This Week in The Journal

Teresa Esch

Inhibiting SK Channels to Allow LTP during Wake and Sleep

Cezar M. Tigaret, Sophie E.L. Chamberlain, Joseph H.L.P. Sadowski, Michael C. Ashby, and Jack. R. Mellor

Episodic learning occurs when hippocampal glutamate release depolarizes postsynaptic neurons enough to elicit spiking. Under these conditions, activation of NMDA receptors (NMDARs) leads to calcium influx and downstream signaling that produces long-term potentiation (LTP). For the memory to persist, it must be consolidated. This occurs during sleep, when synapses are reactivated in the presence of high-frequency oscillations [sharp wave ripples (SWRs)], resulting in additional NMDAR-dependent LTP.

Although LTP occurs as rodents explore an arena, stimulating neurons in slices with patterns recorded during exploration does not elicit LTP. This is because calcium influx into dendritic spines activates small-conductance calcium-activated potassium (SK) channels, which counter depolarization and thus limit NMDAR activity. Activation of muscarinic receptors (M1R) by acetylcholine, which is released in the hippocampus during exploration, leads to inhibition of SK channels, allowing LTP to proceed. During sleep, however, acetylcholine levels are low, so the brake on LTP must be lifted in another way—possibly by activation of metabotropic glutamate receptors (mGluR1) during SWRs.

Tigaret et al. examined LTP at CA3 inputs to CA1 pyramidal neurons in slices. Under baseline conditions, pairing presynaptic stimulation with postsynaptic spikes produced excitatory postsynaptic calcium transients (EPSCaTs) in dendritic spines and triggered LTP. An mGluR1 antagonist reduced EPSCaT amplitude and blocked LTP, effects that were prevented by co-applying an M1R agonist or an SK channel blocker. Activating M1Rs in the absence of mGluR1 antagonist also increased EPSCaT amplitude, but the effect was small and variable, suggesting it was partially occluded by activation of mGluR1 during stimulation. Although stimulating neurons in slices with patterns mimicking spike trains recorded in vivo did not induce LTP under baseline conditions, the stimulation induced LTP when it was combined with administration of SK blockers or M1R agonists. Importantly, natural spike trains also induced LTP when paired with stimulation mimicking SWRs recorded in vivo, but this LTP was blocked when mGluR1 antagonists were administered, suggesting that mGluR1 activation enables LTP induction during SWRs.

These data suggest that LTP induction during initial learning and consolidation processes depends on activation, respectively, of M1R and mGluR1, both of which inhibit SK channels. Future work should examine whether these pathways diverge at other points to differentiate initial learning and consolidation.

Latency Coding of Olfaction in Honeybees

Marco Paoli, Angela Albi, Mirko Zanon, Damiano Zanini, Renzo Antolini, et al.

(see pages 9240–9251)

The early stages of odor processing are similar in mammals and insects. In insects, odors are detected by receptors present in olfactory receptor neurons (ORNs) on antennae. Each ORN expresses 2–4 odorant receptor types, and ORNs expressing the same receptor types converge on glomeruli in the antennal lobe. Although all ORNs innervating a given glomerulus express the same limited set of odorant receptors, every odor activates multiple receptors—and therefore multiple glomeruli. The specific subset of glomeruli activated and the amplitude and temporal order of activation differ across odors, however. Odor identity can be deciphered from the mean response amplitude in individual glomeruli across the antennal lobe, and Paoli, Albi, et al. report that it can also be decoded from relative response latencies.

Odors evoke distinct temporal patterns of activation across antennal lobe glomeruli. The relative latency at which glomeruli are activated can be used to discriminate odors. See Paoli, Albi, et al. for details.

The authors used calcium imaging in honeybees to detect neural responses to five odorants, each of which strongly activated 2–4 of the glomeruli imaged. The latency of responses differed across glomeruli, but the pattern of latencies evoked by a given odor was similar across trials and across individuals. This suggests that response latency carries odor-specific information. Furthermore, response latency and amplitude were uncorrelated, indicating that latency might provide additional information to improve odor encoding. To test this, the authors constructed latency rank vectors representing the order in which glomeruli were activated by each odor in each animal. Template vectors derived from the average latency profiles across bees could be used to identify which odor was presented with 60–70% accuracy for 4 of 5 tested odors—a discrimination performance similar to that of the response amplitude code. Furthermore, prediction accuracy was improved when using a combination of latency and amplitude information. Finally, odor discrimination based on latency, amplitude, or a combination predicted bees' performance on an odor discrimination test.

These results suggest that a neural code incorporating information about relative glomerular response latencies and amplitudes may be used for odorant identification in honeybees. Whether bees take advantage of this information must be verified, but notably, a similar code has been suggested for mice. The latency code might be particularly useful for rapidly categorizing odors, enabling rapid behavioral responses during food seeking and threat avoidance.