Technical Comments

Response to Comment on "Cerebellar LTD and Learning-Dependent Timing of Conditioned Eyelid Responses"

Science  09 Apr 2004:
Vol. 304, Issue 5668, pp. 211
DOI: 10.1126/science.1094374

We disagree with Christian et al. (1) regarding the superior sensitivity of electromyogram (EMG) recording. The advantages of magnetic distance measurement technique (MDMT) over EMG recordings can best be illustrated by using both recording methods simultaneously in the same mice, and by evaluating both technologies with high-speed video recordings (Fig. 1). The video traces demonstrate that MDMT recordings are more accurate than EMG in both the spatial and temporal domain. Due to their variable delay, EMG recordings are especially inappropriate for investigating the timing of eyelid movements in mice. Signals generated by the mouse eyelid muscles are frequently contaminated by those of surrounding muscles (2). However, EMG recordings may be suitable for eyelid recordings in rabbits and cats, in which the various facial muscle groups are much further separated.

Fig. 1.

High-speed video evaluation of MDMT and EMG recordings in mice. To evaluate the spatiotemporal accuracy, high-speed video, MDMT, and EMG were applied simultaneously in the same mice. (A) Schematic showing how the three recording methods were implemented around a single eye. (B) An example of simultaneous recordings using MDMT, high-speed video, and EMG during a paired trial (from top to bottom). The MDMT signal closely follows that of the automated video signal in space and time (with regard to both onset and peak latency), while the converted EMG signal shows more variation in its amplitude and precedes the video signal at different latencies during various parts of the trace. Note that the unconditioned response can be monitored successfully with MDMT, but not with EMG. Micrographs 1 to 5 (right panel) correspond to the numbers in the video trace. See (2) for details.

The main advantages and disadvantages of using EMG, high-speed video, and MDMT recordings for eyeblink conditioning in mice are listed in Table 1. It is important to record in the freely moving preparation, because eyeblink conditioning behavior in mice can be negatively affected by stress due to fixation. Thus, high-speed video can be used to record eyelid movements, but is not practical for conditioning because fixation is required. The high-speed video recordings showed that the EMG method can provide both false positive and false negative data (2). These artifacts largely explain why percentages of conditioned responses obtained over various sessions with EMG (3, 4) differ from those obtained with MDMT (2, 5). Because of the superior recording over time, the traces obtained with MDMT also lend themselves well to accurately identifying different initial components such as startle responses (<75 ms) or initial extra-cerebellar components, which can be readily distinguished from cerebellar conditioned responses and do not interfere with their quantification. This precise quantification was particularly important in our study of the PKCi mutant (5), because the percentage of conditioned responses in the mutant was lower than in wild types, but high enough to perform statistical analysis. The same precision was important to show that learning-dependent timing in mice was not expressed in onset latency, but only in peak latency. In this respect, the mice data partly diverge from those in rabbits in which both parameters appear to be controlled. Fortunately, in both rabbits and mice, peak latency shows a shift toward the moment of the unconditioned stimulus during conditioning when induction of long-term depression (LTD) operates normally. This consistency also makes sense, because this parameter will ultimately reflect whether the eyelid is optimally closed at the appropriate moment. Finally, MDMT— but not EMG recordings—allow us to accurately determine sensitivity to the unconditioned stimulus. Control experiments show that PKCi mutants do not have unidentified deficits in this respect (Fig. 2). Our lesion experiments, which have been supported by silver stainings for degenerated neurons and axons, showed that part of the eyeblink components in mice originates outside the anterior interposed nucleus (AIP) of the cerebellum. If the outcome had been different [as we expected, based upon similar lesion studies in the rabbit (6)], our conclusion in (5) that LTD contributes to learning-dependent timing would have been even more significant. That the impact of AIP lesions on the timing is even stronger than a blockage of LTD does not mean that LTD is negligible. On the contrary, one may assume that lesioning the AIP eliminates the impact of LTD at the parallel fiber-Purkinje cell synapse upstream. Thus, the combination of experiments presented in (5) suggests that both LTD and non-LTD related mechanisms play a role in learning-dependent timing.

Fig. 2.

Sensitivity to unconditioned stimulus in LTD-deficient PKCi mutants does not differ from that in wild types. (A) Plots of peak amplitude of the unconditioned response versus its peak velocity show the same slope for PKCi mutants and wild types. This relation also holds when airpuffs are used as the unconditioned stimulus instead of electric shocks. (B) Experimental design used to deliver airpuffs in freely moving mice in which eyeblink movements are recorded with MDMT.

Table 1.

Comparison of techniques for recording eyeblink responses in mice.

Recording technique Allows freely moving prep. Measurement of amplitude Measurement of timing Electric shock produces artifact
EMG Yes Reasonable Poor Yes
High-speed video No Perfect Perfect No
MDMT Yes Very good Perfect No

Christian et al.(1) claim that our report failed to cite relevant studies from Thompson's laboratory (79). First, their memory localization studies assumed that the red nucleus is the only target area of the AIP (7), but the AIP has since been shown to target at least 10 other projection areas including the thalamus, nucleus reticularis tegmenti pontis, zona incerta, and inferior olive (10, 11). Second, their work on teaching signals that relay information about the unconditioned stimulus assumed that climbing fibers can fire well over 50 Hz (8). However, this frequency is typically around 1 to 2 Hz and cannot be higher than 10 to 15 Hz based on theoretical and experimental data (11). Third, they used EMG eyeblink recordings in mice without the evaluation described above, and did not directly determine the sensitivity to unconditioned stimuli used in the eyeblink paradigm (9). Finally, their mice were trained using tones of 1 kHz at 80 dB as a conditioned stimulus, but mice have a hearing sensitivity threshold of at least 90 dB at frequencies below 2 kHz (12). If mice hear anything at these low frequencies, their perception will be disturbed by variant noisy signals at the edge of the threshold. The subsequent lack of a precise conditioned stimulus will dramatically burden their perception of the moment of its onset. It will therefore be difficult to appropriately address issues of timing in these eyeblink studies (9).

References and Notes

Related Content

Navigate This Article