The ampullae of Lorenzini (AoL) found in sharks, skates, and rays were rst discovered by Marcello Malpighi in 1663; they were described in detail by Stephano Lorenzini in 1678, and bear his name. Anatomically the AoL are relatively simple structures. A jelly- lled canal projects through a pore in the skin. The surface pores of the AoL on the face of a shark give it the appearance of having acne. The canal jelly has a high conductivity due to its high content of Cl- and K+ ions; it is probably secreted by the cells that line the pore canal and that surround the sensory cells. The length of the canal is species dependent, but is on the order of 1 to 2 mm. Its diameter
is about 150 "m. The inner end of the canal opens up into a sacklike chamber about 400 "m in diameter; the chamber is lined with a single layer of epithelial cells, some of which are the electrosensory cells; others are support cells. The electrosensory cells have no cilia or projections into the gel- lled cavity (Murray, 1965). Figure 2.5-1 illustrates a schematic section through an AoL. One myelinated sensory nerve is shown for simplicity. About six myelinated nerves innervate each ampulla. Each nerve provides two to three synapses per sensory cell, and innervates several cells. The synapses have a characteristic tight "ribbon and gutter" morphology, not unlike the synapses of retinal rods, or cochlear hair cells (Murray, 1965). The synapses are evidently chemical in nature, and the bases of the electrosensory cells contain presynaptic vesicles. The myelinated ampullary nerve bers run together to the anterior lateral line ganglion at the base of the brain.
Murray (1965) examined the electrophysiological properties of single AoL from the skate, Raja clavata. In the absence of external current passing through the canal into the AoL, a single AoL nerve would typically re at about 19 pps in the steady state with a regular rhythm. Murray (1965) gives an interval histogram of the zero-current spikes from an ampullary nerve ber, showing that its period is quite regular. If an external current on the order of nanoamps was passed in or out of the AoL canal, there was a nearly linear relation between the frequency of ring of the nerve and the current. This relation is approximately: f = fo + KI. f represents the frequency of the nerve in the rst quarter second follo wing application of the current. fo was 19 pps, and K was 16.5 pps/nA. The instantaneous frequency of the AoL nerve showed adaptation to steps of applied current, and rebound when it was switched off. The ring decreased to zero for inward current greater than 1.15 nA, and increased nearly linearly for outward current up to about 3 nA. Above 3nA outward current, nerve ring w as inhibited (Murray, 1965). From Murray's gures, the author estimated that the threshold canal current to in uence the AoL "clock" frequency should be on the order of ±100 pA. (Note that British sign convention for current ow assumes that current is carried by negative charges; U.S. sign convention assumes that current is carried by positive charges. In copper wires in the U.S., the electrons travel in the opposite direction from current. Thus, Murray's outward current can be visualized as being carried by Cl- ions moving out of the canal, or curiously, by positive ions [e.g., K+, Na+] moving inward.)
Compared with the sh's skin and the canal lining, the canal, sensory cells, and sensory epithelium have a relatively high conductance, directing the minute ionic currents associated with an external eld to o w through the canal and through the apical region of the sensory cells. At the molecular level, the electrosensory cells of the AoL may release neurotransmitter because voltage-gated transmembrane proteins allow Ca++ ions to enter and depolarize them (Adair et al., 1998).
Kalmijn (1998) reported that an AoL, when electrically stimulated, sources a dc current in opposition to the inward, stimulating current (U.S. sign convention). This reaction current could be carried by a variety of ions, and represents negative feedback, as far as the excitatory potential on the electrosensory cell is concerned. This opposing current could come from an electrically activated ion pump, or be from controlled outward K+ leakage or, equivalently, from controlled inward Cl-leakage. Its purpose may be to act as an automatic gain control for the AoL, giving it a log-linear sensitivity characteristic. Kalmijn (1998) claimed threshold sensitivities for shark AoLs (operating as an array in vivo) can be as low as 1 to 2 nV/cm (0.1 to 0.2 rcV/m). This is an incredible sensitivity!
Behavioral studies of shark electroreception by Dijkgraaf and Kalmijn (1962), and Kalmijn (1966; 1971; 1973; 1974) have shown that sharks use their AoL as low-frequency electric eld sensors to locate prey, and as input devices for geomagnetic navigation. Threshold sensitivity for AoL electroreceptors in the dog-sh shark, Scyliorhinus canicula, has been estimated by behavioral experiments to be 100 rcV/m. The skate R. clavata can sense an amazing 1 ^V/m. This sensitivity is not for one AoL, but rather for the whole array of electrosensors on a free-swimming sh. To obtain these gures, sharks and skates were conditioned to nd and eat an injured at sh hidden on the ocean oor . They did this by using the at sh' s electric eld from muscle potentials. Scent w as controlled for, and was not a factor. Then anthropogenic electric elds of kno wn strength were set up on the ocean bottom with buried electrodes; the sharks and skates mistook the dc electric eld for an injured sh and w ould attack the electrodes.
That sharks, skates, and rays (elasmobranch sh) ha ve a real sixth sense (elec-troreception) that they use for prey location, and quite possibly for geomagnetic navigation, is remarkable. However, when these sh are stationary, their electrore-ceptive sense is useful only over a short range, probably less than a meter. Range is limited by the fact that an electric eld in v olts per meter decreases proportional to R-3, where R is the distance from the shark to the source. The electrical potential in volts of a dipole source falls off as R-2. The great sensitivity of the AoL organs may make them subject to interference from elds generated within the shark' s own body, and by its own swimming motions. How the shark compensates for this "noise" in its CNS is just beginning to be understood.
How the spike signals from elasmobranch ampullary arrays interact centrally to permit these sh to sense a weak electric eld and home in on its source promises to be a major challenge for computational neurobiologists. The inputs to the detection system are the many ampullary nerve bers that re re gularly at slightly different fos. The presence of an external electric eld will cause the ampullary nerv e ring frequencies to increase slightly or decrease, depending on its sign (which determines the direction of current o w in the ampullary canals). As a sh swims in the presence of a dc electric eld, its body motions will continuously modulate the ampullary nerve ring frequencies up and do wn, depending on the geometry. How does the sh use this information? That is, what neural circuit architectures might be involved in this system?
Ampullary electroreceptors are also used in passive prey detection by cat sh (Whitehead et al., 1998), and by the paddle sh, Poly don spathula, an eater of zooplankton. Zooplankton such as Daphnia sp. were shown to emit weak, low-frequency (dc-20 Hz) electric elds from dipole-lik e sources of their internal organs. Tested with arti cial, sinusoidal elds from electrodes, paddle sh preferred signals around 10 Hz (Wojtenek et al., 1998; Wilkens et al., 1998).
Was this article helpful?