Neurobiology of Learning and Memory 87 (2007) 72–77

www.elsevier.com/locate/ynlme

Active avoidance conditioning in zebraWsh (Danio rerio)

Xiaojuan Xu ¤, Theodora Scott-Scheiern, Leah Kempker, Katie Simons
Department of Psychology, Grand Valley State University, Allendale, MI 49401, USA

Received 5 January 2006; revised 14 June 2006; accepted 14 June 2006

Available online 24 July 2006

Abstract

The zebraWsh represents a potentially useful organism for studying genes involved in learning and memory function in vertebrates,
because a number of genetic techniques in zebraWsh have been developed to produce a wide variety of genetic mutants. While zebraWsh
mutants are being developed, behavioral studies on learning and memory function in zebraWsh are in urgent need. The present study
investigated active avoidance conditioning in normal zebraWsh. ZebraWsh were trained to swim from a lighted (CS) compartment to a
dark compartment to avoid an electrical body shock (US) in a shuttle-box that consisted of a water-Wlled tank separated by an opaque
barrier into two equal compartments. By varying the number of trials per training session and the duration of the intertrial interval,
Experiments 1 and 2 showed that, with the CS, US, and intertrial interval being 12 s, zebraWsh learned avoidance responses within a train-
ing session consisting of 30 trials and retained the avoidance responses. Experiment 3 showed that zebraWsh learned avoidance responses
following the association between the CS of light and the US of shock in the avoidance conditioning paradigm. Using the avoidance con-
ditioning paradigm, Experiment 4 investigated the amnestic eVects of N-methyl-D-aspartate receptor antagonist MK-801 and nitric oxide
synthase inhibitor L-NAME in zebraWsh. Experiment 4 showed that post-training injection of L-NAME signiWcantly impaired retention
of avoidance responses while MK-801 did not, conWrming previous results with other vertebrates. The results of the present study suggest
the similar involvements of neurochemicals in learning and memory among vertebrates. Thus, future studies with zebraWsh mutants may
identify genes involved in learning and memory in vertebrates.
© 2006 Elsevier Inc. All rights reserved.

Keywords: ZebraWsh; Active avoidance conditioning; Shuttle-box; N-methyl-D-aspartate receptor; Nitric oxide

1. Introduction zebraWsh, such a strategy may be tried in zebraWsh. While

Of the several hundred thousand proteins manufac-
tured by a neuron, some may be more important than
others when it comes to learning and memory. It is possi-
ble that some proteins are uniquely involved in learning
and memory. One way to identify a learning and memory
protein is to delete genes one at a time and see if speciWc
learning and memory deWcits result. This strategy has
been tried with the fruit Xy, resulting in implication of the
cyclic adenosine monophosphate (cAMP) system in
learning and memory (Goodwin et al., 1997). With
increasing understanding of the genetic background of
*
Corresponding author. Fax: +1 616 331 2480.
E-mail address: XUX@GVSU.EDU (X. Xu).
genetic mutants are being developed in zebraWsh, behav-
ioral studies on learning and memory function in
zebraWsh are in urgent need.
Active avoidance conditioning in goldWsh is a well-estab-
lished behavioral paradigm. Using avoidance procedures,
investigators have discovered a good deal about learning
and memory in goldWsh. The behavioral methods and Wnd-
ings with goldWsh may serve as the groundwork for study-
ing learning and memory function in zebraWsh. Thus, based
on the behavioral method and Wndings of the active avoid-
ance paradigm in goldWsh (AgranoV, 1980; Xu, Boshoven,
Lombardo, & Spranger, 1998, 1999, 2001, 2003), the present
study investigated active avoidance conditioning in
zebraWsh.
Behavioral studies show that administrations of
N-methyl-D-aspartate (NMDA) receptor antagonists block

1074-7427/$ - see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.nlm.2006.06.002
 X. Xu et al. / Neurobiology of Learning and Memory 87 (2007) 72–77
certain forms of associative learning (Kim, Decola, Lan-
deira-Fernandez, & Fanselow, 1991; Morris, Anderson,
Lynch, & Baudry, 1986; Shapiro & Caramanos, 1990; Xu &
Davis, 1992; Xu, Klinger, & Davis, 1995, 1998, 2003).
