Using local anaesthetics to block neuronal activity and map specific

European Journal of Neuroscience, Vol. 26, pp. 3193–3206, 2007
doi:10.1111/j.1460-9568.2007.05904.x
Using local anaesthetics to block neuronal activity and map
specific learning tasks to the mushroom bodies of an insect
brain
Jean-Marc Devaud,1,2 Aline Blunk,3 Jasmin Podufall,3 Martin Giurfa1,2 and Bernd Grünewald3
1
CRCA, CNRS UMR 5169, Toulouse, France
GDR 2905 ‘Neurosciences de la Mémoire’, CNRS, France
3
Institut für Biologie – Neurobiologie, Freie Universität Berlin, Berlin, Germany
2
Keywords: anaesthetic mechanisms, insects, olfactory learning, patch-clamp
Abstract
The formation of a stable olfactory memory requires activity within several brain regions. The honeybee provides a valuable model to
map complex olfactory learning tasks onto certain brain areas. To this end, we used injections of the local anaesthetic procaine to
reversibly block spike activity in a specific brain region, the mushroom body (MB). We first investigated the physiological effects of
procaine on cultured MB neurons from adult honeybee brains. Using the whole-cell configuration of the patch-clamp technique, we
show that procaine blocks voltage-gated Na+ and K+ currents in a dose-dependent manner between 0.1 and 10 mm. The effects are
reversible within a few minutes of wash. Lidocaine acts similarly, but is less effective at the tested concentrations. We then studied
the role of the MBs during reversal learning by blocking the neural activity within these structures by injecting procaine. During
reversal learning bees learn to revert their responses to two odorants, one rewarded (A+) and one unrewarded (B–), if their
contingencies are changed (A– vs B+). Injecting procaine into the MBs impaired reversal learning. Procaine treatment during
acquisition left the later retention of the initial learning (A+ vs B–) intact. Similarly, a differential conditioning task involving novel
odorants (C+ vs D–) was intact under procaine treatment. Our experiments show that local injections of procaine can be used to map
learning tasks onto specific regions of the insect brain. We conclude that intact MB activity is required for the acquisition of reversal
learning, but not for simple differential learning tasks.
Introduction
Local anaesthetics can be used to locally and reversibly block spike
activity within the brain to study how specific brain regions control
behaviour. The primary targets of, for example lidocaine and procaine,
commonly used anaesthetics, are voltage-sensitive Na+ channels (e.g.
Hille, 1966; Bräu et al., 1998; Sheets & Hanck, 2003; Leuwer et al.,
2004; for review, see Scholz, 2002). However, they also block other
voltage-sensitive channels (e.g. Sugiyama & Muteki, 1994; Komai &
McDowell, 2001; Bischoff et al., 2003) as well as ligand-gated ion
channels (e.g. Cuevas & Adams, 1994; Nishizawa et al., 2002).
Studies of the physiological effects of local anaesthetics on invertebrate spikes or ionic currents are comparably rare, although the first
voltage-clamp experiments on the actions of procaine were performed
in an invertebrate preparation (Fatt & Katz, 1953; squid: Taylor, 1959;
blowflies: Wolbarsht & Hanson, 1965; Washio, 1972; cockroaches:
Lapied et al., 2001; crustaceans: Anwyl, 1977; Uusitalo et al., 1995;
Theander et al., 1996; C. elegans: Franks et al., 2002). These
experiments indicate a similar mode of action in invertebrates and
vertebrates.
Most studies of the behavioural effects of anaesthetic treatment of
specific brain regions have been performed on mammals (e.g. Bohbot
Correspondence: Dr J.-M. Devaud, 1CRCA, CNRS UMR 5169, ‘Neurosciences de la
Mémoire’, as above.
E-mail: [email protected]
Received 2 May 2007, revised 14 September 2007, accepted 20 September 2007
et al., 1996; Daumas et al., 2005; Martin et al., 2006; Teixeira et al.,
2006). Previous work by Müller et al. (2003), using systemic
injections, showed reversible impairments of motor responses and
memory performances in honeybees. Before using local anaesthetics
in vivo, however, we provide a detailed analysis of their physiological
effects on cultured honeybee neurons where the voltage-sensitive ionic
currents are well investigated (e.g. Schäfer et al., 1994; Pelz et al.,
1999; Grünewald, 2003; Wüstenberg et al., 2004). For behavioural
experiments we used the olfactory conditioning of the proboscis
extension reflex (Kuwabara, 1957; Bitterman et al., 1983), which is a
well-established paradigm to study the neural control of learning and
memory formation in bees. Here, we applied intraneuropilar injections
of local anaesthetics to localize acquisition and memory formation
using this paradigm. We specifically addressed the role of one brain
area, the mushroom bodies (MBs), which has been repeatedly
associated with learning and memory capabilities (Menzel et al.,
1974; Erber et al., 1980; Menzel, 1999, 2001), and more recently with
specific forms of higher-order learning and memory (Giurfa, 2003).
During olfactory conditioning, primary associations between an odour
(the conditioned stimulus, CS) and the reward (sucrose solution: the
unconditioned stimulus, US) may be formed within the antennal lobes
(ALs) and the MBs (Hammer, 1993; Hammer & Menzel, 1998).
Conditioned responses (CRs) during acquisition are apparently
independent of proper MB function (Malun et al., 2002b; Komischke
et al., 2005). By contrast, formation, consolidation and retrieval of
memory, particularly long-term memory, require functional MBs
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
3194 J.-M. Devaud et al.
(Menzel et al., 1974; Erber et al., 1980; Cano-Lozano et al., 1996,
2001; Komischke et al., 2005; Locatelli et al., 2005). Because the
MBs appear to play only a minor role during simple learning forms
like differential conditioning, several researchers argued that the
honeybee MBs might rather be important during complex learning
behaviour (Menzel & Giurfa, 2001; Giurfa, 2003, 2007; Lachnit et al.,
2004). Reversal learning may be regarded as an ambiguous task, as
transition between two discrimination phases implies a change in the
valence of odorants with opposed contingencies. Our hypothesis was
that MB activity was necessary during reversal learning. By injecting
procaine into the main output regions of the MBs, the a-lobes, we
showed that the acquisition of reversal learning was impaired by MB
blockade, leaving differential learning unaffected. This indicates that
output from the MBs is required for the acquisition of reversal learning
in honeybees.
Materials and methods
Electrophysiology
Animals and cell preparation
Adult honeybees (Apis mellifera) were collected from the comb or
entrance of hives from the Neurobiology Institute. All electrophysiological experiments were performed in Berlin.
The procedures for dissecting and culturing of Kenyon cells from
adult honeybees were similar to those described earlier for pupal cells
(Kreissl & Bicker, 1992; Grünewald, 2003). Brains were removed
from the head capsule and transferred into a Leibovitz L15 medium
(Gibco BRL) supplemented with (in mm): sucrose, 123; glucose, 22.2;
fructose, 13.9; proline, 28.7 (500 mOsmol, pH 7.2). The glial sheath
was gently removed and the MBs were dissected out of the brains.
After incubation (10 min) in a calcium-free saline solution (in mm:
NaCl, 130; KCl, 5; MgCl2, 10; glucose, 25; sucrose, 180; HEPES, 10;
pH 7.2), MBs were transferred to L15 medium containing 1 mg ⁄ mL
collagenase ⁄ dispase (30 min). Thereafter, the tissue was rinsed twice
with L15 and dissociated by gentle trituration with a 100-lL
Eppendorf pipette. Cells were plated in 10-lL samples on Falcon
plastic dishes coated with poly-lysine (poly-lysine-L-hydrobromide,
MW > 300 kDa Biochrom, Berlin, Germany) and allowed to settle
and adhere to the substrate for at least 10 min. Thereafter, the dishes
were filled with approximately 2.5 mL of culture medium [13% (v ⁄ v)
heat-inactivated foetal calf serum (Sigma, St Louis, MO, USA), 1.3%
(v ⁄ v) yeast hydrolysate (Sigma), 12.5% (w ⁄ v) L-15 powder medium
(Gibco BRL), 18.9 mm glucose, 11.6 mm fructose, 24.7 mm proline,
93.5 mm sucrose 8 lg ⁄ mL gentamycin (from a stock solution of
10 mg ⁄ mL, Sigma); adjusted to pH 6.7 with NaOH; 500 mOsmol]
and were kept at 26 C in an incubator at high humidity. For
electrophysiological measurements, cells were used between culture
days 3 and 7. Processes of those cells chosen for recordings did not
overlap with neighbouring neurites.
Electrophysiological techniques
Whole-cell gigaohm seal recordings were performed at room temperature following the methods described by Hamill et al. (1981).
Electrodes were pulled from borosilicate glass capillaries (1.5 mm
o.d., 0.8 mm i.d., GB150-8P, Science Products, Hofheim, Germany)
with a horizontal puller (DMZ-Universal Puller, Zeitz-Instrumente,
Munich, Germany), and had tip resistances between 5 and 10 MW in
standard external solution (see below). Recordings were made using a
HEKA EPC9 amplifier (HEKA-Elektronik, Dr Schulz GmbH,
Lamprecht, Germany). Pulse generation, data acquisition and analysis
were carried out using PULSE and PULSE-FIT softwares (version
8.53, HEKA-Elektronik) under Windows XP. Currents were low-pass
filtered with a four-pole Bessel ()3 dB) filter and sampled at 10–
20 kHz. Pipette offset potentials were nulled prior to seal formation;
leakage currents were not subtracted. Series resistances ranged
between 5 and 20 MW, and were compensated at approximately
85%. For data analyses we used IGOR Pro, version 3.15 (Wavemetrics, Lake Oswego, OR, USA).
