Limits...
Trace Eyeblink Conditioning in Mice Is Dependent upon the Dorsal Medial Prefrontal Cortex, Cerebellum, and Amygdala: Behavioral Characterization and Functional Circuitry(1,2,3).

Siegel JJ, Taylor W, Gray R, Kalmbach B, Zemelman BV, Desai NS, Johnston D, Chitwood RA - eNeuro (2015)

Bottom Line: To identify the circuitry involved, we made restricted lesions of the medial prefrontal cortex (mPFC) and found that learning was prevented.Anatomical data from these critical regions showed that mPFC and amygdala both project to the rostral basilar pons and overlap with eyelid-associated pontocerebellar neurons.The data further reveal a specific role for the amygdala as providing a conditioned stimulus-associated input to the cerebellum.

View Article: PubMed Central - HTML - PubMed

Affiliation: Center for Learning and Memory, University of Texas at Austin , Austin, Texas 78712.

ABSTRACT
Trace eyeblink conditioning is useful for studying the interaction of multiple brain areas in learning and memory. The goal of the current work was to determine whether trace eyeblink conditioning could be established in a mouse model in the absence of elicited startle responses and the brain circuitry that supports this learning. We show here that mice can acquire trace conditioned responses (tCRs) devoid of startle while head-restrained and permitted to freely run on a wheel. Most mice (75%) could learn with a trace interval of 250 ms. Because tCRs were not contaminated with startle-associated components, we were able to document the development and timing of tCRs in mice, as well as their long-term retention (at 7 and 14 d) and flexible expression (extinction and reacquisition). To identify the circuitry involved, we made restricted lesions of the medial prefrontal cortex (mPFC) and found that learning was prevented. Furthermore, inactivation of the cerebellum with muscimol completely abolished tCRs, demonstrating that learned responses were driven by the cerebellum. Finally, inactivation of the mPFC and amygdala in trained animals nearly abolished tCRs. Anatomical data from these critical regions showed that mPFC and amygdala both project to the rostral basilar pons and overlap with eyelid-associated pontocerebellar neurons. The data provide the first report of trace eyeblink conditioning in mice in which tCRs were driven by the cerebellum and required a localized region of mPFC for acquisition. The data further reveal a specific role for the amygdala as providing a conditioned stimulus-associated input to the cerebellum.

No MeSH data available.


Related in: MedlinePlus

Learning rates and CR latency data for mice trained at trace 50–350 or trace 50–450. A, Only 4 of the 8 mice were able to learn trace 50–350 (left graph), and only 3 of 8 mice were able to learn trace 50–450 (right graph) within 12 training sessions (black lines indicate learners and gray lines indicate nonlearners; red markers show median CR rate for learners and light red markers show the median CR rate for all mice). B, Left graphs show probe trials from two example mice from each of the three groups trained at different trace intervals. Numbers on x-axes give the median latency to CR onset and peak for each example mouse. Note that the latencies to onset are similar across these examples, while the latencies to peak shift with the longer training intervals. Right graph shows group data for CR latencies and amplitude for the 3 training groups beginning with initial expression until the last training session (black lines show median latencies, shaded gray region indicates interquartile range). Mice showed CR onsets that did not vary across the different trace intervals on the final training days (trace 50–250 vs 50–350, Wilcoxon rank: U = 26, n = 4 and 12, p = 0.42; trace 50–350 vs 50–450, U = 5, n = 3 and 4, p = 0.69). Latencies to CR criterion and peak were different between trace 50–250 and 50–350 (Wilcoxon rank: U = 45, n = 4 and 12, p < 0.001 and U = 44, p < 0.001, respectively), with the peaks of CRs appropriately timed for the training intervals used. Mice trained to trace 50–450 showed early latencies relative to the training interval, and were not different from mice trained to 50–350 for any parameter (Wilcoxon rank, Onset: U = 5, n = 3 and 4, p = 0.69; Criterion: U = 8, p = 0.31; Peak: U = 6, p = 0.57). CR amplitudes were similar between the three training groups (rightmost graph, black lines show median for each training day, gray shaded area shows interquartile range, dotted line indicates median amplitude of trace 50–250 mice on the first day of expression as reference).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4596016&req=5

