Magnetosensationin zebrafish

Denis Shcherbakov1, MichaelWinklhofer2, Nikolai Petersen2,Johannes Steidle1, ReinhardHilbig1 and Martin Blum1*

Many species, from bacteria tovertebrates, have been reportedto use the geomagnetic field as amajor cue for oriented short andlong range migration [1–10], butthe molecular nature of theunderlying receptor has remainedelusive. One of the main reasonsmay be that past attempts to trainanimals to respond to magneticstimuli proved surprisingly difficult[11]. We present a novel approachto magnetic conditioning, using afast, fully automated assaysystem relying on negativereinforcement. Weak electricimpulses were applied to punishfish that failed to escape uponmagnetic field alterations(avoidance behaviour). Using thisassay we first demonstratemagnetosensation in Mozambiquetilapia, a fish migrating regularlybetween freshwater and the sea.Next we wondered whether non-migratory fish have a magneticsense, such as zebrafish, thegenetic fish model organism.Zebrafish were trained in groupsof 4 individuals, and statisticallyhighly significant reactions tomagnetic field changes wererecorded. The demonstration ofmagnetosensation in zebrafishopens a possibility to geneticallyidentify the magnetoreceptor andits downstream signallingcascade.

Magnetoreception haspreviously been shown inmigratory fish, such as tuna,salmon, and rainbow trout [4,11].Before testing zebrafish, a non-migratory fish, we established afully automated module usingtrout as test species. Followingpublished procedures [4], abehavioural assay using positivereinforcement was developed.These experiments (Supplementaldata), confirmed the presence ofmagnetosensation in trout [4].Training single fish, however, took30–40 days, rendering this assay

inappropriate for a systematicgenetic screening approach.

We therefore developed asecond assay relying on negativereinforcement. The experimentalset-up was composed of a fishtank with two conjoinedcompartments (Figure 1). Amagnetic field (100µT) wasapplied in the east–west directionand subsequently fish that failedto escape to the respective othercompartment were punished. Thepunishment was given as weakelectric impulses of 3V.Successful escape was registeredby an infrared light barrier systemand disabled punishment. Trialswere performed in pairs such thata magnetic field trial was alwaysfollowed by a control trial or viceversa, in a randomly generatedorder (Figure 1). A total of 10 trialpairs per training session wereperformed, and fish were trainedfor a total of up to 10 sessions perday. Typically, training sessionswere restricted to 3–4 on a singleday. Data were sampledcontinuously and displayed as theratio of infrared light barriercrossing (sensor signals) inmagnetic field versus controltrials, i.e., movement activity ineach individual control trial wasset to a value of one.

Mozambique tilapia, a robustmigratory species, was chosen asa first experimental fish foravoidance training. Figure 2A

summarizes the results on threefish trained individually in 10sessions each. Learning effectswere obvious from the fourthtraining session onward, with atotal of 28% higher movementactivity in magnetic field trialscompared to control trials. Theresults were highly significant(p<0.01), and differences inresponse rates between magneticfield and control trials were withinthe same range as those obtainedwith trout in positive reinforcementconditioning (Supplemental data).

In a second set of negativeconditioning experiments, wetested zebrafish formagnetosensation. As zebrafishare schooling fish, groups of fourindividuals were trained together.In order to establish experimentalconditions for this much smallerspecies, we initially tested light asa strong conditioning factor.Avoidance behaviour was verypronounced and instantaneouslyobvious (Supplemental data). Theaverage movement activity in atotal of three experiments (12 fish)was 40% higher in light trials thanin control trials (p<0.001). Next, fishwere challenged with magneticfield trials (Supplemental data). Asingle experiment with a strongresponse rate (20% higher inmagnetic field trials) is shown inFigure 2B. While response ratesvaried between individualexperiments (Figure 2C), statistical

Magazine R161

Figure 1. Testing magnetosensation in tilapia and zebrafish.

Fish were placed in an experimental module, consisting of two connected compart-ments. Magnetic field alterations were introduced by magnetic coils (1). Lack of escapeupon magnetic field change triggered penalties – weak electrical impulses generatedby non-magnetic metal plates (2) — at the end of the trial period. Escape into the othercompartment was monitored by infrared light barriers (3) and disabled the punishmenttrigger. A computer-controlled random order generator produced trial pairs in randomorder, in which a magnetic field trial was followed by a control trial or vice versa.

1 1

3

2

Input Output

Random ordergenerator

OR

Computer

Controltrials

no penalty

MF trialsnegative

reinforcement

MF trialsnegative

reinforcement

Controltrials

no penalty

Current Biology

significance was high for the totalof all 10 experiments (p<0.01).