Solutions
The bath (chamber volume: 1 mL) was continuously perfused at
flow rates of 1–4 mL ⁄ min with a standard external solution that
consisted of (in mm): NaCl, 130; KCl, 6; MgCl2, 4; CaCl2, 5; sucrose,
160; glucose, 25; HEPES-NaOH, 10; pH 6.7, 520 mOsmol ⁄ L1. To
record currents through K+ channels, tetrodotoxin (100 nm) was added
to the saline to block voltage-gated Na+ currents. Some experiments
were performed with 50 lm CdCl2 in the solution to block Ca2+
currents. The standard internal solution contained (in mm): KCl, 20;
K-gluconate, 87; Na2-ATP, 3; CaCl2, 0.2; MgCl2, 3; K-EGTA, 10;
glutathione, 3; GTP-Mg, 0.1; HEPES-bis-tris, 10; sucrose, 120; KF,
40; pH 6.7, 500 mOsmol ⁄ L.
To record currents through Na+ channels, K+ ions in the micropipette solution were replaced by TEA or Cs2+ Cs-gluconate, TEA-Cl,
Cs-EGTA and CsF replaced the corresponding K+ salts (in mm: TEACl, 20; Cs-gluconate, 83; Na2-ATP, 3; CaCl2, 0.2; MgCl2, 3;
Cs-EGTA, 10; glutathione, 3; Mg-GTP, 0.1; HEPES-bis-tris, 10;
sucrose, 120; CsF, 40). All chemicals were purchased from Sigma.
Statistical analyses
Only procaine experiments were statistically evaluated, because we
performed very few experiments with lidocaine. To determine
concentration effects of procaine on voltage-sensitive ionic currents
and to test whether procaine differently blocked currents through
K+ and Na+ channels, various statistical procedures were employed. A
three-factor repeated-measure anova was used to test for concentration effects, because each cell was usually tested several times with
different procaine concentrations. Differences between procaine
effects on Na+ and K+ currents were tested with an anova. For
post hoc comparisons we employed the Scheffé post hoc test. All
statistical analyses were performed using STATISTICA (StatSoft,
Tulsa, OK, USA).
Behavioural experiments
Animals
All behavioural experiments were performed in Toulouse. Honeybees (Apis mellifera) were caught from hives of the Research
Institute on Animal Cognition in the morning of each experimental
day, and immobilized by brief cooling on ice. They were then
harnessed in individual metal tubes that allowed free movements of
their antennae and mouthparts. A small drop of melted bee wax was
used to fix the back of the head to the tube to avoid rotational
movements during conditioning. Bees were then prepared for
intracerebral injections (see below). A window was cut in the head
cuticle to give frontal access to the brain. The glands and parts of
the tracheae were carefully removed, but the neurilemma was left
intact. After placing the piece of cuticle back to its original position,
bees were fed with two drops of sucrose solution (50% in water)
and left to rest for 3 h in a dark and humid chamber before the
experiment started.
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 26, 3193–3206
Local anaesthetics and odour learning in honeybees 3195
Conditioning protocols
Injections
A standardized protocol was used for acquisition and retrieval trials,
as described recently by Stollhoff et al. (2005). Each acquisition trial
lasted 40 s. It consisted of putting a bee from its resting position to
the test position into an incoming odourless air flow. An odour
stimulus (the CS) was applied 15 s later (duration 4 s) by passing the
air flow in a syringe containing a filter paper soaked with 4 lL of
odorant. Changes from pure air to odorant and vice versa were
operated automatically using computer-controlled magnetic valves,
keeping the flow intensity constant. Presentation of the US started
3 s after odour onset. For this both antennae were touched by a
toothpick soaked in 50% sucrose solution to induce proboscis
extension and thus ingestion of sugar as the US. Thereafter bees
were allowed to lick sucrose solution with their proboscis for 3 s
(i.e. about 3 lL), with a 1 s overlap with the CS. Unrewarded CS
presentations followed the same timing, except that no sucrose
stimulation was applied.
Procaine was dissolved in a saline that consisted of (in mm): sucrose,
160; glucose, 25; HEPES, 10; MgCl2, 4; NaCl, 130; KCl, 6; CaCl2, 5
(pH 6.7). Except for preliminary experiments (see above), the
concentration of the procaine solution was 20% (740 mm), and all
injections were performed 15 min before the start of phase 2. For
behavioural experiments, we rather indicate dilutions (in percentage)
instead of concentrations as in patch-clamp experiments (in mm) in
order to enable a comparison with the work of Müller et al. (2003).
For clarity, the corresponding concentrations in mm have been
indicated in the text and figure legends. Each injection procedure took
less than 10 min. A volume of 0.5 nL of either procaine solution or
saline alone was injected in each a-lobe using a pulled glass capillary
(GC 100-10, Harvard Apparatus, Les Ulis, France) connected to a
pressure microinjector (IM 300, Narishige, London, UK). The volume
of the injected solution was adjusted by injecting saline into a drop of
paraffin oil prior to the injection phase and immediately thereafter.
Methylene blue (2 mm, Sigma-Aldrich, Lyon, France) was added to
the solutions to control for successful injections and correct location of
the injection.
Pilot experiments. At the beginning of this study, two pilot experiments were designed to establish the optimal conditions for a
behavioural effect of procaine. In both cases, bees were trained to
learn a simple association between a CS (limonene) and the sugar
reward (A+). The acquisition phase consisted of six trials separated by
a 10-min interval. In a first experiment, bees received injections
45 min before the start of conditioning, in both ALs. Independent
groups were injected with saline and procaine at one of the following
concentrations (w ⁄ w in saline, see below): 4%, 10% or 20% (about
148, 370 and 740 mm, respectively). In a second experiment,
injections of 20% procaine (in both ALs) were performed at different
times before conditioning, i.e. either 15, 30, 45, 60 or 90 min. Saline
injections were also performed at all time points in independent
groups.
Reversal learning. In all experiments on reversal learning, bees were
injected in the a-lobes of the MBs. These structures are at the base of
the pedunculus and constitute the output region of MBs. Bees were
first subject to a differential conditioning with two odorants A and B,
one being rewarded with a sucrose solution and the other presented
unrewarded (phase 1: A+ vs B–). After 45 min of rest, bees that
performed correctly the discrimination in at least one block received
an injection of either procaine or saline (see below). Bees were then
trained during a second acquisition phase (phase 2) starting 15 min
after injections (i.e. 1 h after the end of the initial training). Three
different protocols were used during phase 2. (i) In the reversal
learning experiment, the same odorants as in phase 1 were presented,
but with reversed contingencies: odorant B was thus rewarded and
odorant A was presented without reward (A– vs B+). (ii) In the
differential conditioning experiment, phase 2 was similar to phase 1,
but with two novel odorants (C+ vs D–). (iii) In the extinction
experiment, odours A and B were again presented in phase 2, but
without reward (A– vs B–). For all experiments, both phase 1 and 2
consisted of five trials for each odorant (10 trials in total) with 8-min
intertrial intervals. The odorants were presented in a pseudo-randomized
sequence. During the retention tests (1 h after phase 2) the odorants
used for conditioning were presented without reward (CS-only trials).
Finally, the unconditioned response (UR) was assayed for each bee.
Only bees that responded to the sucrose stimulus were considered for
analysis. The odorants used were limonene, eugenol (A and B),
1-heptanal, 1-nonanol (C and D), and were purchased from SigmaAldrich (Lyon, France). All experiments were balanced with respect to
odour identity.
Statistical analysis
All results are presented as percentages of proboscis extension
responses to the odorant applied, ± SEM. Error bars on the graphs
represent the 95% confidence limits, as calculated for proportions (Zar,
1999). As our experiments met the criteria required for the application
of anova to a dichotomous dependent variable (Lunney, 1970),
repeated-measurement anova was used for between-group and
within-group comparisons for each acquisition phase. A Wilcoxon
test was performed to determine differences between response rates to
the CS during specific trials of either phase of acquisition and during
the retention test. In the pilot experiments, comparisons of response
rates at the last trial were performed using a Mann–Whitney test. All
statistical analyses were performed using SPSS 14.0 (SPSS, La
Défense, France).
Results
Procaine and lidocaine inhibit voltage-sensitive ionic currents
and spike activity
Procaine and lidocaine reduce voltage-sensitive ionic currents
When taken into primary neuron culture, Kenyon cells from adult
honeybee brains survive up to 14 days in vitro. We successfully
recorded from the neurons between 2 and 12 days in vitro. The adult
Kenyon cell shape is similar to that described for pupal cells
(Grünewald, 2003). However, only very few cells show neurite
outgrowth in the dish. After rupturing the cell membrane, Kenyon
cells were voltage-clamped to a holding potential of )80 mV.
Depolarizing voltage pulses were then applied to elicit voltagesensitive ionic currents. Subsequently, we determined the effects of
procaine and lidocaine onto the isolated current components.
