Eyeblink conditioning in 12-day-old rats using pontine stimulation as the conditioned stimulus

Abstract

A fundamental issue in developmental science is whether ontogenetic changes in memory are caused by the development of cellular plasticity mechanisms within the brain's memory systems or maturation of sensory inputs to the memory systems. Here, we provide evidence that the development of eyeblink conditioning, a form of associative learning that depends on the cerebellum, is driven by the development of sensory inputs rather than the development of neuronal plasticity mechanisms. We find that rats as young as 12 days old show associative eyeblink conditioning when pontine stimulation is used in place of an external (e.g., a tone) conditioned stimulus. Eyeblink-conditioned responses established with pontine stimulation in 12-day-old rats were reversibly abolished by an infusion of muscimol into the cerebellar interpositus nucleus. The findings suggest that cerebellar neurons are capable of supporting associative learning-specific plasticity in vivo in very immature animals if given sufficient afferent stimulation.

It seems reasonable to assume that the development of memory is caused by the maturation of synaptic plasticity mechanisms within the memory systems of the brain. For example, the development of declarative memory could be related to the development of long-term potentiation in the medial temporal lobe, and conditioning of discrete movements could be related to the development of cerebellar long-term depression. On the other hand, it is possible that the cerebellum and hippocampus are capable of establishing learning-related synaptic plasticity in young animals but simply do not receive sufficient sensory input during learning. We evaluated these possibilities by examining the effects of electrically stimulating a sensory input pathway to the cerebellum on eyeblink conditioning in rats that are too immature to show conditioning with external sensory stimuli.

Eyeblink conditioning, a type of associative learning, typically involves the presentation of a conditioned stimulus (CS) that does not produce a blink reflex before training (e.g., a tone or light) followed by an unconditioned stimulus (US) that reliably elicits the blink reflex. Repeated presentations of the CS and US result in the acquisition of a conditioned response (CR) that precedes the onset of the US. Eyeblink conditioning emerges ontogenetically between postnatal days 17 and 24 in rats (1). Developmental changes in human eyeblink conditioning have also been documented (2).

The cerebellar hemisphere that is ipsilateral to the conditioned eye is essential for acquisition and retention of eyeblink conditioning in adult and infant animals (3, 4). Neurons within the pontine nuclei (PN) are part of the mossy fiber pathway that sends CS information to the cerebellum (5–7). Cerebellar neurons are thought to develop learning-specific changes in synaptic efficacy to CS inputs, which underlies memory for the CS–US association (8). Using either cerebellar slices or cultured neurons from rodents, many sources of cerebellar synaptic plasticity have been characterized that might play critical roles during in vivo associative learning (9–12). Some of the in vitro studies of cerebellar plasticity have used infant cerebellar cells to optimize whole-cell or sharp electrode recordings. The findings of the in vitro studies that used infant tissue suggest that cerebellar neurons might be fully capable of establishing learning-specific plasticity in vivo during the first 2 postnatal weeks (12).

In vivo neurophysiological analyses of the ontogeny of eyeblink conditioning suggest that the PN receive weaker afferent input from sensory nuclei in younger rats (nonlearners), which results in weaker mossy fiber input to the cerebellum (13). Stimulation of the pontine mossy fibers may be able to overcome the developmental limitations in eyeblink conditioning in younger rats by providing sufficient mossy fiber input to the cerebellum (14). The current study pushed the limits of this hypothesis by examining whether PN stimulation could be an effective CS in 12-day-old rats. Rats at this age have closed eyes and ear canals and therefore cannot learn responses to visual and auditory stimuli. The cerebellum is also immature at this age, with ongoing neurogenesis in the external molecular layer and neuronal migration into the granule cell layer (15). However, Purkinje cells and deep nucleus neurons have synaptic inputs that are sufficient for induction of long-term plasticity in vitro during the first 2–3 postnatal weeks (12). The main question in the current study was whether cerebellar learning could be induced in vivo in 12-day-old rats by direct stimulation of the mossy fiber pathway as a CS paired with a peripheral US.

