Abstract
Fast Gamma Oscillations Are Generated Intrinsically in CA1 without the Involvement of Fast-Spiking Basket Cells
Craig MT, McBain CJ. J Neurosci 2015;35(8):3616–3624.
Information processing in neuronal networks relies on the precise synchronization of ensembles of neurons, coordinated by the diverse family of inhibitory interneurons. Cortical interneurons can be usefully parsed by embryonic origin, with the vast majority arising from either the caudal or medial ganglionic eminences (CGE and MGE). Here, we examine the activity of hippocampal interneurons during gamma oscillations in mouse CA1, using an in vitro model where brief epochs of rhythmic activity were evoked by local application of kainate. We found that this CA1 KA-evoked gamma oscillation was faster than that in CA3 and, crucially, did not appear to require the involvement of fast-spiking basket cells. In contrast to CA3, we also found that optogenetic inhibition of pyramidal cells in CA1 did not significantly affect the power of the oscillation, suggesting that excitation may not be essential for gamma genesis in this region. We found that MGE-derived interneurons were generally more active than CGE interneurons during CA1 gamma, although a group of CGE-derived interneurons, putative trilaminar cells, were strongly phase-locked with gamma oscillations and, together with MGE-derived axo-axonic and bistratified cells, provide attractive candidates for being the driver of this locally generated, predominantly interneuron-driven model of gamma oscillations.
Commentary
Neuronal synchrony is the coordinated activity of a group of neurons. Changes in behavior correlate with dynamic changes in neuronal synchrony, thus it is thought that synchrony may drive behavior and information processing (1) (see also Histed and Maunsell [2]). Synchrony is described in different frequency bands: α (8–12 Hz), β (15–30 Hz), δ (> 4 Hz), θ (4–8 Hz), and γ (30–90+ Hz) (3, 4). Conceptually, these rhythms continuously ebb and flow, dance back and forth, like traffic through a multi-way intersection from one neuronal ensemble (or structure) to the next, as each structure entrains the next (like a traffic signal) with the putative flow of information (5). Gamma frequency oscillations (GFOs), by nature of their speed, allow the greatest precision (milliseconds) to entrain specific outputs or neuronal groups that underlie timely coordination of specific behaviors, such as memories, patterns, and so forth (6). Consideration of synchrony is important to epilepsy research as it has been considered a disorder of neuronal synchrony (3). Further, learning and memory (often disrupted by epilepsy) have been linked to synchrony and, when disrupted, are associated with alterations in synchrony, particular GFOs. Whether or not epilepsy is truly a disorder of neuronal synchrony cannot be fully covered here. However, GFOs are thought to precede seizures (7–9), and recent work suggests that they may dissipate just prior to seizure onset (10). Thus, the origins of GFOs are important and have been subjected to intense study. While studied in multiple brain regions, hippocampal GFOs are of particular interest owing to their possible role in temporal lobe epilepsy and in learning and memory (11, 12).
Within hippocampal substructures, GFOs have been differentially characterized because neurons differ anatomically in each structure. These location-specific characterizations have included the factors that trigger GFOs, the characteristic frequency of GFOs, interneuron (IN) subtypes involved, and whether pyramidal neurons (PN) are necessary (PING, pyramidal–interneuron network gamma oscillations) or not (ING, interneuron network gamma oscillations) (13). IN subtypes are based on morphology (shape, sources of input and output), firing patterns (fast or slow), cotransmitters/histology and embryologic origin (medial gangiolonic eminence [MGE] or caudal ganglionic eminence [CGE]). In previous work, the authors found that embryonic origin coincides with cotransmitters/histology but not necessarily with other features such as firing patterns or anatomy (14). Previous work by others (4, 15) has concluded that in CA1 hippocampus (output from hippocampus), GFOs are extrinsically entrained by CA3 or entorhinal cortex GFOs, require fast-spiking basket-cell INs, and involve a PING model, all similar to CA3 GFOs. In the current study, the authors sought to reexamine this finding with the finer detail of understanding these features in terms of IN embryonic origin (CGE vs MGE).
