Temporal coding and the electric organ discharge: mormyrid fish as a model systemLaurieanne Dent, Bruce R. Land, and Carl D. Hopkins

Department of Neurobiology and Behavior, Cornell University, Ithaca, NY, USA, 14853

��

PO418

ntroductionI1The temporal features of an EOD are used for

social recognition Weakly-electric African mormyrid fishes use the timing information in an electric organ discharge (EOD) waveform to identify sex and species of other mormyrids1.

2Timing of EOD waveforms, encoded by Knollenorgan electroreceptors, is re-encoded by small cells in mid-brain nucleus, ELa

Knollenorgan electroreceptors phase-lock to an EOD

Temporal processing in brainstem, ELa

These receptors in the skin are stimulated by the positive-going voltage transient of an EOD, therefore, sensory inputs from different areas of the body are required to encode the entire EOD waveform--“two-spike code”1

In midbrain toral nucleus exterolateralis pars anterior (ELa, pink), ‘small cells’ compare arrival times of spikes from the periphery. ELa small cells output timing information exclusively to ELp (nucleus exterolateralis pars posterior, yellow).

3

M

esultsR

9/12/06180 µm

9/12/06250 µm

Stimulus amplitude (dB re: 50% max integral)

Inte

gra

l of E

Lp e

voke

d p

ote

nti

al (

V*s

)

(12)

(52)(34)

(48)

(16)(44)

(20)

(24)

(8)

(37)

(16)

(0)

(0)

(14)(30)

(15)

(6)

(33)

(53)

(13)

(44)

-10 100 20 30 40 50

9/18/06

2.0

2.5

3.0

3.5

4.0

0

1.0

1.5

0.5

A B C D

E F G H

0.5 ms6/12/07

0.5 ms6/12/07

5/31/07

2.0

2.5

3.0

3.5

4.0

0

1.0

1.5

0.5

6/16/070.5 ms

6/16/07

-10 100 20 30 40 50 -10 100 20 30 40 50-10 100 20 30 40 50

8/21/060.1 ms

2.0

2.5

3.0

3.5

4.0

0

1.0

1.5

0.5

9/4/06100 µm

9/4/06200 µm

A

I

E

B

2.0

2.5

3.0

3.5

4.0

0

1.0

1.5

0.5

9/11/06loc2

Inte

gra

l of E

Lp e

voke

d p

ote

nti

al (

V*s

)

(34)

(12)

(72)

(56)(57) (38)

(18)

(29)

(4)

(11)(0)

(8)(60) (38)

(10)

(8)

(66)

(45)(43)

(12)(23)

(54)

(47)

(65)(71)

5/31/07 6/12/07C 6/16/07D

9/11/06loc1

2.0

2.5

3.0

3.5

4.0

0

1.0

1.5

0.5

F 9/12/06180 µmG 9/12/06

250 µmH

J 9/4/06300 µmK

-10 100 20 30 40 50 -10 100 20 30 40 50 -10 100 20 30 40 50

-10 100 20 30 40 50

“Two-spike code”

Predictions of the model

Small cell size and large presynaptic terminals precludes intracellular recording (antomy, upper right), so we re-corded local field potentials from ELp under different electric field geometries (uniform, differential), and stimulus amplitudes and durations.

1) Uniform geometry: ELp evoked potential will be eliminated2) Differential geometry: ELp evoked potential will reach a plateau

If excitation is always delayed relative to inhibition, then

iscussionD

cknowledgementsA We thank B. Scott Jackson for the custom-written data acquisition sofware and for many helpful discussions regarding data interpretation. Support for LD was contributed by NIH#: 2 T32 GM07 469 and for CDH and BRL by NIH#: DC006206. References available separately.