Behavioral studies also show that administrations of drugs
to block nitric oxide (NO) production or protein kinases
impair certain types of learning and memory (Baldwin,
Sadeghian, Holahan, & Kelly, 2002; Bernabeu, de Stein,
Fin, Izquierdo, & Medina, 1995; Koh, Thiele, & Bernstein,
2002; Schafe, Nadel, Sullivan, Harris, & LeDoux, 1999; Xu
et al., 1998, Xu, Boshoven, & Gunn, 1999, 2001). Thus, the
present study also investigated the involvements of NMDA
receptors and NO in avoidance conditioning in zebraWsh.
2. Materials and methods
2.1. Subjects and experimental drugs
ZebraWsh (Danio rerio) were obtained from local pet stores. ZebraWsh
were kept in large tanks for several weeks prior to experiments. During
experiments, zebraWsh were kept in individual compartments of parti-
tioned tanks at 26 § 1 °C with a 12 h light–dark cycle (0700–1900 light).
Experiments were conducted during the light cycle. Dizocilpine maleate
(MK-801) and nitro-L-arginine methyl-ester (L-NAME) (Sigma–Aldrich,
St. Louis, MO) were dissolved in 3 l saline and administered intramuscu-
larly with a 30 gauge needle and a 10 l Hamilton syringe.
2.2. Apparatus
ZebraWsh were trained and tested individually in three identical
zebraWsh shuttle-boxes connected to a programmer/shocker unit. The
zebraWsh shuttle-box consisted of a water-Wlled tank (18 £ 7.5 £ 10 cm)
separated by an opaque barrier (7.5 £ 10 cm) into two equal compart-
ments. The barrier was raised 0.6 cm above the Xoor of the tank to allow
zebraWsh to swim freely from one side of the tank to the other. The cross-
ing movement of zebraWsh was monitored by infrared light beams and
their corresponding detectors located on the long sides of the tank. There
was a light at each end of the tank and there were two stainless steel elec-
trode plates (6.5 £ 4 cm) at each of the long sides of each compartment.
2.3. Active avoidance paradigm
ZebraWsh were placed in the shuttle-boxes for 5 min, and then a trial
began with the onset of the light on the side of the Wsh’s location. After the
light was on for 12 s, a repetitive mild electrical shock (0.73 V/cm AC,
pulsed 100 ms on and 1400 ms oV) was administered, along with the light,
for 12 s through the water by means of electrodes. At the end of 24 s or at a
crossing response by zebraWsh during the 24 s, both the light and electrical
shock were switched oV and the trial ended. After an intertrial interval
(ITI) ranging from 12 to 30 s, another trial began.
ZebraWsh initially swam through the opening only after receiving sev-
eral shocks. The crossing response made after the onset of both light signal
and electrical shock to escape the electrical body shock is here deWned as
an escape response. During the training sessions, zebraWsh gradually
learned to swim from the lighted end to the dark end to avoid the electrical
body shock. The crossing response made after the onset of the light signal,
but before the onset of electrical shock to avoid the electrical body shock,
is here deWned as an avoidance response. The crossing response made dur-
ing an ITI is here deWned as an ITI crossing, and an ITI crossing was
immediately followed by a single body shock to reduce the rate of cross-
ings during ITI. The measurements were the number of avoidances,
escapes and ITI crossings. All experiments were fully automated through
the programmer/shocker unit and a Gateway 2000 P5-100 computer that
programmed stimuli, monitored and recorded behavior of zebraWsh.
73
ZebraWsh were trained semiweekly on Experimental days 1, 3 or 4, and 8.
Percentage of avoidance responses was used as an indicator of learning.