Figure 3: Learning rates and CR latency data for mice trained at trace 50–350 or trace 50–450. A, Only 4 of the 8 mice were able to learn trace 50–350 (left graph), and only 3 of 8 mice were able to learn trace 50–450 (right graph) within 12 training sessions (black lines indicate learners and gray lines indicate nonlearners; red markers show median CR rate for learners and light red markers show the median CR rate for all mice). B, Left graphs show probe trials from two example mice from each of the three groups trained at different trace intervals. Numbers on x-axes give the median latency to CR onset and peak for each example mouse. Note that the latencies to onset are similar across these examples, while the latencies to peak shift with the longer training intervals. Right graph shows group data for CR latencies and amplitude for the 3 training groups beginning with initial expression until the last training session (black lines show median latencies, shaded gray region indicates interquartile range). Mice showed CR onsets that did not vary across the different trace intervals on the final training days (trace 50–250 vs 50–350, Wilcoxon rank: U = 26, n = 4 and 12, p = 0.42; trace 50–350 vs 50–450, U = 5, n = 3 and 4, p = 0.69). Latencies to CR criterion and peak were different between trace 50–250 and 50–350 (Wilcoxon rank: U = 45, n = 4 and 12, p < 0.001 and U = 44, p < 0.001, respectively), with the peaks of CRs appropriately timed for the training intervals used. Mice trained to trace 50–450 showed early latencies relative to the training interval, and were not different from mice trained to 50–350 for any parameter (Wilcoxon rank, Onset: U = 5, n = 3 and 4, p = 0.69; Criterion: U = 8, p = 0.31; Peak: U = 6, p = 0.57). CR amplitudes were similar between the three training groups (rightmost graph, black lines show median for each training day, gray shaded area shows interquartile range, dotted line indicates median amplitude of trace 50–250 mice on the first day of expression as reference).

Mentions: To further investigate the timing of CRs, two sets of naïve mice were trained to different trace intervals (trace 50–350: n = 8, and trace 50–450: n = 8). Fewer mice were able to learn the longer intervals within 12 training sessions (4/8 mice for trace 50–350 and 3/8 mice for trace 50–450; Fig. 3A), suggesting that an interval even 100 ms longer was more difficult for the mice to learn. Of the mice that were able to learn trace 50–350, the latency to the peak amplitude of responses was significantly later than that observed for mice trained with a trace 50–250 protocol (trace 50–350: median = 407 ms; IQR = 357–456 ms; Mann–Whitney–Wilcoxon Rank test, U = 44, n = 4 and 12, p < 0.001d; Fig. 3B, top and middle, “peak”), and was also well timed to the air puff (H0: mean = 400 ms, t = 0.32, df = 3, p = 0.84e). However, the latency to the onset of CRs (i.e., when the eyelids initially began to close during a given trial; Fig. 1C) was not different between mice that learned trace 50–250 versus those that learned trace 50–350 (tr50–250: median = 105 ms, IQR = 99–112 ms; tr50–350: median = 107 ms, IQR = 98–129 ms; U = 26, n = 4 and 12, p = 0.42f; Fig. 3B, top and middle, “onset”). The results are consistent with previous reports of mice trained in delay eyeblink conditioning using different interstimulus intervals, and demonstrate that mice initiated learned responses 100–125 ms after CS onset regardless of the training interval used, whereas timing the peaks of CRs to coincide with the air puff (Chettih et al., 2011; Heiney et al., 2014). Furthermore, a significant shift in the latency at which CR criterion was met (FEC > 0.1; Fig. 1C) between trace 50–250 and 50–350 suggests that the difference in the timing of CR peaks was modulated by the initial velocity of behavioral responses (tr50–250: median = 164 ms, IQR = 148–181 ms; tr50–350 median = 234 ms, IQR = 192–266 ms; U = 45, n = 4 and 12, p < 0.001g; Fig. 3B, top and middle, “criterion”). Interestingly, the latencies to onset, criterion and peak amplitude of CRs for mice that learned the trace 50–450 task protocol were not different to those observed for trace 50–350 (tr50–450 Onset: median = 100 ms, IQR = 98–114 ms, U = 5, n = 3 and 4, p = 0.69h; Criterion: median = 225 ms, IQR = 223–285 ms, U = 8, p = 0.31i; Peak: median = 410, IQR = 403–411 ms, U = 6, p = 0.57j; Fig. 3B, middle and bottom). The data suggest that the mice able to learn trace 50–450 did so by expressing early CRs that were not well timed to the air puff, and that this interval may be too long for naïve mice to express well timed responses. Similar to trace 50–250, mice trained at longer intervals also initially showed low amplitude CRs at initial expression that gradually increased over several sessions (Fig. 3B, right graphs).