Results

In the first experiment, rat pups were given eyelid conditioning by using a 300-ms train of 100-μs current pulses at 200 Hz within the PN as the CS, which was paired with a peripheral periorbital shock (10.0 ms, 4.0 mA) US on postnatal days 12 and 13. The pups were given three conditioning sessions per day with 100 training trials in each session. An age-matched control group was given unpaired presentations of PN stimulation and the US during each conditioning session. The control group was necessary to show that blink responses in the group given paired training were caused by the formation of a CS–US association and not by nonassociative learning. The eyelids are normally fused at this age in rats. It was, therefore, necessary to separate the eyelids surgically on postnatal day 11 to enable recording eyelid electromyography (EMG) activity during lid closure. Rat pups that were given paired presentations of PN stimulation, and the US showed an increase in eyelid CRs across training and significantly more eyelid responses than in the unpaired group during sessions 3–6 [Fig. 1A; F(5, 50) = 4.60, P < 0.003]. The increase in the percentage of eyelid CRs was, therefore, caused by acquisition of associative learning and cannot be attributed to sensitization. CRs in the group given paired training were initiated during the CS and the maximum amplitude of the responses occurred before the onset of the US, as seen in adult rat eyelid EMG activity (Fig. 1B). Histological analysis indicated that the electrodes were consistently placed within or just dorsal to the basilar PN (Fig. 1C). The findings of the first experiment indicate that associative eyelid conditioning can be established in rat pups as early as postnatal day 12, even though that is well before the age at which eyelid conditioning is observed with an external CS such as a light or tone (postnatal day 20) (16).

Fig. 1.

Associative eyelid conditioning in rats trained on postnatal days 12 and 13. (A) Mean CR percentage across six 100-trial training sessions for rat pups given paired or unpaired presentations of pontine stimulation as a CS and a peripheral US. Error bars indicate SEM. (B) Eyelid EMG activity for rat pups that were given paired (Upper) or unpaired (Lower) conditioning during test trials with only the CS from the last training session. Dashed lines indicate the onset time of the CS. (C) A digital image of a representative electrode placement (arrow) in the PN. (Magnification: ×2.5.)

PN stimulation was clearly sufficient for establishing eyelid conditioning in relatively young rat pups. It was possible, however, that PN stimulation supported learning through a noncerebellar mechanism. That is, the eyelid CRs could have been generated through a different neural circuit than the neural circuit that is needed for learning in older rats. The second experiment examined the effect of inactivating the cerebellum ipsilateral to the conditioned eye on the production of eyelid CRs in rat pups given paired presentations of PN stimulation as the CS and a peripheral US. Previous studies found that muscimol infusions into the ipsilateral cerebellar interpositus nucleus abolished eyelid CRs in adult and infant rats (17), a finding first demonstrated in adult rabbits (18). The rat pups were given five sessions in which the PN stimulation CS was paired with the US on postnatal days 12 and 13, as in the first experiment. They were then given an infusion of muscimol (0.5 μl, 5.0 nmol) into the interpositus nucleus ipsilateral to the conditioned eye. A recovery session was administered the next day to confirm that the infusion did not produce any lasting effects on the production of CRs. As in the first experiment, the pups given paired training showed an increase in eyelid CRs across training sessions [Fig. 2A; F(6, 18) = 8.05, P < 0.001]. Muscimol infusion into the cerebellum produced a substantial decrease in the incidence of eyelid CRs (Fig. 2 A and B; P < 0.05). Complete recovery of eyelid CRs was seen after the muscimol infusion during the last conditioning session (Fig. 2 A and B; P < 0.05). Analysis of fluorescently labeled muscimol indicated that it spread throughout the interpositus nucleus and into the overlying cerebellar cortex in the ipsilateral hemisphere (Fig. 2C). The findings of the second experiment indicate that the cerebellar hemisphere that is ipsilateral to the conditioned eye is necessary for the production of CRs that are established when using PN stimulation as a CS.

Fig. 2.

Eyelid conditioning was severely impaired by cerebellar inactivation. (A) Mean CR percentage across seven 100-trial training sessions for rat pups given paired presentations of pontine stimulation as a CS and an external US on postnatal days 12–14. Muscimol (0.5 μl, 5.0 nmol) was infused into the cerebellar interpositus nucleus that was ipsilateral to the conditioned eye 30 min before session 6 (arrow). Error bars indicate SEM. (B) Eyelid EMG activity from a rat on training trials with the CS and US for the session before (Top, blue), during (Middle, red), and after (Bottom, green) cerebellar inactivation. Dashed lines indicate the onset times of the CS and US. (C) A digital image showing fluorescent muscimol (0.5 μl, 5.0 nmol) within the cerebellar hemisphere. The smallest and largest spread of fluorescent muscimol are depicted by the green and blue dotted lines, respectively. The area of inactivation included the interpositus nucleus (IN) and lobule HVI of the overlying cortex. The dentate nucleus (DN) was not affected. (Magnification: ×2.5.)