Na+ currents. Sodium currents (INa) were isolated by blocking
voltage-gated Ca2+ and K+ currents. Voltage-sensitive Ca2+ currents
were blocked by adding 50 lm CdCl2 to the bath solution, and
K+ currents were blocked by substituting Cs2+ (133 mm) for K+
and adding 20 mm TEA to the pipette solution (Fig. 1). INa activated
at voltages more depolarized than )40 mV and peaked between )10
and +4 mV (Fig. 1C). The peak current ranged between )51.4 and
)418.4 pA, with a mean amplitude of )195.7 ± 51.4 pA (SEM,
N ¼ 27).
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 26, 3193–3206
3196 J.-M. Devaud et al.
Fig. 1. Procaine and lidocaine reduce voltage-sensitive Na+ currents. (A) Kenyon cells express rapidly activating and inactivating Na+ currents. Procaine and
lidocaine reduced the amplitude of the Na+ currents. At 5 mm procaine was a more potent blocker. The effects were reversible even at 10 mm procaine and after
almost complete block (B). Voltage protocols are given in the inset. (C) The current–voltage relationship (current amplitude measured at the peak) indicates that
procaine does not shift the steady-state activation of the Na+ current. (D) Superimposed current traces of INa at +4 mV during bath perfusion with saline before (pre),
during 10 mm procaine (proc) and during wash (wash, 2 min). (E) Procaine inhibited Na+ currents in a dose-dependent way (P < 0.05, repeated-measure anova).
Significant differences between different concentrations (Scheffé post hoc test) are indicated by different letters. Currents are normalized to Imax measured at a
command potential of )10 mV during the last activation protocol before procaine application. Plotted are the mean relative currents (± SEM) before and during
anaesthetic (drug) applications and during wash. (F) Lidocaine blocked Na+ currents, but appears less effective than procaine (no block at 1 mm). However, note
that the number of experiments during these experiments was smaller, statistical differences were therefore not tested.
Procaine inhibited currents through sodium channels in a concentration-dependent manner between 0.1 and 10 mm (P < 0.05;
F ¼ 8.55, repeated-measure anova). The effects were completely
reversible after 1–3 min of wash (Fig. 1A and B). Procaine
applications reduced the mean peak amplitude of the Na+ current to
64.7 ± 2.8% (SEM, 1 mm, N ¼ 29; Fig. 1E) and 49.7 ± 5.4% (5 mm,
N ¼ 25). At a procaine concentration of 10 mm, the mean Na+ current
amplitude reached only 38.9 ± 4.6% of its initial level (N ¼ 14). The
reduction of the current at this concentration ranged between 94.2 and
37.7%. In three cells, the reduction was above 80%, and in two cells
less than 40%. In several cells the seal stability of the recording
decreased at this concentration. Therefore, we did not test higher
concentrations than 10 mm.
K+ currents. Honeybee neurons express various voltage-sensitive
K+ currents. These currents were described in detail on cells cultured
from pupal brains (Schäfer et al., 1994; Pelz et al., 1999; Grünewald,
2003; Wüstenberg et al., 2004). Adult Kenyon cells show very similar
K+ currents with respect to steady-state activation as well as the
overall dynamics (Fig. 2). For comparison of the procaine effects we
measured the current amplitudes at two different time points during
the depolarizing test pulse. Firstly, the peak current reflects the
transient, inactivating K+ current that consists mainly of an A-type
K+ current (Pelz et al., 1999). Secondly, the sustained current,
measured at the end of the 100-ms test pulse, comprises mainly the
delayed rectifier K+ current. The mean peak amplitude of the transient
current at a holding potential of +50 mV was 1630.5 ± 121.2 pA
(SEM, N ¼ 22; range 587.0–2982.2 pA). The current amplitude of the
sustained current component was 364.0 ± 34.1 pA (SEM, N ¼ 22;
range 144.6–693.4 pA). The ratio of peak (transient) vs sustained
K+ current was 4.87 ± 0.37. This is higher than the ratio of Kenyon
cells from pupal brain tissue obtained in an earlier study (2.8 ± 0.25;
N ¼ 10, Grünewald, 2003).
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 26, 3193–3206
Local anaesthetics and odour learning in honeybees 3197
Fig. 2. Effects of local anaesthetics on K+ currents. Depolarizing voltage pulses (inset) elicited large K+ currents (A). They comprised of rapidly activating and
inactivating components, and a sustained current component. Both components are sensitive to procaine and lidocaine, and the effects are readily reversible (all
current traces were measured on a given cell). (B) Representative K+ current trace (pulse potential +60 mV) before, during (procaine, above; lidocaine, below) and
after (wash) anaesthetic delivery. (C) The transient K+ current component can be isolated by prepulse experiments. In a first experiment the total K+ current was
elicited after a prepulse to )120 mV. From this, the current after a prepulse to )20 mV was subtracted. The time course of this difference current (Itrans) is shown at a
command potential of +60 mV. Top trace: current amplitude is reduced by 5 mm procaine (grey line, arrow); lower trace: normalizing the peak currents to the
maximum currents (at the peak, Vhold +60 mV) show perfect overlap of the currents before and during 5 mm procaine. This indicates that procaine did not alter the
current dynamics. (D) Current reduction of the transient (top) and sustained K+ current at various procaine (left) and lidocaine (right) concentrations. Relative
currents (I ⁄ Imax) are reduced in a concentration-dependent manner by procaine (P < 0.0001, repeated-measure anova, differences between concentrations are
indicated).
Procaine reversibly blocked voltage-sensitive K+ currents (Fig. 2A).
At a concentration of 1 mm it reduced the peak amplitude to
87.1 ± 2.3% and the sustained K+ current component to 77.1 ± 3.2%
(N ¼ 25; VHold ¼ 50 mV; Fig. 2D). The procaine block was dosedependent (P < 0.00001, F ¼ 46.68, d.f. ¼ 16, repeated-measure-
ment anova). The maximum block (at 10 mm) ranged between 65
and 33.7%, with a mean of 51.6 ± 2.2% of the original current
(measured at the peak, N ¼ 18; Fig. 2D). Two cells showed a block of
more than 60%, and two cells of less than 40%. Concentrations higher
than 10 mm were not tested due to seal instabilities (see above).
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 26, 3193–3206
3198 J.-M. Devaud et al.
Procaine affected the transient and the sustained K+ current equally
(Fig. 2B and D; P ¼ 0.25, F ¼ 1.31, d.f. ¼ 201, anova). It reduced
the current amplitude and did not alter the current kinetics (Fig. 2C).
The voltage-sensitive Na+ current was more sensitive to procaine
than the K+ currents at a concentration of 1 mm (P < 0.01, Scheffe
test, N ¼ 25). No differences were detected for the other procaine
concentrations.
Lidocaine. Earlier studies (Müller et al., 2003) used lidocaine for
behavioural experiments to prevent neural activity in honeybees. We
wanted to test whether lidocaine was similarly effective than procaine.
Therefore, and because we needed to evaluate which drug was best
suited for intraneuropil injections, we also tested the effect of lidocaine
on a few neurons. Overall we observed similar effects as for procaine.
However, lidocaine appeared to be a less potent Na+ channel blocker
than procaine. At a concentration of 1 mm no reduction of the peak
Na+ current amplitude was observed (99.4 ± 5.8%, SEM, N ¼ 4,
Fig. 1F). At higher concentrations the blockade was dose dependent
with 70.2 ± 15.6% at 5 mm (N ¼ 5) and 56.6 ± 12.4% at 10 mm
(N ¼ 4). Lidocaine also inhibited voltage-sensitive K+ currents
(Fig. 2D). The lidocaine effects were dose dependent and reduced
the peak current amplitude between 64.9 ± 5.6% at 1 mm (N ¼ 5)
and 24.1 ± 3.1% at 10 mm (N ¼ 4). All lidocaine-induced blockades
were completely reversible after a few minutes of wash. Thus, it
appears that lidocaine inhibits K+ currents more potently than
Na+ currents. We used procaine for injections during the behavioural
experiments, because it is more efficiently blocking Na+ currents and it
appears to be less toxic than lidocaine (Müller et al., 2003).
)65 mV. Under these conditions Kenyon cells do not generate
spontaneous action potentials, as was similarly shown for pupal cells
(Wüstenberg et al., 2004). About 26% of the recorded Kenyon cells
(12 of 45 cells) responded to a depolarizing current injection (100 ms
)1 s) by firing at least one action potential. The remaining cells failed
to generate action potentials. Three cells out of 45 cells spiked
repetitively when depolarized with a 1-s, constant-current pulse
(Fig. 3A), whereas the remaining cells fired only a single spike to a
suprathreshold depolarization. Action potentials were normally overshooting, but otherwise varied considerably in amplitude and duration
among different cells. Even within a given cell the spike shape
changed during the time course of the recording.
The spike activity was blocked by 2% procaine (corresponding to a
concentration of approximately 74 mm) applied to the bath perfusion
(N ¼ 5; Fig. 3A and B). Lower concentrations of procaine (0.2–
1 mm) did not affect spiking in cultured Kenyon cells. Upon switching
to the voltage-clamp mode of the amplifier, we determined the
membrane currents of the cell presented in Fig. 3. At a concentration
of 74 mm procaine reversibly inhibited the inward current almost
completely (Fig. 3B). In addition, procaine also drastically reduced the
amplitude of the total outward current. At 5–10 mm procaine the spike
duration was substantially prolonged (N ¼ 6; data not shown). The
effects of procaine on spikes were reversible within 60 s of wash.