2.4. Experiment 1
2.4.1. EVects of number of trials per training session on learning
In goldWsh, we found that too few trials per training session resulted in
no or unobservable learning over several weeks, but too many trials per
training session did not produce further learning in a session and may
cause distress in Wsh. For goldWsh, 20 trials per training session seem to
produce faster learning (Xu et al., 1998, 1999, Xu, Russell, Bazner, & Ham-
ilton, 2001, 2003). With an active avoidance paradigm that did not deliver
a body shock following an ITI crossing, our preliminary studies with
zebraWsh explored the eVects of 15 vs. 30 trials per training session, and the
eVects of diVerent durations of ITI ranging from 30 to 96 s on the rate of
learning. The preliminary results showed that 30 trials per training session
with ITI being 30 s produced best learning. However, all groups of Wsh
showed very high rate of ITI crossings. To reduce the rate of crossings dur-
ing ITI, the computer program was revised and any ITI crossing was
immediately followed by a single electrical body shock. The series of
experiments in the present study used the active avoidance paradigm that
delivered a body shock immediately following an ITI crossing. While
keeping the ITI at 30 s, this experiment tried to determine the optimal
number of trials per training session in producing fast learning. A total of
30 zebraWsh were randomly divided into three groups (n D 10 per group):
20-trial, 30-trial, and 40-trial groups. The three groups received 20, 30, or
40 trials per training session for three semiweekly sessions.
2.4.2. Results
A two-way ANOVA with one between (three groups) and one repeated
measures (three sessions) on the avoidance responses indicated a signiW-
cant session diVerence [F(2, 27) D 34.2, p < 0.01]. The two-way repeated
measures ANOVA with multiple comparisons on the within factor indi-
cated that the increase in avoidance responses from Session 1 to Session 2
was statistically signiWcant (p < 0.01), but the increase in avoidance
responses from Session 2 to Session 3 was not (Fig. 1). The two-way
repeated measures ANOVA with multiple comparisons on the between
factor indicated that the 30-trial group learned avoidance responses sig-
niWcantly faster than the 20-trial group (p < 0.05), but the 40-trial group
did not learn the avoidance responses faster than the 30-trial group
(Fig. 1). The mean percentages of ITI crossings of the 20-trial group were
77.5, 49.5, and 50.5% for the three sessions, respectively. The mean percent-
Fig. 1. Avoidance learning of zebraWsh trained with diVerent numbers of
trials per session. Each point represents the mean percentage of avoidance
responses § SE for 10 Wsh.
74 X. Xu et al. / Neurobiology of Learning and Memory 87 (2007) 72–77
ages of ITI crossings of the 30-trial group were 48.7, 38.7, and 31.0% for
the three sessions, respectively. The mean percentages of ITI crossings of
the 40-trial group were 60.0, 56.3, and 43.0% for the three sessions, respec-
tively. Thus, the 30-trial group learned avoidance responses faster than the
20-trial group without exhibiting increased ITI crossings, while the 40-trial
group did not show faster learning or lower ITI crossings than the 30-trial
group.
The 20-trial group received Training Session 4 and reached the same
level of performance as the other two group groups (data not shown).
Thus, the 20-trial group were capable of higher performance. It was also
observed that close to the end of a 40-trial session, some Wsh in the 40-trial
group stopped crossing responses.
2.5. Experiment 2
2.5.1. EVects of duration of intertrial interval on learning
In goldWsh, the ITI between 25 and 55 s produced good learning (Xu
et al., 1998, 1999, 2001, Xu, Bazner, Qi, Johnson, & FreidhoV, 2003). The
preliminary data showed that 30-s ITI produced faster learning than a
longer duration of ITI in zebraWsh. To determine whether zebraWsh
showed faster learning with a shorter duration of ITI, this experiment
compared the eVect of ITI being 30 s with that of ITI being 12 s on learn-
ing. A total of 20 zebraWsh were randomly divided into two groups (n D 10
per group): 30-s ITI and 12-s ITI groups. The 30-s ITI group received
trainings with ITI being 30 s, and the 12-s ITI group received trainings
with ITI being 12 s. The groups received three semiweekly training sessions
with 30 trials per session.
