Paradoxical enhancement of long-term potentiation in poor-learning rats at low test stimulus intensities

Much empirical evidence and numerous theoretical models point to modification of synaptic efficacy as a mechanism for memory formation. To evaluate theoretical models, it is necessary to obtain quantitative experimental data relating learning to experimentally induced synaptic efficacy changes (such as long-term potentiation, LTP). An important problem in this type of experiment is how to quantify the LTP induced by a given stimulation protocol. Of relevance is the informally well-known observation that LTP magnitude appears to vary as a function of the intensity of the stimulus used to evoke baseline responses. The present study found that using a measure of LTP that circumvents this variation, a strong negative correlation of learning with potentiation emerges. Spatial learning ability was compared with the magnitude of subsequent LTP induction as follows: rats underwent a day of spatial training in a watermaze followed by 5 days of bilateral perforant path tetanisation. Baseline electrophysiological responses were evoked over a range of stimulus intensities (input/output [IO] curves) before and after tetanisation. Although LTP was observed across the whole of the IO curve, it showed a smooth decline with increasing current. The animals were then grouped according to their watermaze performance and IO curves compared between good and poor learners. After tetanisation, there was a negative within-animal correlation between learning and evoked potential size with weak test stimuli and a positive correlation with strong stimuli. The decline of LTP across the IO curve differed between good and poor spatial learners; the poor learners showed higher percentage potentiation with test stimuli close to zero intensity, but a faster decrease in LTP across the curves. The findings are therefore: (1) the measured amount of LTP declined systematically with increasing stimulus strength, and (2) the parameters of the decline correlated with spatial learning ability. These results raise two important issues. First, because measured LTP varied systematically across the IO curve, it appears that for quantitative analyses the widely used method of LTP measurement using a single test stimulus intensity risks missing significant features of the data. It is suggested that a measure be used that incorporates data from a range of stimulus intensities. Second, when such a measure is used there is a striking negative correlation of spatial learning ability with LTP. These apparently paradoxical results are discussed.