After establishing the physiological efficiency of procaine and
lidocaine in vitro, we wanted to test the effect of local anaesthesia of
MB a-lobes on reversal learning. Before starting these experiments,
we first performed two pilot experiments designed to determine the
dose and time of injection in order to block neural activity rapidly
enough and during the complete duration of the reversal experiment.
Procaine blocks action potentials in adult Kenyon cells in vitro
Upon switching the amplifier into the current-clamp mode, a constant
current was injected to keep the cell membrane at about )80 mV.
When the holding current was removed the cells maintained a
membrane potential that varied considerably between )45 and
Local injections of procaine induce dose-dependent and stable
behavioural effects
We first examined the learning performance of bees that had received a
bilateral injection of saline or procaine (4%, 10% or 20% in
Fig. 3. Procaine inhibits spike activity. Cultured Kenyon cells generate overshooting action potentials upon injecting depolarizing current pulses (+14 pA in A, 13
pA in B). At about 74 mm (¼ 2%) procaine added to the external saline spikes are blocked (A). The dashed line indicates a membrane potential of 0 mV.
(B) The voltage-sensitive ionic currents of the same cell measured under voltage-clamp. Outward as well as inward currents are reduced. Both the effects on spikes
and ionic currents are reversed after 2 min of wash (lower traces in A and B).
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 26, 3193–3206
Local anaesthetics and odour learning in honeybees 3199
saline ) corresponding to about 148, 370 and 740 mm, respectively)
into the ALs. We targeted this structure because of its well-known
implication in odour processing, so that we expected olfactory
conditioning to be impaired during procaine-induced blockade of neural
activity in vivo. The conditioning protocol lasted 60 min, and injections
were performed 45 min before it started, so that we compared
performances in the last trial at a time that corresponds to 105 min
post-injection. This delay was chosen in order to be comparable with the
required duration of procaine action during the next experiments
(acquisition or extinction phases of 80 min). As shown in Fig. 4A,
procaine-injected bees exhibited a dose-dependent decrease of learning
performance as compared with the control group. Relative to control
levels (100%), the rates of CRs were reduced after injection of 4% (CR,
72.9%), 10% (CR, 53.7%) and 20% (CR, 47.0%). This dose-dependent
learning impairment was significant for the highest dose (20%),
although at 10% the effect was marginally significant (Mann–Whitney
test: for 10%: P ¼ 0.068; for 20%: P < 0.01). We thus selected 20% as a
working dilution for the next experiments.
Once the efficient dose was selected, we varied the delay between
injection and conditioning, from 15 min to 90 min (bilateral injection
into the ALs; Fig. 4B). After a six-trial conditioning identical to that used
in the previous experiment, all groups of procaine-treated bees showed a
significant decrease in learning, compared with bees treated with saline
at a similar time (P < 0.05 in all cases, Mann–Whitney). This decrease
varied from 37.1% to 42.9% of control bees’ rate of CRs (P < 0.05 in all
cases). However, we found no clear effect of the delay among the values
tested, thus we selected the shortest one (15 min before conditioning).
For this value, the effect on behaviour was maintained over more than
1 h (delay between injection and last conditioning trial: 45 min).
Because reversal learning would consist of two successive acquisition
phases (80 min each), separated by a 1-h period during which injections
had to be performed (see Materials and methods), we decided to inject
15 min before the second (reversal) phase began. This would ensure to
leave consolidation after phase 1 undisturbed for 45 min and to affect
neural activity during the whole duration of phase 2.
Injections of procaine into the a-lobes impair acquisition
of reversal learning
During this and all following experiments, bees were trained to
discriminate two odours A and B during phase 1. One was rewarded
(A+) and the other not (B–). The response probability to A+ was
higher than to B– during acquisition (Figs 5–8). At the end of the first
phase of acquisition (i.e. before injection of procaine), the proportions
of CRs to A+ and B– were about 80% and 10%, respectively
(P < 0.005 in both groups, Fig. 5).
One hour later all bees were trained the reversed differential rule,
i.e. A– vs B+. Injections of procaine into both a-lobes differentially
affected the acquisition curves during the second phase, as indicated
by a significant effect of treatment on responses (F1,162 ¼ 4.336,
P < 0.05), as well as a significant treatment–blocks interaction
(F1,162 ¼ 6.296, P < 0.005; repeated-measurement anova: treatment · block · CS). This difference could be attributed specifically
to the treatment, as no difference in acquisition was found between the
two groups before injection, during phase 1, like in all the other
experiments. It should be noted that, in our conditions, procaine
injections affected neither survival nor the probability of the UR
(P > 0.05 in both cases, Mann–Whitney).
As shown in Fig. 5, saline-injected bees were able to learn the
reversed contingencies. They responded significantly more to odour B
than to A during the last trial of this second training phase (P < 0.001,
Wilcoxon). By contrast, bees that received procaine injections were
not able to reverse the rule, because they did not respond differently to
both odorants. Their acquisition curve showed an increased response
probability to B+. However, responses to A– did not decrease and the
bees did not show significant differences between odour responses to
A vs B at the last trial of the reversal training phase. Similarly, they
responded equally to odours A and B during the retention tests (1 h
later), while saline-injected bees responded more often to odour B
(P < 0.005, Wilcoxon). Thus, a-lobe blockade by procaine impaired
reversal learning.
Reversal learning impairment after MB blockade is not
due to impaired retrieval
As observed in Fig. 5, procaine injections resulted in a reduction of the
initial level of response to A in the first trial of phase 2 for this
odorant, compared with the level achieved at the end of phase 1
( 40% and 80%, respectively). This decay in retention level to the
initially conditioned odour A after procaine administration is in
general agreement with a role of the a-lobes during the recall of an
Fig. 4. Dose dependence and time-course of procaine action in vivo (ALs). Percentage of proboscis extension responses (PER) evoked by limonene, on the last of
six paired presentations with sugar (A+), i.e. 60 min after the beginning of conditioning. Values are expressed as proportions of the performance of controls (salineinjected bees). (A) Comparison of performance levels of animals injected with either saline or procaine (4%, 10% or 20% in saline ) respectively, about 148, 370
and 740 mm) into both ALs, 45 min before conditioning (i.e. performance was measured 105 min after injection). (B) Comparison of final performance levels
(sixth conditioning trial) of bees injected with saline or 20% procaine into both ALs between 15 and 90 min before conditioning. *P < 0.05; **P < 0.001 (n ¼ 25 in
all groups).
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 26, 3193–3206
3200 J.-M. Devaud et al.
Fig. 5. Reversal learning after blocking alpha lobes. Percentage of proboscis extension response (PER) evoked by odorants A (filled bars) and B (open bars). In
phase 1 of acquisition, bees from both experimental groups were untreated and had to learn to discriminate between a rewarded odorant (A+) and an unrewarded one
(B–). After injections of either saline (above) or procaine (below), phase 2 took place, during which the reversed rule (A–, B+) had to be acquired. A retention test
was applied 1 h after the end of acquisition: bees were presented successively with both odorants without reward. **P < 0.01; ***P < 0.005. n ¼ 86 (saline) and 79
(procaine).
odour memory (Cano-Lozano et al., 2001; Menzel, 2001; Locatelli
et al., 2005). However, this might explain why procaine-treated bees
could not perform reversal learning: because they started phase 2 with
a low level of responses to A, they might not have reduced sufficiently
their response rates to attain a level lower than that observed in
response to B at the end of phase 2. Thus, we asked whether this
impaired retrieval could be the actual cause of the absence of reversal
by looking specifically at the performance of bees with intact retrieval,
i.e. those that did respond to A at the beginning of phase 2. Hence, we
analysed again our data, considering only the subgroup of bees that
responded to A at the beginning of phase 2 (respectively, 67% and
63% of saline- and procaine-treated bees). The corresponding learning
curves (Fig. 6A) clearly show that, in this subgroup of bees, reversal
learning was impaired after procaine treatment (P > 0.05), but not
after saline treatment (data not shown; P < 0.05; Wilcoxon). The
same trend was found in the responses during the retention test: no
significant difference was found between the response rates to A and B
of procaine-treated bees (Fig. 6A), while the saline-treated bees
correctly discriminated (data not shown; P ¼ 0.001; Wilcoxon). Thus,
the effect of MB blockade on reversal learning can be dissociated from
that on retrieval.