Trace Eyeblink Conditioning in Mice Is Dependent upon the Dorsal Medial Prefrontal Cortex, Cerebellum, and Amygdala: Behavioral Characterization and Functional Circuitry(1,2,3).

Siegel JJ, Taylor W, Gray R, Kalmbach B, Zemelman BV, Desai NS, Johnston D, Chitwood RA - eNeuro (2015)

Learning rates and CR latency data for mice trained at trace 50–350 or trace 50–450. A, Only 4 of the 8 mice were able to learn trace 50–350 (left graph), and only 3 of 8 mice were able to learn trace 50–450 (right graph) within 12 training sessions (black lines indicate learners and gray lines indicate nonlearners; red markers show median CR rate for learners and light red markers show the median CR rate for all mice). B, Left graphs show probe trials from two example mice from each of the three groups trained at different trace intervals. Numbers on x-axes give the median latency to CR onset and peak for each example mouse. Note that the latencies to onset are similar across these examples, while the latencies to peak shift with the longer training intervals. Right graph shows group data for CR latencies and amplitude for the 3 training groups beginning with initial expression until the last training session (black lines show median latencies, shaded gray region indicates interquartile range). Mice showed CR onsets that did not vary across the different trace intervals on the final training days (trace 50–250 vs 50–350, Wilcoxon rank: U = 26, n = 4 and 12, p = 0.42; trace 50–350 vs 50–450, U = 5, n = 3 and 4, p = 0.69). Latencies to CR criterion and peak were different between trace 50–250 and 50–350 (Wilcoxon rank: U = 45, n = 4 and 12, p < 0.001 and U = 44, p < 0.001, respectively), with the peaks of CRs appropriately timed for the training intervals used. Mice trained to trace 50–450 showed early latencies relative to the training interval, and were not different from mice trained to 50–350 for any parameter (Wilcoxon rank, Onset: U = 5, n = 3 and 4, p = 0.69; Criterion: U = 8, p = 0.31; Peak: U = 6, p = 0.57). CR amplitudes were similar between the three training groups (rightmost graph, black lines show median for each training day, gray shaded area shows interquartile range, dotted line indicates median amplitude of trace 50–250 mice on the first day of expression as reference).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4596016&req=5