Two sets of controls verifiedthe presence of a magnetic sensein zebrafish. When infrared lightsensor activations were countedduring simulated, randomlygenerated trial pairs in whichneither signal nor punishmentwas applied (softwaresimulation), no differences wererecorded (p=0.96). Another seriesof control experiments appliedpunishment in the absence ofmagnetic field changes. Again,

no significant tendency wasfound (p=0.88).

Why should non-migratory fishhave a magnetic sense? Short-range animals depend on deadreckoning, keeping track ofoutward legs while foraging andtake the net displacement to plota route home. Many species usethe sun or polarized light for thistask. In turbid water, underovercast skies and at night,however, the dead reckoning toolof choice may be the Earth’smagnetic field. The

demonstration ofmagnetosensation in zebrafishfor the first time offers theopportunity to identify themagnetoreceptor and itsdownstream signalling cascade.

Supplemental dataSupplemental data containingexperimental procedures areavailable at http://www.current-biology.com/cgi/content/full/15/5/R161/DC1/

References1. Wiltschko, W., and Wiltschko, R.

(1972). Magnetic compass ofEuropean robins. Science 176,62–64.

2. Blakemore, R. (1975).Magnetotactic bacteria. Science19, 377–379.

3. Lohmann, K.J., and Lohmann,C.M.F. (1996). Detection ofmagnetic field intensity by seaturtles. Nature 380, 59–61.

4. Walker, M.M., Diebel, C.E., Haugh,C.V., Pankhurst, P.M.,Montgomery, J.C., and Green, C.R.(1997). Structure and function ofthe vertebrate magnetic sense.Nature 390, 371–376.

5. Deutschlander, M.E., Borland, S.C.,and Phillips, J.B. (1999).Extraocular magnetic compass innewts. Nature 400, 324–325.

6. Kirschvink, J.L., Walker, M.M., andDiebel, C.E. (2001). Magnetite-based magnetoreception. Curr.Opin. Neurobiol. 11, 462–467.

7. Boles, L.C., and Lohmann, K.J.(2003). True navigation andmagnetic maps in spiny lobsters.Nature 421, 60–63.

8. Fleissner, G., Holtkamp-Rötzler, E.,Hanzlik, M., Winklhofer, M.,Fleissner, G., Petersen, N., andWiltschko, W. (2003).Ultrastructural analysis of aputative magnetoreceptor in thebeak of homing pigeons. J. Comp.Neurol. 458, 350–360.

9. Ritz, T., Thalau, P., Phillips, J.B.,Wiltschko, R., and Wiltschko, W.(2004). Resonance effects indicatea radical-pair mechanism for avianmagnetic compass. Nature 429,177–180.

10. Mouritsen, H., Feenders, G.,Liedvogel, M., and Kropp, W.(2004). Migratory birds use headscans to detect the direction of theearth’s magnetic field. Curr. Biol.14, 1946–1949.

11. Wiltschko, R., and Wiltschko, W.(1995). Magnetic orientation inanimals (Berlin: Springer).

1Institut für Zoologie, UniversitätHohenheim, D-70593 Stuttgart,Germany. 2Department für Geo- undUmweltwissenschaften, SektionGeophysik, Ludwig-Maximilians-Universität München, D-80333München, Germany.*E-mail: mblum@uni-hohenheim.de

Current Biology Vol 15 No 5R162

Figure 2. Magnetosensation in tilapia and zebrafish.

Results are expressed as relative response rates defined as the percentage of infraredsensor signals per magnetic field (MF) trial (or per control (co) trial) during a trainingsession with 10 trial pairs. (A) Magnetosensation in tilapia. Mean response rates of threeindividual fish trained in 10 sessions of 10 trial pairs each. Bars indicate standard error.(B) Magnetic field conditioning experiment in zebrafish. Response rates of four fishtrained as a group in 10 sessions of 10 trial pairs each. (C) Results from 10 group train-ing experiments in zebrafish. Note that while response rates vary between experiments,they were higher in magnetic field trials than in control trials in all cases.

1 2 3 4 5 6 7 8 9 10

Experiment

1.1

1.2

1.0

1 2 3 4 5 6 7 8 9 10

MF trialsControl trials

MF trialsControl trials

2.2

2.0

1.8

1.6

1.4

1.2

1.0

0.8

1 2 3 4 5 6 7 8 9 10Session

Session

Sen

sor s

igna

ls [M

F/c

o]S

enso

r sig

nals

[MF

/co]

Sen

sor s

igna

ls [M

F/c

o]2.2

2.0

1.8

1.6

1.4

1.2

1.0

0.8

A

B

C

Current Biology