2.5.2. Results
A two-way ANOVA with one between factor (two groups) and one
repeated measures (three sessions) on the avoidance responses indicated a
signiWcant session diVerence [F(2, 18) D 16.1, p < 0.01]. The two-way
repeated measures ANOVA with multiple comparisons on the within fac-
tor indicated that the increase in avoidance responses from Session 1 to
Session 2 was signiWcant (p < 0.01), but the increase in avoidance responses
from Session 2 to Session 3 was not (Fig. 2). Fig. 2 showed that the12 s ITI
group learned avoidance responses slightly faster than the 30 s ITI group,
although the diVerence was not statistically signiWcant. The rates of ITI
crossings did not increase over sessions in either group, and the 12-s ITI
group showed slightly lower ITI crossings (ranging from 30.0 to 38.0%)
than the 30-s ITI group (ranging from 31.0 to 48.7%).
Fish in both groups did not show much increase in avoidance
responses from Session 2 to Session 3 (Fig. 2). For the 12-s ITI group, the
Fig. 2. Avoidance learning of zebraWsh trained with diVerent durations of
ITI. Each point represents the mean percentage of avoidance
responses § SE for 10 Wsh.
mean percentage of avoidance responses for the Wrst 10 trials during Ses-
sion 2 was 72.0%, which was similar to mean percentages of avoidance
responses for 30 trials during Sessions 2. For the 30-s ITI group, the mean
percentage of avoidance responses for the Wrst 10 trials during Session 2
was 62.0%, which was similar to mean percentages of avoidance responses
for 30 trials during Sessions 2. Therefore, most learning occurred during
Session 1, and a 10-trial Session 2 could be used as a testing session.
2.6. Experiment 3
2.6.1. EVects of sensitization produced by repeated body shocks on the
avoidance responses
Experiments 1 and 2 demonstrated that zebraWsh showed increases in
avoidance responses without displaying increases in ITI crossings over
training sessions. Those result suggested that Wsh learned to associate the
CS of light with the US of shock and responded to the light by crossing the
barrier to avoid the shock. However, it might be possible that the increases
in avoidance responses over training sessions resulted from sensitization,
i.e., repeated shocks sensitized Wsh and caused Wsh to respond to the light.
This experiment investigated whether repeated shocks sensitized Wsh and
caused Wsh to show increases in avoidance responses over sessions. A total
of 16 zebraWsh were randomly divided into two groups (n D 8 per group):
conditioning and pseudo-conditioning groups. The conditioning group
received 30 trials of paired light and light + shock per training session as
described in the Section 2.3. The pseudo-conditioning group received the
same amount and the same order of the light and light + shock as the con-
ditioning group did, except that they received unpaired light and
light + shock. That is, the pseudo-conditioning group received lights alone
trials, and lights + shocks trials in which light and shock were presented
simultaneously and terminated simultaneously. The light, light + shock,
and ITI were 12 s each for both groups and both groups received three
semiweekly sessions.
2.6.2. Results
A two-way ANOVA with one between factor (two groups) and one
repeated measures (three sessions) on the avoidance responses, i.e., cross-
ing responses to the light, indicated a signiWcant group diVerence
[F(1, 14) D 28.1, p < 0.01] and a signiWcant group £ session interaction
[F(2, 28) D 11.5, p < 0.01]. Fig. 3 showed that while the conditioning group
displayed increases in avoidance responses over training sessions, the
pseudo-conditioning group did not display much change in avoidance
responses over training sessions. The mean percentages of ITI crossings of
the conditioning group were 36.6, 40.8, and 32.1% for the three sessions,
Fig. 3. Avoidance responses of zebraWsh trained with paired or unpaired
light and light + shock. Each point represents the mean percentage of
avoidance responses § SE for eight Wsh.
