Presented at the Neurofly 2008 conference in Würzburg, Germany, September 9, 2008.
Double Dissociation of Protein Kinase C and Adenylyl
Cyclase Manipulations on Operant and Classical Learning in Drosophila
Björn Brembs
FU Berlin, Institut für Biologie - Neurobiologie,
Königin-Luise-Strasse 28/30, 14195 Berlin, Germany
bjoern@brembs.net, http://brembs.net
Composite learning consists of two components with reciprocal, hierarchical interactions.
The AC-dependent classical or allocentric learning system inhibits the PKC-dependent operant or egocentric learning system via the mushroom-bodies. Operant behavior controlling predictive stimuli facilitates learning about these stimuli by the classical system via unknown, non-mushroom-body pathways. These interactions lead to efficient learning, generalization and prevent premature habit-formation.
Composite learning consists of two components with reciprocal, hierarchical interactions.
The AC-dependent classical or allocentric learning system inhibits the PKC-dependent operant or egocentric learning system via the mushroom-bodies. Operant behavior controlling predictive stimuli facilitates learning about these stimuli by the classical system via unknown, non-mushroom-body pathways. These interactions lead to efficient learning, generalization and prevent premature habit-formation.
classical/
allocentric
operant/
egocentric
Facili- tation
Inhi- bition
mb-dependent
rut-dependent
mb-independent
PKC-dependent
Composite learning
6. Conclusion
torque meter
yaw torque signal
diffusor light guides light source
IR laser diode
solenoid
color filter
classical component
operant component operant + classical
Composite training:
operant and classical components
Training Test
(1)
(2)
(3)
3. Learning-by-doing is most effective (in flies, too)
3. Learning-by-doing is most effective (in flies, too)
Fig. 3: Comparison of operant and classical pattern learning in flies.
The same sequence of sensory input sufficient for inducing a substantial learning effect if controlled operantly (left), only induces a small learning score if it is perceived passively (”classical”, right). Thus, active learning (”by doing”) is more effective than passive learning.
Left/red - Operant flies. N=30.
Right/blue - Classical flies. N=30.
Fig. 3: Comparison of operant and classical pattern learning in flies.
The same sequence of sensory input sufficient for inducing a substantial learning effect if controlled operantly (left), only induces a small learning score if it is perceived passively (”classical”, right). Thus, active learning (”by doing”) is more effective than passive learning.
Left/red - Operant flies. N=30.
Right/blue - Classical flies. N=30.
-0.2 0.0 0.2 0.4 0.6
PI [r el. units]
30 30
operant classical
AC PKC
classical
operant both
5. Mushroom-bodies prevent premature habit formation
5. Mushroom-bodies prevent premature habit formation
PI [r el. units]
*
*
20 19
mb247 x TNT
-0.2
0.0 0.2 0.4 0.6
21
A
genetic controls
* B
*
27 32 15
17 17 20
* *
-0.2 0.0 0.2 0.4
0.6 C
17D x TNT 17 14
14
* *
(2) Isolated operant component
(1) composite control (operant + classical)
(3) Isolated classical component
WT - extended training
17 21
12
c205 x TNT
*
*
-0.2 0.0 0.2 0.4 0.6
D
E
Fig. 5: The mushroom-body α
and β lobes but not the γ lobes
are necessary for inhibition of the operant component and ge- neralization of classical memory.
A. Flies with blocked MB output perform well in composite learning (red), but do not inhibit the operant component during com- posite training (green). Without inhibition of operant system, these transgenic flies are unable to generalize the isolated classi- cal component to a novel behavior (blue).
B. The genetic control flies (the two hete- rozygote strains did not differ and were pooled) reproduce the wild-type results:
significant composite learning, inhibition of the operant system and successful genera- lization of the isolated classical compo- nent. C. Flies with blocked output only from the α and β lobes of the MB mimic the flies expressing tetanus toxin in all MB lobes. They perform well in composite lear- ning, do not inhibit the operant system and do not generalize. D. Specificity of our mushroom-body effects is provided by expressing TNT in the fan-shaped body.