Discussion

Stimulation of the basilar PN as a CS was sufficient for eyeblink conditioning in 12-day-old rats. There was an increase in the percentage of eyeblink CRs in the pups given paired presentations of the pontine stimulation CS and a periorbital shock US across training sessions, but no increase in eyeblink CRs in the control group given unpaired training. The increase in CR percentage across sessions in the group given paired training can therefore be attributed to acquisition of associative learning. Infusion of muscimol into the cerebellar interpositus nucleus reversibly abolished eyeblink CRs in pups given paired training, suggesting that the expression of associative learning established with pontine stimulation depended on cerebellar activity. These findings are consistent with previous studies using adult and juvenile animals (5, 14).

Various synaptic plasticity mechanisms have been demonstrated in vitro by using cerebellar preparations from the first 3 postnatal weeks (9–12). However, in vivo cerebellar learning with external stimuli is not robust until postnatal day 24 in rats (1). This observation raises the possibility that some of the in vitro models might be characterizing plasticity mechanisms that do not support learning in vivo. The associative eyeblink conditioning observed in rats at postnatal day 12 with pontine stimulation suggests that the findings of in vitro studies that use infant tissue as young as postnatal day 12 to examine cerebellar plasticity are directly applicable to in vivo learning.

The demonstration of associative eyeblink conditioning in 12-day-old rats with pontine stimulation has important implications for elucidating the mechanisms underlying the ontogeny of cerebellar learning. At this age, the rat cerebellum is relatively immature (15, 19, 20). Nevertheless, immature cerebellar neurons are capable of synaptic plasticity and increased intrinsic excitability in vitro (9–12). The current findings suggest that immature cerebellar neurons are also capable of supporting associative learning in vivo if given sufficient afferent stimulation. Cerebellar neurons in developing animals are therefore ready to learn before they have something to learn about.

The sufficiency of pontine stimulation as a CS in 12-day-old rats further suggests that the mossy fiber projection to the cerebellum could support associative learning with an externally presented CS (e.g., a tone) in immature animals if sensory input to the PN was stronger. Our current hypothesis is that the ontogenetic emergence of eyeblink conditioning is driven primarily by the development of sensory input to the PN. Pontine stimulation in the 12-day-old rat pups was, in essence, fulfilling the role of sensory input to the PN that occurs during eyeblink conditioning in more mature animals.

Other brain systems involved in learning and memory also show neuronal plasticity in vitro early in postnatal development (21). The neuronal plasticity mechanisms underlying learning may be the same as or derived from mechanisms underlying early activity-dependent connectivity in the nervous system (22). Neuronal plasticity, therefore, plays a more general role in establishing connectivity early in development but then plays a role in learning-specific modifications of that connectivity as sensory and response systems mature. An example of early emergence of neuronal plasticity relative to the emergence of learning can be seen in the hippocampus, where long-term potentiation can be established in slices from animals that are too young to exhibit spatial memory (21, 23).

The findings of this study and previous studies (24) suggest that the ontogeny of eyeblink conditioning depends on the development of sensory input to the cerebellum. It is probably not the case, however, that the ontogeny of all other forms of learning can be related solely to the development of sensory inputs to the learning system(s). In fear conditioning, for example, sensation precedes conditioning and different sensory systems become functional at different ages (25), but the ontogeny of fear conditioning also depends on the differential development of response systems such that conditioned freezing precedes heart-rate conditioning, which precedes potentiated startle (26). The ontogenetic emergence of fear conditioning is thus caused by interactions between the development of sensory systems and response systems (26). Systematic neurobiological analyses are necessary to determine the specific developmental mechanisms underlying the ontogeny of fear conditioning and other types of learning.

Methods

Subjects.

The subjects were 16 Long-Evans rat pups derived from 12 litters. A subset of the pups was implanted with a cannula in the cerebellum (n = 4). The pups were housed with their dams in the animal colony in Spence Laboratories of Psychology at the University of Iowa. The rats were maintained on 12/12-hr light/dark cycles, with light onset at 7 a.m. Training sessions occurred between 7 a.m. and 7 p.m. Eyelid opening occurred on postnatal day 11 to enable recording eyelid EMG activity during eyelid movement. The pups were lightly anesthetized with isoflurane (1–3%). The eyelids were then opened by gently cutting through the point of eyelid fusion with a scalpel blade.