Blocking the a- lobes does not impair extinction
Reversal learning requires that the CR to the previously rewarded
stimulus (B) should extinguish, while the response rate to the formerly
non-rewarded stimulus (A) should increase. Thus, a possibility is that
the impairment of reversal learning in procaine-injected animals might
be due to a failure in extinction learning for A. Because the responses
of procaine-treated bees to A did not diminish when no longer
rewarded (Fig. 5: phase 2), we tested whether extinction was impaired
in these animals. For this, we repeatedly presented unrewarded odours
A and B to individual bees during phase 2 (A– vs B–, Fig. 7). A first
repeated-measurement anova on all responses during phase 2
(extinction phase) revealed that saline- and procaine-treated groups
behaved differently (treatment effect: F1,105 ¼ 7.361, P < 0.01), and
that this difference depended on the CS presented (treatment–CS
interaction: F1,105 ¼ 4.839, P < 0.05). Saline-treated bees exhibited a
significant decrease in their response rates to A when it was no longer
rewarded (extinction of a previous acquisition memory, Stollhoff
et al., 2005; F1,52 ¼ 3.937, P < 0.01, anova). As expected,
responses to B– were very low and did not change. By contrast,
procaine-injected bees maintained a constantly increased level of
proboscis extension responses (PER) to A throughout all trials. As a
result, both groups reached similar response levels to A at the end of
phase 2 (group effect for trial 5: F1,105 ¼ 1.08, P > 0.05).
This result may imply a blockade of extinction by procaine. It may,
however, also be the consequence of an impaired retrieval as, like in
experiment 1, the initial response level was significantly lower in
procaine-treated bees (treatment effect on trial 1: F1,105 ¼ 14.28,
P < 0.005, anova). Then, extinction may not occur in bees that failed
to recall the initial association of odour A with the reward. Thus, we
calculated the response rates of both groups during the extinction
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 26, 3193–3206
Local anaesthetics and odour learning in honeybees 3201
Fig. 6. Reversal learning and extinction in procaine-treated bees with intact retrieval. Percentage of proboscis extension response (PER) evoked by odorants A
(filled bars) and B (open bars) in the subsets of bees that successfully responded to A at the beginning of phase 2, either in the reversal learning experiment (A) or
the extinction experiment (B). The levels of significance indicated for phase 2 in (B) correspond to the decrease of response levels to A across trials. ***P < 0.005.
The data for the subset of individuals in (A) were already included in the results presented in Fig. 5 (lower graphs), together with those from bees with impaired
retrieval. (A) n ¼ 44 (saline) and 24 (procaine); (B) n ¼ 41 (saline) and 23 (procaine).
phases, considering only those bees that responded to A with a CR
during the first extinction trial as we did previously for reversal
learning (Fig. 6B). In these bees, the response levels to A decreased
along the trials in both groups (block effect; saline: F1,40 ¼ 8.224,
P < 0.005; procaine: F1,23 ¼ 4.052, P < 0.01), reaching almost
identical levels in block 5 (saline: 63.4%, n ¼ 41; procaine: 62.5%,
n ¼ 24). Thus, whenever retrieval is not affected by procaine,
extinction occurs normally. This result was again observed in the
retention scores. It should be noted that bees still responded more
often to A than to B (P < 0.005 for both groups, Wilcoxon test),
suggesting that more trials may be necessary for completing
extinction.
Acquisition of differential conditioning remains unaffected
by a-lobe blockade
Bilateral procaine injections may produce a general learning deficit
such that the impairment observed for reversal learning may be
unspecific, rather than due to the particular learning task employed. To
test whether procaine-injected bees are still able to learn an odour, we
performed a discriminatory learning task during the second acquisition
phase. After injections, bees had to solve a simple differential
conditioning task, similar to that learned in phase 1, but using two new
odorants (C+ vs D–). This control is appropriate as it presents bees
with two discrimination problems (phase 1: A+ vs B–; phase 2: C+ vs
D–) as in reversal learning, but with the crucial difference of having
eliminated the ambiguity in stimulus valence underlying phase
transition in reversal learning (A+ fi A– and B) fi B+). In other
words, the new discrimination learning proposed to bees was an
elemental task for which MBs may not be required. In such conditions,
no effect of treatment was observed, neither during acquisition in
phase 2 nor during the retention test (Fig. 8). A treatment · blocks · CS repeated-measurement anova revealed that
treatment had no significant effect nor interaction with either of the
two other factors, thus indicating that procaine injections did not affect
the animals’ performance. This was confirmed by an anova for trial
5: animals from both groups were equally able to learn to respond to C
but not to D (CS effect: F1,97 ¼ 66.69, P < 0.01; treatment–CS
interaction: F1,97 ¼ 0.03, P > 0.05). They also retained this new
association for at least 1 h (P < 0.005 in both cases, Wilcoxon). These
results thus show that blocking neural activity within the a-lobes did
not abolish the ability to learn odours in a discriminatory classical
conditioning task. Furthermore, procaine-injected bees, like salineinjected bees, also retained their memory for the initial acquisition
phase (phase 1, A+ ⁄ B–; Fig. 8). They responded significantly more to
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 26, 3193–3206
3202 J.-M. Devaud et al.
Fig. 7. Extinction after blocking alpha lobes. Percentage of proboscis extension response (PER) evoked by odorants A (filled bars) and B (open bars). Phase 1 of
acquisition was identical to that in previous experiments (differential conditioning: A+ vs B–). In phase 2, saline- and procaine-injected bees were submitted to an
extinction protocol (A– vs B–). As before, a retention test was applied 1 h after the end of acquisition. n ¼ 53 in both groups. The levels of significance indicated for
phase 2 correspond to the decrease of response levels to A across trials. n.s., non-significant; **P < 0.01; ***P < 0.005.
odour A than to odour B during the retention tests (P < 0.005 in both
cases, Wilcoxon). The effect of procaine injection in the MBs was
therefore specific for reversal learning and not general to any kind of
olfactory discrimination.
tion. Thus, using the two complementary approaches of behavioural
analysis and patch-clamp, we have characterized these anaesthetics as
valuable tools to locally block neuronal activity and to trace down
specific memory traces in an insect brain.
Discussion
Procaine and lidocaine affect insect neuronal ion channels
To map specific behavioural tasks to particular regions within the
vertebrate brain, local anaesthetics have been successfully used as
general blockers of neuronal activity in a space- and time-dependent
manner (e.g. Griffin & Berry, 2004; Keller et al., 2004; Winters &
Bussey, 2005; Teixeira et al., 2006). We used here for the first time a
combined physiological and behavioural approach to show that local
anaesthetics inhibit spiking and voltage-sensitive ionic currents also in
insect neurons, and that procaine can be used to reversibly block
neural activity within the insect brain. Lidocaine and procaine, among
the most commonly used in mammals, have been recently shown to
impair basic forms of olfactory learning and motor responses in bees
(Müller et al., 2003). In that study, however, the local anaesthetics
were applied through systemic injections in the thorax, i.e. without
specifically targeting the brain and without knowing whether the drugs
affect neuronal spiking in honeybee neurons.
Here we focused on the effect of these drugs on adult MB neurons.
These same neurons were targeted in vivo in order to inactivate the
function of MBs during several olfactory learning tasks. Under such
conditions, we show that bees cannot perform reversal learning, while
they still retain the abilities for differential conditioning and extinc-
While general anaesthetics were frequently tested on insect preparations (Kirschfeld & Baierrogowski, 1987; Kirschfeld, 1987, 1988;
Rajaram & Nash, 2004; van Swinderen, 2006), local anaesthetics have
been rarely used. Our patch-clamp analysis on cultured Kenyon cells
(MB neurons) has revealed that lidocaine and procaine are efficient
blockers of neuronal spiking activity in insects, like in mammals. Our
recordings of sodium and potassium currents showed that both were
affected by the drugs in a dose-dependent manner, like in mammals
(Scholz, 2002). However, our data suggest some possible differences
in their respective specificities of action, as procaine appeared to be
more efficient to reduce sodium channels, while lidocaine seemed
more efficient on potassium currents. Procaine inhibited currents
through the voltage-sensitive Na+ channel more efficiently than
through K+ channels. The transient and sustained K+ current
components were equally sensitive to procaine. Voltage-gated sodium
and potassium channels are among the main targets of both drugs, as
indicated by previous studies (Scholz, 2002). Thus, although we have
not tested other possible molecular targets identified in some
vertebrate preparations [e.g. calcium channels, nicotinic or
N-methyl-d-aspartate (NMDA) receptors; Sugiyama & Muteki,
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 26, 3193–3206
Local anaesthetics and odour learning in honeybees 3203
Fig. 8. Differential conditioning after blocking alpha lobes. Percentage of proboscis extension response (PER) evoked by limonene and eugenol (during phase 1,
before treatment), and 1-heptanal and 1-nonanol (during phase 2, after treatment) of acquisition. In both phases, one odorant was rewarded while the other was not
(A+ vs B–, then C+ vs D–). Memory was measured at 1 h by recording the responses to all four unrewarded odorants. **P < 0.01; ***P < 0.005; n ¼ 49 in each
group.
1994; Komai & McDowell, 2001; Bischoff et al., 2003], our results
argue in favour of a conserved mode of action of procaine and
lidocaine across phyla.
The binding site for local anaesthetics resides within the channel
pore (Hille, 2001). Thus, local anaesthetics may act in a use-dependent
way as open channel blockers (lipophilic pathway). In addition, a useindependent, hydrophilic block of the closed channel state exists. Our
data suggest that insect channels do not require activation prior to
block. We always observed maximum block during the first depolarizing voltage pulses. Furthermore, we could not observe that procaine
altered the gating kinetics of either the Na+ or the K+ currents (neither
activation nor inactivation). Therefore, procaine appears to reduce the
channel conductance via a hydrophilic pathway in insect neurons.
More detailed biophysical experiments are required before we
understand the mode of action of local anaesthetics on insect ion
channels.
