather rely on a bed of calcium carbonate otoliths which apply a force on a population of sensory hair cells [40]Similarly, the auditory system exploits a complex array of accessory structures that convert airborne sound pressure waves into gating spring tension on mechanosensitive ion channels in hair cells [41] As magnetism and electricity are inseparable, it is conceivable that a secondary structure might convert information about the Earth's magnetic field into an electric stimulus. This proposition, which reflects Faraday’s law of electromagnetic induction, predicts that movement of an animal in a fixed magnetic field would induce an electromotive force in a conductor [42]. If a circuit exists, a change in electric potential could be detected by cells expressing voltage-sensitive channels. This idea has primarily been considered with respect to aquatic animals who reside within a conductive medium (i.e., the ocean) and possess electroreceptors that are sensitive to electric stimuli as small as 5 nVcm⁻¹ [43–45]. However, it is conceivable that electromagnetic induction may also occur within the endolymph of the semicircular canals of birds, with hair cells on the periphery of the cupula functioning as electroreceptors [46] (Fig 2C). If a pigeon were to rotate its head around an axis in the plane of a semicircular canal, there will be no displacement of the endolymph but electromagnetic induction could occur. Conversely, rotation around an axis perpendicular to the plane of the semicircular canal would induce fluid flow and provide information with respect to angular momentum, but would not induce electromagnetic induction [46]. In this way, inertial and electromagnetic input could be distinguished from one another. It is conceivable that the sensitivity of such a magnetoreceptor could be tuned by manipulating the concentration of cations within the endolymph, thereby altering its conductivity. We have recently discovered an iron-rich organelle, the "cuticulosome," which resides apically in avian hair cells and is composed primarily of ferritin nanoparticles [47]. Intriguingly, this structure is associated structurally and molecularly with vesicular trafficking, prompting us to propose that it may regulate the concentration of iron within inner ear compartments.

Although magnetosensation by electromagnetic induction was proposed as early as 1882 by the Frenchman Camille Viguier, there is limited behavioral, anatomical, physiological, or experimental data that directly supports it [48]. Perhaps of most relevance is the work of Dickman and colleagues who have reported that cells within the vestibular nuclei of pigeons are responsive to magnetic stimuli of a particular inclination, polarity, and intensity [49,50]. Given these data and the astonishing electrosensitivity of multiple different species on the planet, in our opinion, electromagnetic induction is an idea that deserves a closer look [51,52] (Box 2).

How to solve the mystery

Let’s imagine a laboratory situation in which a team of researchers aims to understand how vertebrates detect and use magnetic information. For such an endeavor to be successful, we think 2 things are imperative: (1) an interdisciplinary approach, and (2) a model system with a reliable magnetoreceptive assay that is amenable to genetic modification. If ever a scientific problem demanded the efforts of multiple disciplines, it is magnetoreception. It requires a knowledge of the physical sciences, increasingly at a quantal level, to understand the nature of the stimulus and appreciate how it can interact with biological matter. Expertise in the behavioral sciences is required to establish informative assays that exclude other sensory stimuli and enable investigators to ascertain what component of the magnetic field is influential. A detailed characterization of the neuroanatomical circuits is imperative if we are to understand where magnetic information is processed in the central nervous system. This will need to be coupled to physiological methods, such as calcium imaging, if we are to appreciate how this information is encoded and subsequently integrated with other sensory cues that are vital for navigation. The identification of the primary sensory cells, which has proved to be particularly difficult, will need a broad range of techniques including elemental analysis, electron microscopy, immunohistochemistry, live-cell imaging, and ablation studies.

Whether the underlying mechanism relies on magnetite, light, or electromagnetic induction, it is likely that the problem will then coalesce around a particular intracellular signaling pathway or molecule. At this point, the critical question will be whether a particular protein is necessary and/or sufficient for magnetoreception. To establish an iron-clad causal relationship between a molecule and a behavioral output, genetic loss (or gain) of function studies will be necessary. With the advent of the Clustered Regularly Interspaced Short Palindromic Repeats-Cas9 (CRISPR-Cas9) genome editing system, this is now a feasible option in almost any species that can be bred in captivity [54]. Although the generation of germline genetic perturbations may still be time consuming due to the reproductive cycle of a particular species, the creation of somatic genetic perturbations by in ovo or in utero electroporation is realistic in the short term [55]. Moreover, such an approach will enable tissue-specific genetic ablation that will be important if a candidate molecule (e.g., Cry 4) is ubiquitously expressed. The CRISPR-Cas9 editing system is elegantly complemented by astonishing advancements in next generation sequencing, such as the PacBio system, that enable fast, cheap, and high-quality DNA sequencing in any species. These tools, which are rapidly decreasing in cost, promise to open new experimental vistas in non-model systems. The challenge is to couple these revolutionary technologies to an informative readout, whether it be behavioral or physiological. With such an approach we are confident that this mystery, which has steadfastly refused to reveal its secrets, will succumb.