These flies behave as wildtype and control heterozygote flies with significant compo- site learning and inhibition of the operant system, which in turn allows for a success- ful generalization of the classical compo- nent to a novel behavior. E. Extended trai- ning in wildtype flies constitutes a pheno- copy of the transgenic animals (A). The longer training duration does not lead to an overtraining decrement. Testing for the operant component shows a release from the inhibition of operant learning. Without inhibition of the operant system, the flies are unable to generalize.
bjoern@brembs.net, http://brembs.net
WT
26
*
0.0 0.2 0.4
0.6 * *
noHS PKCi HS
PKCi rut
16 20 23
n.s.
AC PKC
PI [r el. units]
0.0 0.2 0.4 0.6
0 15 30 45 60 75 90 105 120
Time [s]
Yaw torque [arb. units]
-80 -40 0 40 80
0
Pattern position [°]
-180 -90 90
B
180C
electric motor torque meter
yaw torque signal
diffusor light guides light source
IR laser diode
solenoid
color filter
1. Classical or
Pavlovian learning 1. Classical or
Pavlovian learning
Fig. 1: Manipulation of AC, but not of PKC disrupts learning of a classical predictor.
A – Experimental setup. The fly controls the angular position of a drum with four identical vertical bars in a flight simulator- like situation. The coloration of the arena is switched between bars, such that flying towards one pair of opposing bars leads to green coloration and towards the other pair to blue coloration. During training, heat is made contingent on one color, irre- spective of the turning maneuver which changed flight direction. B – Sample data from a wildtype fly during the first test period after the final training with heat on blue coloration. The fly uses both left and right turning maneuvers (red trace) to change flight direction (blue trace) and hence coloration of the environment (background color of the graph). The fly shows a clear preference for green with only brief excursions into flight directions which lead to blue color, even though the heat is switched off. C – Pooled performance indices (PI) from the first test period after training. In this and all subsequent bar graphs: Displayed are means, error bars are s.e.m. Numbers at bars – number of animals; rut – rut-mutant flies affecting AC; WT – wildtype; HS PKCi – heat shock induced expression of the specific PKC inhibitor; noHS PKCi – PKCi expression not induced; n.s. – not significant, * – p<0.05.
A
torque meter
yaw torque signal
diffusor light guides light source
IR laser diode
30
*
WT noHS
PKCi HS
PKCi
0.0 0.2 0.4 0.6
17
* *
rut
23
20
n.s.
AC PKC
PI [r el. units]
0.0 0.2 0.4 0.6
43 41
noHS het.c.
HS het.c.
*
*
45 60 75 90 105 120
Time [s]
C
Fig. 4: Manipulation of PKC but not of AC disrupts learning of a purely operant predictor.
A – Experimental setup. There are no visual cues for the fly. During training, heat is made contingent on either left- or right-turning yaw torque. B – Sample data from a wildtype fly during the first test period after the final training with heat on positive (right- turning) yaw torque. The fly only briefly generates right-turning yaw torque during the test phase (unsaturated red/blue bar under- neath dark red yaw torque trace), even though the heat is switched off. C – Pooled performance indices (PI) from the first test period after training. HS het.c. – Heat shock-treated heterozygous parental controls strain; noHS het.c. – Heterozygous parental control strain without heat shock.
4. Purely operant learning is different
40
Yaw torque [arb. units]
B
-40 0
0 15 30
A
bjoern@brembs.net, http://brembs.net
32
*
WT
0.0 0.2 0.4 0.6
17
* *
noHS PKCi HS
PKCi rut
27 21
n.s.
AC PKC
PI [r el. units]
0.0 0.2 0.4 0.6
15 30 45 60 75 90 105 120
Time [s]
B
torque meter
yaw torque signal
diffusor light guides light source
IR laser diode
solenoid
color filter
2. Operant or
instrumental learning 2. Operant or
instrumental learning
Fig. 2: Operant learning requires the same gene as learning a classical predictor.
A – Experimental design. Throughout the experiment, one yaw torque domain is coupled to one color and the other to the other color (e.g., right turning causes green illumination and left turning blue illumination of the environment). During trai- ning, heat is made contingent on one of the two yaw torque/color combinations. B – Sample data from a wildtype fly during the first test period after the final training with heat on positive (right-turning) yaw torque (red trace) and blue illumination (background coloration). The fly shows the yaw torque domain/color preference and only briefly ventures into the previously punished situation, even though the heat is switched off. C – Pooled performance indices (PI) from the first test period after training.
C
-80 -40 0 40 80
Yaw torque [arb. units] 0