Figure 3: Learning rates and CR latency data for mice trained at trace 50–350 or trace 50–450. A, Only 4 of the 8 mice were able to learn trace 50–350 (left graph), and only 3 of 8 mice were able to learn trace 50–450 (right graph) within 12 training sessions (black lines indicate learners and gray lines indicate nonlearners; red markers show median CR rate for learners and light red markers show the median CR rate for all mice). B, Left graphs show probe trials from two example mice from each of the three groups trained at different trace intervals. Numbers on x-axes give the median latency to CR onset and peak for each example mouse. Note that the latencies to onset are similar across these examples, while the latencies to peak shift with the longer training intervals. Right graph shows group data for CR latencies and amplitude for the 3 training groups beginning with initial expression until the last training session (black lines show median latencies, shaded gray region indicates interquartile range). Mice showed CR onsets that did not vary across the different trace intervals on the final training days (trace 50–250 vs 50–350, Wilcoxon rank: U = 26, n = 4 and 12, p = 0.42; trace 50–350 vs 50–450, U = 5, n = 3 and 4, p = 0.69). Latencies to CR criterion and peak were different between trace 50–250 and 50–350 (Wilcoxon rank: U = 45, n = 4 and 12, p < 0.001 and U = 44, p < 0.001, respectively), with the peaks of CRs appropriately timed for the training intervals used. Mice trained to trace 50–450 showed early latencies relative to the training interval, and were not different from mice trained to 50–350 for any parameter (Wilcoxon rank, Onset: U = 5, n = 3 and 4, p = 0.69; Criterion: U = 8, p = 0.31; Peak: U = 6, p = 0.57). CR amplitudes were similar between the three training groups (rightmost graph, black lines show median for each training day, gray shaded area shows interquartile range, dotted line indicates median amplitude of trace 50–250 mice on the first day of expression as reference).
Mentions: To further investigate the timing of CRs, two sets of naïve mice were trained to different trace intervals (trace 50–350: n = 8, and trace 50–450: n = 8). Fewer mice were able to learn the longer intervals within 12 training sessions (4/8 mice for trace 50–350 and 3/8 mice for trace 50–450; Fig. 3A), suggesting that an interval even 100 ms longer was more difficult for the mice to learn. Of the mice that were able to learn trace 50–350, the latency to the peak amplitude of responses was significantly later than that observed for mice trained with a trace 50–250 protocol (trace 50–350: median = 407 ms; IQR = 357–456 ms; Mann–Whitney–Wilcoxon Rank test, U = 44, n = 4 and 12, p < 0.001d; Fig. 3B, top and middle, “peak”), and was also well timed to the air puff (H0: mean = 400 ms, t = 0.32, df = 3, p = 0.84e). However, the latency to the onset of CRs (i.e., when the eyelids initially began to close during a given trial; Fig. 1C) was not different between mice that learned trace 50–250 versus those that learned trace 50–350 (tr50–250: median = 105 ms, IQR = 99–112 ms; tr50–350: median = 107 ms, IQR = 98–129 ms; U = 26, n = 4 and 12, p = 0.42f; Fig. 3B, top and middle, “onset”). The results are consistent with previous reports of mice trained in delay eyeblink conditioning using different interstimulus intervals, and demonstrate that mice initiated learned responses 100–125 ms after CS onset regardless of the training interval used, whereas timing the peaks of CRs to coincide with the air puff (Chettih et al., 2011; Heiney et al., 2014). Furthermore, a significant shift in the latency at which CR criterion was met (FEC > 0.1; Fig. 1C) between trace 50–250 and 50–350 suggests that the difference in the timing of CR peaks was modulated by the initial velocity of behavioral responses (tr50–250: median = 164 ms, IQR = 148–181 ms; tr50–350 median = 234 ms, IQR = 192–266 ms; U = 45, n = 4 and 12, p < 0.001g; Fig. 3B, top and middle, “criterion”). Interestingly, the latencies to onset, criterion and peak amplitude of CRs for mice that learned the trace 50–450 task protocol were not different to those observed for trace 50–350 (tr50–450 Onset: median = 100 ms, IQR = 98–114 ms, U = 5, n = 3 and 4, p = 0.69h; Criterion: median = 225 ms, IQR = 223–285 ms, U = 8, p = 0.31i; Peak: median = 410, IQR = 403–411 ms, U = 6, p = 0.57j; Fig. 3B, middle and bottom). The data suggest that the mice able to learn trace 50–450 did so by expressing early CRs that were not well timed to the air puff, and that this interval may be too long for naïve mice to express well timed responses. Similar to trace 50–250, mice trained at longer intervals also initially showed low amplitude CRs at initial expression that gradually increased over several sessions (Fig. 3B, right graphs).

Bottom Line: To identify the circuitry involved, we made restricted lesions of the medial prefrontal cortex (mPFC) and found that learning was prevented.Anatomical data from these critical regions showed that mPFC and amygdala both project to the rostral basilar pons and overlap with eyelid-associated pontocerebellar neurons.The data further reveal a specific role for the amygdala as providing a conditioned stimulus-associated input to the cerebellum.

View Article: PubMed Central - HTML - PubMed

Affiliation: Center for Learning and Memory, University of Texas at Austin , Austin, Texas 78712.

ABSTRACT
Trace eyeblink conditioning is useful for studying the interaction of multiple brain areas in learning and memory. The goal of the current work was to determine whether trace eyeblink conditioning could be established in a mouse model in the absence of elicited startle responses and the brain circuitry that supports this learning. We show here that mice can acquire trace conditioned responses (tCRs) devoid of startle while head-restrained and permitted to freely run on a wheel. Most mice (75%) could learn with a trace interval of 250 ms. Because tCRs were not contaminated with startle-associated components, we were able to document the development and timing of tCRs in mice, as well as their long-term retention (at 7 and 14 d) and flexible expression (extinction and reacquisition). To identify the circuitry involved, we made restricted lesions of the medial prefrontal cortex (mPFC) and found that learning was prevented. Furthermore, inactivation of the cerebellum with muscimol completely abolished tCRs, demonstrating that learned responses were driven by the cerebellum. Finally, inactivation of the mPFC and amygdala in trained animals nearly abolished tCRs. Anatomical data from these critical regions showed that mPFC and amygdala both project to the rostral basilar pons and overlap with eyelid-associated pontocerebellar neurons. The data provide the first report of trace eyeblink conditioning in mice in which tCRs were driven by the cerebellum and required a localized region of mPFC for acquisition. The data further reveal a specific role for the amygdala as providing a conditioned stimulus-associated input to the cerebellum.

No MeSH data available.


Related in: MedlinePlus