 X. Xu et al. / Neurobiology of Learning and Memory 87 (2007) 72–77
respectively. The mean percentages of ITI crossings of the pseudo-condi-
tioning group were 77.9, 89.2, and 58.3% for the three sessions, respec-
tively. The pseudo-conditioning group showed higher ITI crossings than
the conditioning group. While the conditioning group received a single
body shock following an ITI crossing, the pseudo-conditioning group
did not receive a single body shock following an ITI crossing, which may
be the reason that they had higher ITI crossings than the conditioning
group. Thus, the conditioning group learned to associate the light and
the shock and showed increases in avoidance responses, while the
pseudo-conditioning group did not show increases in avoidance
responses over sessions.
2.7. Experiment 4
2.7.1. EVects of post-training injections of MK-801 and L-NAME on
retention of avoidance conditioning
Experiment 3 showed that Wsh learned avoidance responses following
the association between the light and shock in the avoidance conditioning
paradigm. By varying the number of trials per training session and the
duration of ITI, Experiments 1 and 2 showed that most learning occurred
during a 30-trial training session. Furthermore, a 10-trial Session 2 could
be used as a testing session. Using those parameters and the 12-s ITI in the
avoidance conditioning paradigm, this experiment investigated the amnes-
tic eVects of NMDA receptor antagonist MK-801 and NO synthase inhib-
itor L-NAME in zebraWsh. The purposes of this experiment were to
determine whether NMDA receptors and NO were involved in learning
and memory of avoidance conditioning in zebraWsh and whether this
experiment produced results that would conWrm previous studies in other
vertebrates.
Previous studies in other species showed that post-training injec-
tions of L-NAME produced retention deWcits in avoidance learning, but
post-training injections of MK-801 did not (Bernabeu et al., 1995; Xu
et al., 1998). The hypothetical physiological process that is normally
completed sometime following the learning experience and that is nec-
essary for normal long-term memory is called memory consolidation
(Hebb, 1949). Retention deWcits produced by immediate post-training
pharmacological manipulations suggest an inXuence on memory con-
solidation processes. This experiment investigated whether injections of
MK-801 and L-NAME immediately following training impaired mem-
ory consolidation, resulting in retention deWcits in avoidance condition-
ing in zebraWsh. Groups of Wsh received injections of saline, 10 g of
MK-801, or 100 g of L-NAME immediately following a training
session on Experimental day 1. The 10 g of MK-801 and 100 g of
L -NAME were chosen because those doses were used in previous stud-
ies with goldWsh (Xu et al., 1998, 1999). Another group of Wsh received
no injection. On Experimental day 3, Wsh were tested without drug
treatments. While the training session consisted of 30 trials, the testing
Fig. 4. Amnestic eVects of MK-801 and L-NAME. Each bar represents the
mean percentage of avoidance responses § SE for eight Wsh during the
testing session. ¤p < 0.05, compared with the control group.
75
session consisted of 10 trials. ZebraWsh that showed ITI crossings of
30% or less during testing were retained in this experiment.
2.7.2. Results
The mean percentage of avoidance responses of no-injection Wsh dur-
ing testing was 70%, while the mean percentage of avoidance responses of
saline-injected Wsh during testing was 72.5%. Because there was no diVer-
ence in avoidance responses between Wsh that received no injection and
Wsh that received saline-injection, the two groups were combined into one
control group. A one-way ANOVA with multiple comparisons on the
avoidance responses of the three groups (n D 8 per group) during the test-
ing session showed that L-NAME signiWcantly impaired retention while
MK-801 did not (Fig. 4).
3. Discussion
Experiments 1 and 2 showed that with the CS of light,
US of shock, and ITI each being 12 s, zebraWsh learned
active avoidance responses within a training session con-
sisting of 30 trials and retained avoidance responses. Exper-
iment 3 showed that zebraWsh learned avoidance responses
as a result of pairing of the light and the shock. Using the
avoidance conditioning paradigm, Experiment 4 showed
that post-training L-NAME signiWcantly impaired reten-
tion while post-training MK-801 did not, conWrming the
previous studies in other vertebrates and thus also validat-
ing the avoidance conditioning paradigm in zebraWsh.
