Theta synchronization between medial prefrontal cortex and cerebellum is associated with adaptive performance of associative learning behavior

Abstract

Associative learning is thought to require coordinated activities among distributed brain regions. For example, to direct behavior appropriately, the medial prefrontal cortex (mPFC) must encode and maintain sensory information and then interact with the cerebellum during trace eyeblink conditioning (TEBC), a commonly-used associative learning model. However, the mechanisms by which these two distant areas interact remain elusive. By simultaneously recording local field potential (LFP) signals from the mPFC and the cerebellum in guinea pigs undergoing TEBC, we found that theta-frequency (5.0-12.0 Hz) oscillations in the mPFC and the cerebellum became strongly synchronized following presentation of auditory conditioned stimulus. Intriguingly, the conditioned eyeblink response (CR) with adaptive timing occurred preferentially in the trials where mPFC-cerebellum theta coherence was stronger. Moreover, both the mPFC-cerebellum theta coherence and the adaptive CR performance were impaired after the disruption of endogenous orexins in the cerebellum. Finally, association of the mPFC -cerebellum theta coherence with adaptive CR performance was time-limited occurring in the early stage of associative learning. These findings suggest that the mPFC and the cerebellum may act together to contribute to the adaptive performance of associative learning behavior by means of theta synchronization.

Figure 1. Classification of trace CRs into adaptive and non-adaptive responses.

Individual examples of eyelid movement stack plots showing adaptive and non-adaptive CRs. The first trial for each 50-trial training session is at the bottom of the stack plot and each eyeblink trace represents one training trial. Left panel: A representative animal consistently showed adaptive CRs with peak latency close to the US onset. As shown, the adaptive CR persisted until the UR was exhibited. Right panel: A representative animal emitted non- adaptive CRs with an earlier peak. In particular, these responses returned to baseline before the 200-ms pre-US period.

Figure 2. Acquisition of trace CRs in guinea pigs.

(A) Incidence, (B) Magnitude and (C) Peak latency of trace CRs measured from the adaptive learners (red circles), non-adaptive learners (blue triangles), and unpaired animals (open squares), respectively. Data are given as the mean ± SEM. Symbols are the same for (A–C).

Figure 3. Theta-band oscillations (5.0–12.0 Hz) in mPFC and cerebellum.

(A) Schematic diagram of simultaneous local field potential (LFP) recording in the mPFC and the cerebellum during TEBC. (B) In a representative CS-US trial, the LFP signals are recorded from the mPFC (red trace) and the cerebellum (blue trace) during 850-ms periods before and after CS onset, respectively. (C,D) Fast Fourier transform (FFT) runs on the mPFC and cerebellar LFP signals during the 850-ms periods before and after CS onset, as shown in (B).

Figure 4. Coherent theta-band (5.0–12.0 Hz) oscillations between mPFC and cerebellum.

Across a conditioning session, theta-band oscillations consistently occur in 850-ms periods before (A) and after (B) the CS onset, whereas theta-band activities are more evident in the latter period. (C,D) In the theta-frequency (5.0–12.0 Hz) range, the LFP signals recorded from the mPFC are highly coherent with the LFP signals from the cerebellum. Green-shaded areas indicate 95% confidence intervals.

Figure 5. mPFC-cerebellum theta-band synchrony remains stable across training, but is stronger during the performance of adaptive CRs.

(A) Stronger mPFC-cerebellum theta-band (5.0–12.0 Hz) coherence was consistently observed in the adaptive learners (n = 11, red circles) compared with the non-adaptive learners (n = 12, blue triangles) and unpaired animals (n = 11, open squares). Nevertheless, mPFC- cerebellum coherence was comparable among three groups of animals in the delta-band (B, 0.5–4.5 Hz) and beta-band (C, 12.5–30.0 Hz). Data are given as the mean ± SEM. Symbols are the same for (A–C).

Figure 6. Theta-band phase coherence in the early- and late-learning stages.

(A) Representative histogram of theta-band phase differences. Instantaneous theta phase of two LFP signals were subtracted from each other, and differences in theta phase between the mPFC-cerebellum were plotted as a histogram in an adaptive learner, a non-adaptive learner, and an unpaired animal. (B) Width of theta-band phase difference histogram at half the peak height averaged across animals for mPFC-cerebellum in the adaptive learners (n = 11, red circles), non-adaptive learners (n = 12, blue triangles) and unpaired animals (n = 11, open squares) in the early learning stage (ELS, days 2–4) and the late learning stage (LLS, days 8–10). Data are given as the mean ± SEM.

Figure 7. Theta-band power correlation in the early- and late-learning stages.

(A) Representative examples of theta power correlation scatter plots for mPFC-cerebellum in an adaptive learner, a non-adaptive learner and an unpaired learner from a conditioning training session. (B) Averages of linear correlation coefficients of theta-band power across animals in the adaptive learners (n = 11, red circles), non-adaptive learners (n = 12, blue triangles) and unpaired animals (n = 11, open squares) in the early learning stage (ELS, days 2–4) and late learning stage (LLS, days 8–10). Data are given as the mean ± SEM.

Figure 8. mPFC-cerebellum theta-band coherence was higher when adaptive CR occurred.

Representative raw LFP signals simultaneously recorded from the mPFC (black traces) and the cerebellum (gray traces) during performance of an adaptive CR (A) and a non-adaptive CR (B), respectively. Horizontal scale bar indicates 0.5 sec. Vertical scale bar indicates 2 mV for the eyelid movements and 50 μV for the LFP signals.

Figure 9. Time-limited association of mPFC-cerebellum theta coherence with adaptive CR performance.

(A) In the early learning stage (ELS, days 2–4), mPFC-cerebellum theta-band (5.0–12.0 Hz) coherence was specifically higher in the trials where adaptive trace CRs occurred. By contrast, no such differences were found in the delta- and beta-frequency bands. (B)In the late learning stage (LLS, days 8–10), no differences were observed in either frequency band. Data are given as the mean ± SEM. Symbols are the same for (A,B).

Figure 10. Relative theta power in neither the mPFC nor the cerebellum was associated with adaptive CR performance.

(A) In both the ELS (days 2–4) and the LLS (days 8–10), there were no obvious differences in prefrontal relative theta power when adaptive and non-adaptive CRs occurred. (B) Likewise, in both the ELS and the LLS, there were no apparent differences in cerebellar theta power when the adaptive and non-adaptive CRs were observed. Data are given as the mean ± SEM. Symbols are the same for (A,B).

Figure 11. Blockade of orexinergic receptor-1 in the cerebellum disrupted both mPFC- cerebellum theta-band coherence and adaptive CR performance.

(A,B): Injections of SB-334867 into the intermediate cerebellum decreased mPFC-cerebellum coherence in theta-band (5.0–12.0 Hz) instead of in the delta-band (0.5–4.5 Hz). (C) Relative theta power in the cerebellum was not significantly attenuated by the SB-334867 injections. (D) Compared with DMSO (n = 7, red circles), the SB-334867 injections (n = 7, blue triangles) decreased the latency to CR peak in the early learning stage (ELS, days 2–4). (E,F) The SB-334867 injections seemed not to reduce the incidence and magnitude of trace CRs. Data are shown as mean ± SEM. Symbols are the same for (A–F). The averaged effects of SB-334867 (black traces) and DMSO (gray traces) injections on the CR performance on days 4 and 8 are illustrated, respectively.