Output from the MBs is required during reversal learning
MBs are important centres for learning and memory in the insect brain
(e.g. Menzel, 2001; Heisenberg, 2003). In bees they receive input from
different sensory modalities, in particular from olfactory, visual and
gustatory sensory neuropils (for review, see Mobbs, 1982; Strausfeld
et al., 1998). Associations between an odorant and a sugar reward can
be formed within the MBs during classical olfactory conditioning
(Menzel, 2001). However, proper MB functioning is not an absolute
requirement for learning, as bees can learn elemental tasks in which
stimuli are unambiguously associated with reward or with its absence
(Giurfa, 2003), even with absent or severely atrophied MBs (Malun
et al., 2002b; Komischke et al., 2005) ) contrarily to flies (Heisenberg et al., 1985; deBelle & Heisenberg, 1994; Dubnau et al., 2001;
Schwaerzel et al., 2003). Nevertheless, this conclusion does not extent
to all forms of olfactory learning tested. Rather, it seems as if MBs
were specifically required for some learning tasks only (Komischke
et al., 2005), while dispensable in others. What is the specificity of
such tasks? One possibility is that MB-dependent tasks include several
different stimuli, such as several odours (Komischke et al., 2005).
Else, MB activity may be required to solve conflicts between
contradictory CS–US associations. Examples of the latter are opposite
association rules processed by each brain hemisphere (side-specific
conditioning, Sandoz & Menzel, 2001), or alternative presence or
absence of reward associated with the same CS, depending whether it
is presented alone or in a mixture (configural learning, Komischke
et al., 2003, 2005; Giurfa, 2003; Lachnit et al., 2004). If this was the
case, then we could predict that MBs would be necessary for other
olfactory tasks involving conflict resolution. Reversal learning belongs
to this category, as it involves the sequential processing of contradictory information about associations between CS and US. It was
previously shown that intact honeybees are able to perform this task
(Komischke et al., 2002).
Here we have shown that injecting procaine into the a-lobes (one of
the MB output pathways) profoundly affected the animals’ perfor-
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 26, 3193–3206
3204 J.-M. Devaud et al.
mance in this task. Acquisition of reversed CS–US associations was
completely impaired. Successful reversal learning implies that two
processes take place: (i) extinction, i.e. reduction of the response
probability to the initially rewarded odorant (A+, then A–); and (ii)
increase of the response probability to the newly rewarded stimulus
(B–, then B+). Because the response rates to B+ increase during
reversal learning and because discriminatory learning under procaine
is intact, we show that odours can be learned as prediction signals to a
reward. Furthermore, extinction was not completely impaired by
blocking MB function. Rather, some bees failed to remember the
initial A ⁄ sugar association shortly after injection, and thus could not
show extinction. Thus, procaine injections partially impaired retrieval
of the learned odour, in accordance with previous studies showing the
importance of intact output from the MBs for retrieval (Cano-Lozano
et al., 2001). As a consequence, they failed to extinguish their
responses to A+, as shown by the drop in the response rate between
the last A+ trial during phase 1 and the first A– trial of phase 2. This
partial retrieval failure is probably not due to an impaired consolidation because, after differential conditioning, retrieval of the initial
association rule (A+, B–) was intact after more than 2.5 h postinjection (see Fig. 8).
Importantly, this retrieval failure is unlikely to explain the reversal
learning impairment. Indeed, bees that do not remember phase 1
should perform phase 2 more easily, which would then be similar to a
normal differential conditioning task. Besides, those bees with intact
retrieval did not display a better performance. Thus, blocking a-lobes
leads to an impairment of reversal learning acquisition, which is
dissociable from the impairment of retrieval.
of the targeted brain regions. Here, drawing conclusions regarding the
role of MBs in reversal learning is only possible if we can safely
assume that our injections did not affect other brain areas. A
preliminary experiment with sulphorhodamine B, a fluorescent tracer
with a molecular weight similar to that of procaine, suggested that
massive leakage out of the MBs after injection was unlikely (see
Supplementary material, Fig. S1). However, we cannot exclude that
small volumes diluted in the head haemolymph would have been
unnoticed. If ALs would have then received procaine, odour
processing might have been altered. A trivial explanation for the
impaired performance of procaine-treated bees could then be a
reduced olfactory perception. However, this does not appear to be the
case, for two reasons. First, the treatment did not prevent odour
discrimination as shown by the results of our differential conditioning
experiment with two novel odours, C and D, under procaine treatment.
Second, procaine-treated bees displayed little generalization between
A and B. The response rates to B– remained low in the reversal
learning and extinction experiments, while they should have increased
if B were more easily confounded with A after treatment. Thus,
blocking MB function does not seem to crucially alter olfactory
discrimination, consistently with previous studies (Malun et al.,
2002b; Komischke et al., 2005).
Finally, procaine injections did not seem to affect sugar sensitivity
or motor abilities, as the proportion of bees displaying PER upon
sugar presentation did not differ according to the injection (respectively, 92.0% and 85.3%, P > 0.05, Mann–Whitney test). Hence, local
injections of procaine into an insect brain have transient and regionspecific consequences on behaviour, thus reflecting a likely temporary
inactivation of spike activity in the injected area.
Reversible and local effects of intracerebral procaine injections
The finding that retrieval of the initial associative rule was intact 2.5 h
after injection indicates that the action of procaine was reversible within
2.5 h. Our pilot experiments showed the effect of injecting the same
dose of procaine into the ALs to last at least 90 min. Thus, we conclude
that: (i) procaine remains effective on Kenyon cells during the whole
second phase of acquisition; (ii) procaine is progressively washed out
during the 1 h consolidation phase, i.e. between 1.5 and 2.5 h postinjection; (iii) the procaine effects are completely reversible. This
reversibility is also suggested by the analysis of individual responses
curves, which showed that some individuals that failed to respond
according to the initial rule at the first trial of phase 2 succeeded in
subsequent trials (data not shown). Hence, while some bees correctly
reduced their response probability to A, others actually increased it,
leading to a steady response rate reflected at the group level.
Our data, together with those of Müller et al. (2003), indicate that
procaine is a powerful tool for local and transient inactivation of
localized brain regions. Its use provides a valuable alternative to the
techniques used so far, i.e. local cooling of brain regions (Menzel
et al., 1974; Erber et al., 1980) and chemical ablation of the MBs by
hydroxyurea treatment during development (Malun et al., 2002a,b).
On the one hand, local cooling proved to be helpful to establish the
dynamics of the olfactory memory trace after a single-trial experiment
because of its rapid and very short-lived effect, but would not be
helpful to block MB function for longer durations, as required in our
multiple-trial training. On the other hand, MB ablation during
development results in a permanent loss of MBs, unsuitable for
a time-controlled blockade during only some phases of learning ⁄
memory processes.
The consequences of local injections of procaine or any other
molecule should be analysed with caution in terms of the function(s)
A new role for MBs in olfactory learning
Because retrieval seems to be only transiently impaired, a memory of
the initial association rule is again accessible in later trials of phase 2
in many individuals. Thus, retrieval impairment alone cannot explain
the failure to reverse the rule. Indeed, complete blockade of retrieval
should facilitate learning of the new rule, as the memory of the first
phase (A+ vs B–) would not interfere with reversal learning. In such a
case, procaine-treated bees should behave exactly the same way in the
two protocols used, i.e. reversal learning and differential conditioning
(control), which is obviously not the case. Thus, other processes that
are controlled by MB function must be required for reversal learning,
which are impaired by procaine.
First, it is noteworthy that extinction seems to take place normally
without functional MBs. This is in contrast with the situation in
Drosophila (Schwaerzel et al., 2002). This discrepancy may be a
consequence of the use of different networks for learning processes in
these species. Because CS–US associations may form outside the ALs
in bees, the neural processes of ‘re-learning’ that are believed to
underlie extinction (Myers & Davis, 2002; Stollhoff et al., 2005) may
take place outside the MBs, in contrast to flies. Second, during our
differential conditioning experiment four odours were involved, while
reversal learning involves only two odours. Thus, the efficiency of
procaine-treated bees to realize the former but not the latter strongly
argues against the number of stimuli as the critical parameter for MB
requirement in olfactory associative learning, as proposed earlier
(Komischke et al., 2005). Another critical difference is that, during
differential conditioning, processing the new CS–US associations did
not generate any contradiction with those learnt before. Thus, these
results suggest a possible role for MBs for the resolution of conflicting
information. In honeybees, the question of the respective roles of MBs
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 26, 3193–3206
Local anaesthetics and odour learning in honeybees 3205
and ALs in olfactory learning has been raised since early experiments
showed that several memory traces of a given CS–US association
could be formed in both structures (Hammer & Menzel, 1998; see
Menzel, 2001), but that normal AL function was sufficient to form and
consolidate the trace (Malun et al., 2002b; Komischke et al., 2005).