A comparison between the level of learning of Wsh
trained for 20 trials per session and that of Wsh trained for
30 trials per session showed that the 30-trial training ses-
sion produced faster learning (Fig. 1). A comparison
between the level of learning of Wsh trained for 30 trials per
session and that of Wsh trained for 40 trials per session
showed that 40 trials per session did not produce faster
learning (Fig. 1). Fish trained for 40 trials per session were
observed to respond less during the last 10 trials. Thus, 30
trials per session produced faster learning and more trials
seemed to produce stress, instead of learning. Our prelimi-
nary study trained some Wsh with 96 s ITI. The long ITI
slowed down learning (data not shown). A comparison
between the level of learning of Wsh trained with 30 s ITI
and that of Wsh trained with 12 s ITI showed that 12 s ITI
produced slightly faster, but not signiWcantly faster, learn-
ing (Fig. 2). Fish trained with 12 s ITI were observed not to
act diVerently from Wsh trained with the longer ITI. Thus,
30-trail training session with 12 s ITI seemed to be a good
combination to produce faster learning and shorten train-
ing time.
In active avoidance conditioning, animals learn to asso-
ciate the CS with US and respond to the CS to avoid the
US. The rate of avoidance responses serves as an indicator
of learning. In any active avoidance paradigm, high rates of
ITI crossings confound the results. With high rates of ITI
crossings, the increases in avoidance responses may not be
the result of learning to associate the CS with the US to
avoid the US, but rather the result of increased overall
crossing responses. Thus, with high rates of ITI crossings, it
is not clear whether high rates of avoidance responses actu-
ally indicate that animals have learned to associate the CS
76 X. Xu et al. / Neurobiology of Learning and Memory 87 (2007) 72–77
with the US and responded to the CS to avoid the US.
Therefore, it is important to reduce the rate of ITI crossings
in active avoidance conditioning. In our preliminary study,
ITI crossings were not followed by an electrical body shock
and ranged from 0 to 496.7% for individual Wsh during
training. In the series of experiments in this study, ITI
crossings were immediately followed by a single body shock
and rates of ITI crossing ranged from 0 to 240% for indi-
vidual Wsh. Only 5% of zebraWsh in our preliminary study
had low rates of ITI crossings (30% or less) during the last
training session. With the active avoidance paradigm that
delivered a single body shock following ITI crossings,
about 50% of zebraWsh in Experiments 1 and 2 had ITI
crossings of 30% or less during the last training session.
Thus, Experiment 4 used the ITI crossings of 30% during
testing as a cut-oV to further reduce the inXuence of ITI
crossings on the results.
Experiments 1 and 2 demonstrated that zebraWsh
showed increases in avoidance responses without display-
ing increases in ITI crossings over training sessions. These
results suggested that Wsh learned to associate the CS of
light with the US of shock and responded to the light by
crossing the barrier to avoid the shock. As about half of Wsh
still have relatively high rates of ITI crossings, it might be
possible that the increases in avoidance responses over
training sessions resulted from sensitization. That is,
repeated shocks sensitized Wsh and sensitization caused Wsh
to respond to the light. Thus, Experiment 3 investigated
whether repeated shocks sensitized Wsh and caused Wsh to
show an increase in avoidance responses over sessions.
Experiment 3 demonstrated that Wsh that received unpaired
light and light + shock did not show increases in avoidance
responses, while Wsh that received paired light and
light + shock showed increases in avoidance responses over
training sessions. Therefore, the results suggested that the
increase in avoidance responses of Wsh trained in the avoid-
ance conditioning paradigm was not due to sensitization.