Surgery.

The detailed surgical methods have been described (13, 14). Isoflurane (1–3%) was used for anesthesia during surgery. Differential EMG electrodes were implanted in the left upper eyelid and a ground electrode was secured beneath the skull surface. A bipolar stimulating electrode was implanted into or just dorsal to the right basilar PN (14). The stereotaxic coordinates for the PN taken from lambda were +1.0 anterior, −1.0 medial-lateral, and −8.2 dorsal-ventral. The rat pups that were given muscimol infusions into the cerebellum had a 25-gauge guide cannula implanted to the brain surface above the left anterior interpositus nucleus. A 32-gauge stylet was inserted into the guide cannula. The stereotaxic coordinates for the cannula were also taken from lambda: −2.2 posterior, +2.0 medial-lateral, and −4.7 dorsal-ventral. A bipolar stimulating electrode used for delivering the US was implanted subdermally, immediately caudal to the left eye (13, 14).

Conditioning Apparatus.

The conditioning apparatus has been described in detail (13, 14, 18). The rat pups were conditioned within an operant chamber that was contained within a sound-attenuation chamber. Lightweight cables with connectors for the EMG, US, and CS electrodes were attached to a commutator. Computer software (JSA Designs) controlled the delivery of stimuli and the recording of eyelid EMG activity. EMG activity was recorded differentially, filtered (500–5,000 Hz), amplified (×2,000), and integrated (time constant = 20 ms). Pontine stimulation was triggered through a programmable stimulator, (Master 8; A.M.P.I.), which controlled signal input to a stimulus isolator (model number 365A; World Precision Instruments), which delivered the electrical stimulation.

Paired Training.

Pups in the conditioning group were given six paired training sessions, three sessions per day. The paired training sessions consisted of 100 trials each with 90 trials of the stimulation CS paired with the shock US (10 ms, 4.0 mA) and 10 stimulation CS-alone test trials, occurring on every 10th trial. The CS-alone trials were included to assess behavioral responses (integrated EMG activity) without the unconditioned reflex. The interstimulus interval (ITI) for paired trials was 290 ms. Trials were separated by an intertrial interval that averaged 30 s. Behavioral data were examined from computer records of EMG responses. CRs were defined as responses that crossed a threshold of 0.4 V above the baseline activity during the CS period, but at least 80 ms after CS onset, to avoid contamination of the CR measures by the startle (alpha) response (13, 14, 18).

Unpaired Training.

Pups in the unpaired control group were given six 200-trial sessions of explicitly unpaired presentations of the CS and US. The same time durations for the CS and US were used as in the paired procedure. The ITI was set to average 15 s to match the total time spent in the conditioning chamber and the temporal distribution of CS and US presentations with the paired groups. The method for defining CRs was the same as used in the paired procedure.

Pontine Stimulation.

Electrical stimulation of the basilar PN functioned as the CS, which was administered in a 200-Hz train of 0.1-ms biphasic pulses for 300 ms. The stimulation threshold for the CS was found before training by setting the stimulating current to 50 μA, and either increasing or decreasing the current in 5-μA increments, until a slight movement was detected (14). Observable movements included, but were not limited to, eye blinks, eyeball movements, and head movements. The level of stimulation during training was set to 10 μA below the threshold intensity. Pontine stimulation did not elicit alpha responses.

Muscimol Infusion.

Before muscimol infusions, the stylet was removed from the guide cannula and replaced with a 32-gauge infusion cannula. Muscimol (0.5 μl, 5 nmol in saline, pH 7.4) was infused at a rate of 30 μl/hr. The test session administered after muscimol infusion was identical to the other paired sessions.

Histology.

After training was completed, the rat pups were killed with a lethal injection of sodium pentobarbital (90 mg/kg) and transcardially perfused with ≈100 ml of physiological saline, followed by ≈300 ml of 3.0% formalin. The rat pups that were implanted with cerebellar cannula were given a 0.5-μl infusion of fluorescently labeled muscimol (BODIPY TMR-X Muscimol; Molecular Probes) 30 min before being killed. The brains were postfixed in 30.0% sucrose in formalin for at least 2 days before sectioning. The brains were sectioned at 50 μm with a sliding microtome. The location of fluorescently labeled tissue was assessed by using a microscope equipped with a yellow-orange florescent (572-nm wavelength) filter. All other sections were stained with thionin. The cannula and stimulating electrode placements were determined by examining multiple sections.