This situation contrasts with those found in fruitflies (Heisenberg
et al., 1985; de Belle & Heisenberg, 1994; Dubnau et al., 2001) and
crickets (Scotto-Lomassesse et al., 2003), where ‘simple’ associative
learning such as the differential conditioning task used here as a
control cannot be learned without normal MB activity. Thus, the
present results suggest that in the honeybee, MBs might be necessary
during acquisition of ‘complex’ tasks, i.e. those requiring the
resolution of contradictory CS–US associations. This functional
recruitment of additional areas in more demanding learning tasks
reminds several observations on memory consolidation after fear
conditioning in rodents. While not required for the processing of
associative components of fear conditioning (mostly done in the
amygdala), the hippocampus appears to be selectively involved in
consolidation and retrieval of the contextual components (Corcoran &
Maren, 2001, 2004; Bast et al., 2003; Malin & McGaugh, 2006).
From our hypothesis it can be predicted that other learning tasks
involving conflict resolution between multiple CS–US associations are
MB dependent (Giurfa, 2003, 2007). This has already been hypothesized for side-specific conditioning, in which the two halves of the brain
must deal with different learning rules (Komischke et al., 2003). From
this perspective, MBs would play more complex and diverse functions
in olfactory learning than previously expected. In particular, they may
compute a comparison between conflicting information in order to lead
to an appropriate motor output. This is supported by the indication from
our data that some memory trace of the first rule (A+ B–) must be
normally present in the MBs when acquisition of the new rule (A– B+)
starts. Optophysiological recordings have shown learning-induced
plasticity in odorant representations in the MBs as early as 10 min after
conditioning (Faber & Menzel, 2001). It is thus conceivable that reversal
learning involves changes in odorant coding by MB neurons, leading to
form different memory traces (e.g. of A+, B–, C+, D–) or to eliminate
inaccurate ones whenever conflict arises (i.e. A+, B– in reversal
learning). Further experiments will provide insights into the mechanisms underlying conflict resolution in these neural circuits.
Supplementary material
The following supplementary material may be found on
http://www.blackwell-synergy.com
Fig. S1. Fluorescent staining of a-lobes with sulphorodamine B,
showing that diffusion outside the lobes is limited.
Acknowledgements
The authors thank Marion Ganz for technical assistance with cell culturing. We
are grateful to Stefanie Seifert for experimental support on the effects of
procaine on spiking, and Steven Sievers for providing additional data on
procaine actions on ionic currents. We thank Drs Randolf Menzel and Valerie
Raymond-Delpech for critically reading the manuscript. This work was funded
by a DAAD ⁄ French Government PROCOPE grant to B.G. and M.G. J.M.D.
and M.G. also thank the support of CNRS and the Paul Sabatier University.
Abbreviations
ALs, antennal lobes; CR, conditioned response; CS, conditioned stimulus;
MBs, mushroom bodies; PER, proboscis extension response; UR, unconditioned response; US, unconditioned stimulus.
References
Anwyl, R. (1977) The effect of foreign cations, pH and pharmacological agents
on the ionic permeability of an excitatory glutamate synapse. J. Physiol.,
273, 389–404.
Bast, T., Zhang, W.N. & Feldon, J. (2003) Dorsal hippocampus and classical
fear conditioning to tone and context in rats: effects of local NMDA-receptor
blockade and stimulation. Hippocampus, 13, 657–675.
deBelle, J.S. & Heisenberg, M. (1994) Associative odor learning in
Drosophila abolished by chemical ablation of mushroom bodies. Science,
263, 692–695.
Bischoff, U., Brau, M.E., Vogel, W., Hempelmann, G. & Olschewski, A. (2003)
Local anaesthetics block hyperpolarization-activated inward current in rat
small dorsal root ganglion neurones. Br. J. Pharmacol., 139, 1273–1280.
Bitterman, M.E., Menzel, R., Fietz, A. & Schäfer, S. (1983) Classical
conditioning of proboscis extension in honeybees (Apis mellifera). J. Comp.
Psych., 97, 107–119.
Bohbot, V., Otahal, P., Liu, Z., Nadel, L. & Bures, J. (1996) Electroconvulsive
shock and lidocaine reveal rapid consolidation of spatial working memory in
the water maze. Proc. Natl Acad. Sci. USA, 93, 4016–4019.
Bräu, M.E., Vogel, W. & Hempelmann, G. (1998) Fundamental properties of
local anesthetics: half-maximal blocking concentrations for tonic block of
Na+ and K+ channels in peripheral nerve. Anesth. Analg., 87, 885–889.
Cano-Lozano, V., Bonnard, E., Gauthier, M. & Richard, D. (1996)
Mecamylamine-induced impairment of acquisition and retrieval of olfactory
conditioning in the honeybee. Behav. Brain Res., 81, 215–222.
Cano-Lozano, V., Armengaud, C. & Gauthier, M. (2001) Memory impairment
induced by cholinergic antagonists injected into the mushroom bodies of the
honeybee. J. Comp. Physiol. [A], 187, 249–254.
Corcoran, K.A. & Maren, S. (2001) Hippocampal inactivation disrupts
contextual retrieval of fear memory after extinction. J. Neurosci., 21,
1720–1726.
Corcoran, K.A. & Maren, S. (2004) Factors regulating the effects of
hippocampal inactivation on renewal of conditional fear after extinction.
Learn. Mem., 11, 598–603.
Cuevas, J. & Adams, D.J. (1994) Local anaesthetic blockade of neuronal
nicotinic ACh receptor-channels in rat parasympathetic ganglion cells. Br. J.
Pharmacol., 111, 663–672.
Daumas, S., Halley, H., Frances, B. & Lassalle, J.M. (2005) Encoding,
consolidation, and retrieval of contextual memory: differential involvement
of dorsal CA3 and CAI hippocampal subregions. Learn. Mem., 2, 375–382.
Dubnau, J., Grady, L., Kitamoto, K. & Tully, T. (2001) Disruption of
neurotransmission in Drosophila mushroom body blocks retrieval but not
acquisition of memory. Nature, 411, 476–480.
Erber, J., Masuhr, T. & Menzel, R. (1980) Localization of short-term memory
in the brain of the bee, Apis mellifera. Physiol. Entomol., 5, 343–358.
Faber, T. & Menzel, R. (2001) Visualizing mushroom body response to a
conditioned odor in honeybees. Naturwissenschaften, 88, 472–476.
Fatt, P. & Katz, B. (1953) The electrical properties of crustacean muscle fibres.
J. Physiol., 28, 171–204.
Franks, C.J., Pemberton, D., Vinogradova, I., Cook, A., Walker, R.J. & HoldenDye, L. (2002) Ionic basis of the resting membrane potential and action
potential in the pharyngeal muscle of Caenorhabditis elegans. J. Neurophysiol., 87, 954–961.
Giurfa, M. (2003) Cognitive neuroethology: dissecting non-elemental learning
in a honeybee brain. Curr. Opin. Neurobiol., 13, 1–10.
Giurfa, M. (2007) Behavioral and neural analysis of associative learning in the
honeybee: a taste from the magic well. J. Comp. Physiol. A, 193, 801–824.
Griffin, A.L. & Berry, S.D. (2004) Inactivation of the anterior cingulate cortex
impairs extinction of rabbit jaw movement conditioning and prevents
extinction-related inhibition of hippocampal activity. Learn. Mem., 11, 604–
610.
Grünewald, B. (2003) Differential expression of voltage-sensitive K+ and Ca2+
currents in neurons of the honeybee olfactory pathway. J. Exp. Biol., 206,
117–129.
Hamill, O.P., Neher, E., Marty, A., Sakman, B. & Sigworth, F.J. (1981)
Improved patch-clamp techniques for high-resolution current recording from
cells and cell-free membrane patches. Pflügers Arch. Eur. J. Physiol., 391,
85–100.
Hammer, M. (1993) An identified neuron mediates the unconditioned stimulus
in associative olfactory learning in honeybees. Nature, 366, 59–63.
Hammer, M. & Menzel, R. (1998) Multiple sites of associative odor learning as
revealed by local brain microinjections of octopamine in honeybees. Learn.
Mem., 5, 146–156.
Heisenberg, M. (2003) Mushroom body memoir: from maps to models. Nat.
Rev. Neurosci., 4, 266–275.
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 26, 3193–3206
3206 J.-M. Devaud et al.
Heisenberg, M., Borst, A., Wagner, S. & Byers, D. (1985) Drosophila
mushroom body mutants are deficient in olfactory learning. J. Neurogenet.,
2, 1–30.
Hille, B. (1966) Common mode of action of three agents that decrease the
transient change in sodium permeability in nerves. Nature, 210, 1220–1222.
Hille, B. (2001) Classical mechanisms of block. Ion Channels of Excitable
Membranes. Sinauer Associate, Sunderland, MA, pp. 503–536.
Keller, M., Perrin, G., Meurisse, M., Ferreira, G. & Levy, F. (2004) Cortical and
medial amygdala are both involved in the formation of olfactory offspring
memory in sheep. Eur. J. Neurosci., 20, 3433–3441.
Kirschfeld, K. (1987) The action of volatile anesthetics on the resting and
receptor potential in insect photoreceptors. Naturwissen, 74, 29.
Kirschfeld, K. (1988) The influence of general-anesthetics on neurons sensitive
to motion in the visual-system of the fly. Naturwissen, 75, 468–470.
Kirschfeld, K. & Baierrogowski, V. (1987) The neuronal basis of the anesthetic
state – a comparative physiological approach. 1. The influence of anesthetics
on the behavior of insects. Biol. Cybern., 55, 345–354.