Administration of NMDA receptor antagonist amino-
phosphonopentanoic acid (AP5) before training, but not
after training, impaired retention of fear conditioning in
rats (Kim et al., 1991). Administration of the NO synthase
inhibitor nitro-arginine either before or immediately fol-
lowing training was reported to impair retention in inhibi-
tory avoidance learning in rats (Bernabeu et al., 1995). Our
previous study with goldWsh showed that although MK-801
administered before training was very potent in producing
retention deWcits in avoidance conditioning, it did not
impair retention when given immediately following training
(Xu et al., 1998). In contrast, L-NAME administered before
training was not as potent as MK-801 in producing reten-
tion deWcits. In fact, 10 g of MK-801 given before training
produced a more severe retention deWcit than 100 g
of L-NAME given before training in goldWsh. However,
L-NAME produced retention deWcits when given immediately
following training in goldWsh (Xu et al., 1998). Thus, it was
proposed that NMDA receptor antagonists impaired learn-
ing or the process that is completed during training,
whereas NO synthase inhibitors impaired memory consoli-
dation or the process that is normally completed some time
following the learning experience (Xu et al., 1998). Based on
the goldWsh studies, Experiment 4 in the present study
employed 10 g of MK-801 and 100 g of L-NAME. Exper-
iment 4 showed that the percentage of avoidance responses
for control, MK-801, and L-NAME Wsh during testing were
71.3, 62.5, and 42.5%, respectively. One MK-801 Wsh had
0% avoidance response during testing and caused the mean
percentage of avoidance responses for the MK-801 group
to be lower than the control group. However, only the
diVerence between the control and L-NAME groups was
statistically signiWcant. Thus, the present results that post-
training injection of L-NAME signiWcantly impaired retention,
while post-training injection of MK-801 did not, conWrmed
the previous results with other vertebrates.
Learning and memory have been studied in many ani-
mal models. Studies from the fruit Xy, Drosophila and nem-
atode, Caenorhabditis elegans, have involved genetic
mutants, allowing for rapid identiWcation of genes involved
in learning and memory (Goodwin et al., 1997; Wen et al.,
1997). Experiments involving Xies and worms require
extrapolation from the invertebrates, to the vertebrates,
and to the human. In contrast, models of learning and
memory function in mice and rats require less extrapola-
tion. However, these organisms are only beginning to
achieve the genetic background found in the invertebrates.
The zebraWsh represents a potentially useful organism for
studying genes involved in learning and memory function
in vertebrates, because a number of genetic techniques in
zebraWsh have been developed to produce a wide variety of
genetic mutants (WesterWeld, 1994) and a number of labo-
ratories have been working to construct the genome map
for zebraWsh (Postlethwait et al., 1994). Experiments involv-
ing zebraWsh only require extrapolation from the verte-
brates to the human. The present study showed that
zebraWsh can learn and remember avoidance responses and
that the involvements of NMDA receptors and NO in
learning and memory in zebraWsh are similar to the involve-
ments of NMDA receptors and NO in learning and mem-
ory in other vertebrates. Thus, future studies with zebraWsh
mutants may identify genes involved in learning and mem-
ory in vertebrates.
Acknowledgments
This work was supported in part by GVSU Grant-in-aid
to Xiaojuan Xu. Authors gratefully acknowledge that
Dr. B. AgranoV at Molecular and Behavioral Neuroscience
Institute, University of Michigan, Ann Arbor, MI 48109,
USA, provided the automated zebraWsh avoidance set-up
for the present study.
References
AgranoV, B. W. (1980). Biochemical events mediating the formation of
short-term and long-term memory. In Y. Tsukada & B. W. AgranoV
 X. Xu et al. / Neurobiology of Learning and Memory 87 (2007) 72–77
(Eds.), Neurobiological basis of learning and memory. New York: Wiley
and Sons.
Baldwin, A. E., Sadeghian, K., Holahan, M. R., & Kelly, A. E. (2002).
Appetitive instrumental learning is impaired by inhibition of cAMP-
dependent protein kinase within the nucleus accumbens. Neurobiology
of Learning and Memory, 77, 44–62.
Bernabeu, R., de Stein, M. L., Fin, C., Izquierdo, I., & Medina, J. H. (1995).