The authors took advantage of two mouse lines engineered to express green fluorescent protein (GFP) selectively in IN derived from either the CGE or MGE. The authors prepared horizontal hippocampal slices for electrophysiological recordings from juvenile mice. They simultaneously recorded extra-cellular field potentials in stratum radiatum (sr) and whole-cell current clamp recordings of neuronal spiking of one or two visually identified fluorescent INs (or nonfluorescent PNs); recordings were made in either CA3 or CA1. GFOs were evoked by focal application of the glutamatergic excitatory agonist kainate (KA) in sr of either CA3 or CA1; these transient GFOs developed immediately after KA application and persisted up to 10 seconds. Cuts were made to disconnect CA1 from CA3 or subiculum in some experiments. Field-recorded GFOs were analyzed for frequency content and compared with the phase of IN or PN spiking. Anatomic reconstructions of PNs or INs (using biocytin in the whole-cell recording electrode to label the cell) determined neuronal type.
The authors found that transient GFOs, rapidly triggered by KA in CA1, were broadly faster (60–80 Hz) than GFOs triggered in CA3 (40 Hz). Disconnecting CA1 determined that these higher frequencies were intrinsic to CA1. Further, peak frequencies accelerated further by disconnecting the subiculum. This finding is different from previous studies that demonstrated dependence of CA1 GFOs on CA3, where GFOs were triggered altogether differently and required prolonged application of carbachol (15).
Consistent with previous studies, the authors found that CA3 GFOs were phase-locked to MGE-derived fast-firing basket INs (4) that project locally to PN cell bodies in stratum pyramidale (sp). However, this was not the case in CA1 since CA1 PNs did not fire during CA1 GFOs. Rather, CA1 GFOs were phase-locked with CGE trilaminar INs (that project locally to sp, sr, and stratum oriens [so] and forward to the subiculum) and CGE back-projecting INs (that project back to CA3 sr and so). CA1 GFOs were also phase-locked with MGE axo-axonic INs (that project locally to sp) and bistratified INs (that project locally to so and sr).
Finally, a third mouse line engineered to express the light-activated Archaerhodopsin (Arch) in PNs was used to determine the effect of PN silencing. In CA3, light-triggered PN silencing substantially reduced local GFOs. This finding is consistent with a PING model of CA3 GFOs previously described (4). However, in CA1, PN silencing, if anything, increased GFO frequencies. This is consistent with an ING model of GFO generation in CA1.
In summary, the authors uniquely found that dynamically activated (by KA) GFOs in CA1 are of relatively high frequency, intrinsically activated, and do not rely on either fast-spiking basket INs or PNs, the latter being consistent with an ING model in CA1. These features are all distinct from CA3 GFOs (lower gamma bands, rely on basket INs and PNs, and a PING model). CA1 GFOs primarily rely on MGE-derived INs but with some CGE-derived exceptions. CA1 GFOs are not dependent on CA3 GFOs; rather, the presence of back-projecting INs suggests that CA1 GFOs influence CA3. These distinctions are important, as CA1 is the hippocampal output, and further distinguish the unique structure and independent function of CA1.
While KA-evoked GFOs are quickly generated and thus consistent with dynamic synchrony, their resemblance to native, physiological, in vivo GFOs is unknown. There are similar criticisms to carbachol-evoked GFOs (15). While the authors provide results from a technically challenging large number of recordings, in some cases of rare INs, the recordings are sparse. This issue also prevents any generalizations about differences in dorsal versus ventral hippocampus.
As the authors point out, the next logical step is to investigate, through optogenetic control of specific INs, the role of INs in CA1 versus CA3 GFOs in vivo. Specifically, this could allow the determination of how the flow of GFOs from CA3 to CA1 to entorhinal cortex is modulated and entrained by region and type-specific INs. The presence of multiple IN-types in CA1 GFOs queries whether each IN regulates its own unique GFO or not. Of importance, given the putative role of GFOs in epilepsy and behavior, definitive role(s) for region and IN-specific GFOs can finally emerge. While it is clear that optogenetic activation of INs (subtype not yet clear) can terminate seizures (16), the role of GFOs is yet to be determined. Epilepsy may be the result of specific loss of INs in charge of their unique GFOs. Secondary loss of specific INs and their GFOs may underlie epilepsy-mediated cognitive loss. Alternatively, maladaptive, overactive GFOs may take over to mediate epilepsy and cognitive loss. Each of these features could be optogenetically targeted to counter pathophysiology and restore the balance to the appropriate dance of GFOs.