Komai, H. & McDowell, T.S. (2001) Local anesthetic inhibition of voltageactivated potassium currents in rat dorsal root ganglion neurons. Anesthesiology, 94, 1089–1095.
Komischke, B., Giurfa, M., Lachnit, H. & Malun, D. (2002) Successive
olfactory reversal learning in honeybees. Learn. Mem., 9, 122–129.
Komischke, B., Sandoz, J.-C., Lachnit, H. & Giurfa, M. (2003) Non-elemental
processing in olfactory discrimination tasks need bilateral input in honeybees. Behav. Brain Res., 145, 135–143.
Komischke, B., Sandoz, J.-C., Malun, D. & Giurfa, M. (2005) Partial unilateral
lesions of the mushroom bodies affect olfactory learning in honeybees Apis
mellifera L. Eur. J. Neurosci., 21, 477–485.
Kreissl, S. & Bicker, G. (1992) Dissociated neurons of the pupal honeybee
brain in cell culture. J. Neurocytol., 21, 545–556.
Kuwabara, M. (1957) Bildung des bedingten Reflexes von Pavlovs Typus bei
der Honigbiene, Apis mellifica. J. Fac. Sci. Hokkaido Univ. Ser. VI Zool., 13,
458–464.
Lachnit, H., Giurfa, M. & Menzel, R. (2004) Odor processing in honeybees: is
the whole equal to, more than, or different from the sum of its parts ? Adv.
Study Behav., 34, 241–264.
Lapied, B., Grolleau, F. & Sattelle, D.B. (2001) Indoxacarb, an oxadiazine
insecticide, blocks insect neuronal sodium channels. Br. J. Pharmacol., 132,
587–595.
Leuwer, M., Haeseler, G., Hecker, H., Bufler, J., Dengler, R. & Aronson, J.K.
(2004) An improved model for the binding of lidocaine and structurally
related local anaesthetics to fast-inactivated voltage-operated sodium channels, showing evidence of cooperativity. Br. J. Pharmacol., 141, 47–54.
Locatelli, F., Bundrock, G. & Müller, U. (2005) Focal and temporal release of
glutamate in the mushroom bodies improves olfactory memory in Apis
mellifera. J. Neurosci., 25, 11614–11648.
Lunney, G.H. (1970) Using analysis of variance with a dichotomous dependent
variable: an empirical study. J. Educat. Meas., 7, 263–269.
Malin, E.L. & McGaugh, J.L. (2006) Differential involvement of the
hippocampus, anterior cingulate cortex, and basolateral amygdala in memory
for context and footshock. Proc. Natl Acad. Sci. USA, 103, 1959–1963.
Malun, D., Plath, N., Giurfa, M., Moseleit, A.D. & Müller, U. (2002a)
Hydroxyurea-induced partial mushroom body ablation in the honeybee Apis
mellifera: volumetric analysis and quantitative protein determination.
J. Neurobiol., 50, 31–44.
Malun, D., Giurfa, M., Galizia, C.G., Plath, N., Brandt, R., Gerber, B. &
Eisermann, B. (2002b) Hydroxyurea-induced partial mushroom body
ablation does not affect acquisition and retention of olfactory differential
conditioning in honeybees. J. Neurobiol., 53, 343–360.
Martin, C., Gervais, R., Messaoudi, B. & Ravel, N. (2006) Learning-induced
oscillatory activities correlated to odour recognition: a network activity. Eur.
J. Neurosci., 23, 1801–1810.
Menzel, R. (1999) Memory dynamics in the honeybee. J. Comp. Physiol., 185,
323–340.
Menzel, R. (2001) Searching for the memory trace in a mini-brain, the
honeybee. Learn. Mem., 8, 53–62.
Menzel, R., Erber, J. & Masuhr, T. (1974) Learning and memory in the
honeybee. In Barton Browne, L. (Ed.), Experimental Analysis of Insect
Behaviour. Springer, Berlin, pp. 195–217.
Menzel, R. & Giurfa, M. (2001) Cognitive architecture of a mini-brain: the
honeybee. Trends Cogn. Sci., 5, 62–71.
Mobbs, P.G. (1982) The brain of the honeybee Apis mellifera I. The
connections and spatial organization of the mushroom bodies. Philos. Trans.
R. Soc. Lond. B Biol. Sci., 298, 309–354.
Müller, D., Staffeldt, D., Fiala, A. & Menzel, R. (2003) Procaine impairs
learning and memory consolidation in the honeybee. Brain Res., 977, 124–
127.
Myers, K.M. & Davis, M. (2002) Behavioral and neural analysis of extinction.
Neuron, 14, 567–584.
Nishizawa, N., Shirasaki, T., Nakao, S., Matsuda, H. & Shingu, K. (2002) The
inhibition of the N-methyl-D-aspartate receptor channel by local anesthetics
in mouse CA1 pyramidal neurons. Anesth. Analg., 94, 325–330.
Pelz, C., Jander, J., Rosenboom, H., Hammer, M. & Menzel, R. (1999) IA in
Kenyon cells of the mushroom body of honeybees resembles shaker currents:
kinetics, modulation by K+, and simulation. J. Neurophysiol., 81, 1749–
1759.
Rajaram, S. & Nash, H.A. (2004) A specific alteration in the electroretinogram
of Drosophila melanogaster is induced by halothane and other volatile
general anesthetics. Anesth. Analg., 98, 1705–1711.
Sandoz, J.-C. & Menzel, R. (2001) Side-specificity of olfactory learning in
the honeybee: generalization between odors and sides. Learn. Mem., 8,
286–294.
Schäfer, S., Rosenboom, H. & Menzel, R. (1994) Ionic currents of Kenyon
cells from the mushroom body of the honeybee. J. Neurosci., 14, 4600–4612.
Scholz, A. (2002) Mechanisms of (local) anaesthetics on voltage-gated sodium
and other ion channels. Br. J. Anaesth., 89, 52–61.
Schwaerzel, M., Heisenberg, M. & Zars, T. (2002) Extinction antagonizes
olfactory memory at the subcellular level. Neuron, 35, 951–960.
Schwaerzel, M., Monastirioti, M., Scholz, H., Friggi-Grelin, F., Birman, S. &
Heisenberg, M. (2003) Dopamine and octopamine differentiate between
aversive and appetitive olfactory memories in Drosophila. J. Neurosci., 23,
10495–10502.
Scotto-Lomassesse, S., Strambi, C., Strambi, A., Aouane, A., Augier, R.,
Rougon, G. & Cayre, M. (2003) Suppression of adult neurogenesis impairs
olfactory learning and memory in an insect. J. Neurosci., 23, 9289–9296.
Sheets, M.F. & Hanck, D.A. (2003) Molecular action of lidocaine on the
voltage sensors of sodium channels. J. Gen. Physiol., 121, 163–175.
Stollhoff, N., Menzel, R. & Eisenhardt, D. (2005) Spontaneous recovery from
extinction depends on the reconsolidation of the acquisition memory in an
appetitive learning paradigm in the honeybee (Apis mellifera). J. Neurosci.,
25, 4485–4492.
Strausfeld, N.J., Hansen, L., Li, Y., Gomez, R.S. & Ito, K. (1998) Evolution,
discovery and interpretations of arthropod mushroom bodies. Learn. Mem.,
5, 11–37.
Sugiyama, K. & Muteki, T. (1994) Local anesthetics depress the calcium
current of rat sensory neurons in culture. Anesthesiology, 80, 1369–1378.
van Swinderen, B. (2006) A succession of anesthetic endpoints in the
Drosophila brain. J. Neurobiol., 66, 1195–1211.
Taylor, R.E. (1959) Effect of procaine on electrical properties of squid axon
membrane. Am. J. Physiol., 196, 1071–1078.
Teixeira, C.M., Pomedli, S.R., Maei, H.R., Kee, N. & Frankland, P.W. (2006)
Involvement of the anterior cingulate cortex in the expression of remote
spatial memory. J. Neurosci., 26, 7555–7564.
Theander, S., Fåhraeus, C.& Grampp, W. (1996) Analysis of leak current
properties in the lobster stretch receptor neurone. Acta Physiol. Scand., 157,
493–509.
Uusitalo, R.O., Juusola, M. & Weckstrom, M. (1995) Graded responses and
spiking properties of identified first-order visual interneurons of the fly
compound eye. J. Neurophysiol., 73, 1782–1792.
Washio, H. (1972) The ionic requirements for the initiation of action potentials
in insect muscle fibers. J. Gen. Physiol., 59, 121–134.
Winters, B.D. & Bussey, T.J. (2005) Transient inactivation of perirhinal cortex
disrupts encoding, retrieval, and consolidation of object recognition memory.
J. Neurosci., 25, 52–61.
Wolbarsht, M.L. & Hanson, F.E. (1965) Electrical activity in the chemoreceptors of the blowfly. 3. Dendritic action potentials. J. Gen. Physiol., 48, 673–
683.
Wüstenberg, D., Boytcheva, M., Grünewald, B., Byrne, J.H., Menzel, R. &
Baxter, D.A. (2004) Current- and voltage-clamp recordings and computer
simulations of kenyon cells in honeybee. J. Neurophysiol., 92, 2589–2603.
Zar, J.H. (1999) Biostatistical Analysis, 4th Edn. Prentice Hall, Upper Saddle
River, New Jersey, USA.
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 26, 3193–3206