Role of hippocampal NO in the acquisition of inhibitory avoidance
learning. NeuroReport, 6, 1498–1500.
Goodwin, S., DelVechio, M., Velinzon, K., Hogel, C., Russell, S., Tully, T.,
& Kaiser, K. (1997). Defective learning in mutants of the Drosophila
gene for a regulatory subunit of cAMP-dependent protein kinase. Jour-
nal of Neuroscience., 17, 8817–8827.
Hebb, D. O. (1949). The organization of behavior. New York: John Wiley
and Sons Inc..
Kim, J. J., Decola, J. P., Landeira-Fernandez, J., & Fanselow, M. S.
(1991). N-methyl- D-aspartate receptor antagonist APV blocks
acquisition but not expression of fear conditioning. Behavioral Neu-
roscience, 105, 126–133.
Koh, M. T., Thiele, T. E., & Bernstein, Z. L. (2002). Inhibition of protein
kinase A activity interferes with long-term, but not short-term mem-
ory of conditioned taste aversions. Behavioral Neuroscience, 116,
1070–1074.
Morris, R. G. M., Anderson, E., Lynch, G. S., & Baudry, M. (1986). Selective
impairment of learning and blockade of long-term potentiation by a N-
methyl-D-aspartate receptor antagonist, AP5. Nature, 319, 774–776.
Postlethwait, J. H., Johnson, S. L., Midson, C. N., Talbot, W. S., Gates, M.,
Ballinger, E. W., et al. (1994). A genetic linkage map for the zebraWsh.
Science, 264, 699–703.
Schafe, G. E., Nadel, N. Y., Sullivan, G. M., Harris, A., & LeDoux, J. E.
(1999). Memory consolidation for contextual and auditory fear condi-
77
tioning is dependent on protein synthesis, PKA, and MAP kinase.
Learning & Memory, 6, 97–110.
Shapiro, M., & Caramanos, Z. (1990). NMDA antagonist MK-801 impairs
acquisition but not performance of spatial working and reference
memory. Psychobiology, 18, 231–243.
Wen, J. Y., Kumar, N., Morrison, G., Rambaldini, G., Runciman, S., Rous-
seau, J., et al. (1997). Mutations that prevent associative learning C. ele-
gans. Behavioral Neuroscience, 111, 354–368.
WesterWeld, M. (1994). The ZebraWsh Book: A guide to the laboratory use of
the zebraWsh (Brachydanio rerio). Eugene, OR: University of Oregon
Press.
Xu, X., Bazner, J., Qi, M., Johnson, E., & FreidhoV, R. (2003). The role of
telencephalic NMDA receptors in avoidance learning in goldWsh
(Carassius auratus). Behavioral Neuroscience, 117(3), 548–554.
Xu, X., Boshoven, W., & Gunn, D. (1999). NMDA receptor antagonist
MK-801 and nitric oxide synthase inhibitor L-NAME did not produce
state-dependent learning in goldWsh. Psychobiology, 27(3), 426–431.
Xu, X., Boshoven, W., Lombardo, B., & Spranger, F. (1998). Comparison
of the amnestic eVects of NMDA receptor antagonist MK-801 and
nitric oxide synthase inhibitors: L-NAME and L-NOARG in goldWsh.
Behavioral Neuroscience, 112(4), 892–899.
Xu, X., & Davis, R. E. (1992). NMDA receptor antagonist MK-801
impairs learning but not memory Wxation or expression of classical fear
conditioning in goldWsh. Behavioral Neuroscience, 106(2), 307–314.
Xu, X., Klinger, P., & Davis, R. (1995). Comparative anterograde amnestic
and anticonvulsant eVects of two types of NMDA receptor antago-
nists: MK-801 and HA-966. Psychopharmacology, 117, 333–339.
Xu, X., Russell, T., Bazner, J., & Hamilton, J. (2001). NMDA receptor
antagonist AP5 and nitric oxide synthase inhibitor 7-NI aVect diVer-
ent phases of learning and memory in goldWsh. Brain Research, 889,
274–277.