Short duration stimuli, 0.2 ms (unless otherwise noted)

LONG duration stimuli, 1.0 ms (unless otherwise noted)

UniformHead-tailTransverse

‘Differential’

Uniform geometry: evoked potentials not abolished (more than 60% excitation “getting through”)Differential geometry: evoked potentials tend to reach a plateau at long stimulus durations, but not alwaysSmall duration effect (ca. 18% difference) on geometry: inhibition is greater in uniform than in differential(Is small cell activity being inhibited at high stimulus amplitudes, or is ELp firing more synchronously? ELp units with “closed tuning?”)

idelay = 50+/-25 usec

� ���

� ���

� ���

� ��

� ��

�

� �

� � �

� � �

HT

U

dur = 0.5 ms

� ���

� ���

� ���

� ��

� ��

�

� �

� �

� �

� �

� � �

� � �

� � �

� � �

HT

U

dur = 1.0 ms

� ���

� ���

� ���

� ��

� ��

�

� �

� �

� �

� �

� � �

� � �

HT

� � �

U

dur = 0.2 ms U

% Decrease in excitationDshort = 32.0±23.8% Ushort = 39.1±23.2%

1Experimental: evoked potentials

2Computer simulations

Dlong = 20.0±12.2%Ulong = 37.3±14.6%

We modeled the Knollenorgan pathway to see the effects of in-hibitory and excitatory delay lines on small cell activity as a func-tion of stimulus durations. Inhibitory and excitatory delays were measured from anatomical data.

Excitation arrives ahead of inhibition for some ELa small

cells

i.

ii.

iii.

Future directions...

?

excitatory delays

inhibitory delays

MORE COMPUTER SIMULATIONS, varying relative delays, thresholds, additional excitatory inputs (NOTE: the inhibitory synapse at the small cell is a chemical synapse with an estimated delay of 300 ms)

100

Stimulus Duration (µsec)

1.0

0 100-100-200

dE-dI = [100:-353]

-200-300

dE > dI, excitation lags

58 %

dE < dI, excitation leads

83 %

0

Using excitatory conduction delays from anatomical studies, and a fixed inhibitory delay of 360 µs--overlapping distribu-tions--, the majority of ELa small cell tuning would fall within a 100 µs stimulus duration range.

Differential geometry

Uniform geometry

Stimulus amplitude

ELp

resp

on

se

3 Quantify the evoked potential using the integral method

Inhibition is hypothesized to arrive in advance of excitation at ELa small cell ‘anti-coincidence’ detectors.2,3 Stimulus-evoked spikes from differ-ent body regions arrive at small cells, through either a delayed excitatory pathway or non-delayed inhibitory path. Only stimulus durations ex-ceeding axonal delays excite small cells (i.e., excitation from pulse onset arrives at small cell before inhibition from pulse offset, small cell ‘2’). We predict uniform geometry stimulation will activate inhibitory inputs in advance of excitation, preventing ELa small cells from firing.

Re-encoding by relative timing of excitation and inhibition: delay-line ‘anticoincidence’ model of EOD waveform processing in ELa

ethods

Stimulate fish with either uniform or differential elec-tric field geometry

2 Record evoked potentials from ELp

1

1o Aff

Knollenorgan

NELL

nPLL

OT

TEL

OB

valvula removed

VA

1 mm

Alarge cells

small cellsdelay line

ELaELp

NELL

ELpELa

ELL

B

EOD

commanddischarge inhibition

0.1 ms

lat. lem.

ipsicontra

spherical cells

midline

(modified from Xu-Friedman & Hopkins, 1999)

D = differential (electrodes*)U = uniform (electrodes*)

Switchbox U

DSIU

+

+-

-

Stimulator

A/D Converter

Amplifier

Spike detector

EOD command

ELp recording microelectrode

Computer

*

*

*

*

*

We measured the evoked potential by integrat-ing the area under the curve, in the time interval delimited by the dashed lines. Since the large majority of the evoked potential was negative, we considered only the negative values.

Uniform

Head-tail

Transverse

Uniform

Head-tail

Uniform

Head-tail

Transverse

Ratio of integral at highest stimulus amplitude compared to the peak integral of that stimulus geometry

UD

Computer simulation(inhibitory delay = 50±25µs)

Figure by John Sullivan & Carl Hopkins

(Mugnaini & Maler, 1987)

-+ -+

ELa small cell

(Friedman & Hopkins, 1998)

mµ20

(modified from Xu-Friedman and Hopkins, 1999)

5/31/07

2nd peakinitial peak

integration limits

ELa activity

ELp: increasing stimulus amplitude uniformgeometry

*artifact

5 ms

100 mV