Reciprocal Inhibitory Connections and Network Synchrony in the Mammalian Thalamus

See allHide authors and affiliations

Science  22 Jan 1999:
Vol. 283, Issue 5401, pp. 541-543
DOI: 10.1126/science.283.5401.541


Neuronal rhythmic activities within thalamocortical circuits range from partially synchronous oscillations during normal sleep to hypersynchrony associated with absence epilepsy. It has been proposed that recurrent inhibition within the thalamic reticular nucleus serves to reduce synchrony and thus prevents seizures. Inhibition and synchrony in slices from mice devoid of the γ-aminobutyric acid type-A (GABAA) receptor β3 subunit were examined, because in rodent thalamus, β3 is largely restricted to reticular nucleus. In β3 knockout mice, GABAA-mediated inhibition was nearly abolished in reticular nucleus, but was unaffected in relay cells. In addition, oscillatory synchrony was dramatically intensified. Thus, recurrent inhibitory connections within reticular nucleus act as “desynchronizers.”

Inhibitory circuits arising in the reticular thalamic nucleus (RTN) play important roles in various oscillatory activities related to sleep and some epilepsies (1–5). The major projections of inhibitory neurons in RTN are onto relay neurons in dorsal thalamus, but recurrent collaterals also provide intranuclear inhibition (6). It has been hypothesized that the latter connections regulate RTN inhibitory output during thalamic oscillations and prevent the hypersynchrony of generalized absence epilepsy (7–9). Inhibitory postsynaptic currents (IPSCs) in RTN neurons are mediated by the major inhibitory neurotransmitter, γ-aminobutyric acid, through GABA type A receptors. IPSCs in RTN differ from those in relay neurons (10), presumably due to differences in GABAA receptor subunit composition (11), which ultimately affect ligand affinity, channel gating, and modulation (12). In rodent thalamus, β3 is one of a limited number of GABAAsubunit mRNAs expressed in RTN and is absent from relay nuclei (11). Despite a lack of widespread gene expression in the adult rodent brain, β3 knockout mice (β3 −/−) exhibit many neurological impairments and are considered a model of Angelman's syndrome in humans (13). We examined inhibitory function in thalamic slices of β3 knockout mice to test whether elimination of this subunit would suppress intra-RTN inhibition and thus promote intrathalamic synchrony (14).

Voltage clamp recordings (15) were performed in the presence of GABAB and ionotropic glutamatergic blockers to specifically isolate GABAA receptor–mediated IPSCs (10, 16). Spontaneous IPSCs (sIPSCs) in RTN neurons from controls were long lasting, with an average weighted decay time constant (τD,W) of 74.7 ± 5.9 ms (n= 19; Fig. 1, C and D). Infrequent sIPSCs were observed in RTN neurons of β3 knockout mice and were much smaller and more brief than in wild-type (β3 +/+) littermates. sIPSC decay was almost three times faster (τD,W = 27.4 ± 2.4 ms,n = 26, P < 0.0001) in knockouts, whereas sIPSC amplitude and frequency in RTN were reduced by more than 50% (P < 0.0001; Fig. 1, A through D). Overall inhibitory efficacy was estimated by integrating total sIPSC charge per 1-s interval. In controls, the total charge was 3840 ± 1130 pC/s (n = 19), compared to a much reduced value of 130 ± 20 pC/s (n = 26, P < 0.0005) in β3 knockouts. By contrast, excitatory connections were intact in RTN neurons of knockout mice. Spontaneous excitatory postsynaptic currents (sEPSCs) were comparable in amplitude (11 versus 14 pA in β3 +/+ and β3 −/−, respectively), half-width (1.1 versus 1.2 ms), and frequency (2.1 versus 3.0 Hz, n = 7 each).

Figure 1

sIPSCs in RTN and VB neurons from wild-type (β3 +/+) and knockout (β3 −/−) mice. (A and B) Continuous traces depicting sIPSCs in individual RTN neurons from β3 +/+and β3 −/− mice. Note reduced sIPSC amplitude and frequency in the β3 −/− RTN neuron. (C) Averaged sIPSCs from cells in (A) (n = 93 IPSCs) and (B) (n = 51) superimposed on same time scale to illustrate decreased peak amplitude and duration in β3 −/− RTN neurons. (D) Population data (mean ± SE) for RTN sIPSC properties in β3 +/+(n = 19 cells) and β3 −/− mice (n = 26). *** = P < 0.0001. (E and F) Continuous sIPSC traces of individual VB neurons from β3 +/+ and β3 −/− mice. Scale is same as in (A) and (B). (G) Average sIPSCs in VB cells from (E) (n = 167 IPSCs) and (F) (n = 294). (H) Population data (mean ± SE) for VB sIPSC properties of β3 +/+ (n = 9 cells) and β3 −/− (n = 11) mice.

Inhibition in thalamic relay neurons of the ventrobasal (VB) complex was unchanged in β3 knockout mice—sIPSC amplitudes, decay kinetics, and frequency were comparable in wild-type control and knockout mice (Fig. 1, E through H). As in rat (10), sIPSC decay was faster in VB neurons (Fig. 1G) than in RTN neurons (Fig. 1C).

To confirm the specific reduction in intra-RTN inhibition, we also assessed electrically evoked responses (17). Monosynaptic evoked IPSC (eIPSC) amplitudes in RTN neurons from β3knockout mice were reduced by about half in both amplitude and duration (Fig. 2). By contrast, eIPSCs in VB neurons of wild-type and knockout mice were indistinguishable in terms of amplitude (1.31 ± 0.33 nA, n = 6 versus 0.96 ± 0.28 nA, n = 6, P > 0.4) and duration (τD,W of 21.1 ± 3.4 ms versus 21.1 ± 2.5 ms, P > 0.9).

Figure 2

Properties of monosynaptic eIPSCs recorded in RTN neurons. (A) Averages of eIPSCs evoked in individual RTN neurons of β3 +/+ (n = 5 neurons) and β3 −/− (n = 7) mice as indicated. The shaded area flanking the averaged eIPSC represents the mean ± one standard error; • = onset of stimulus. (B) Histogram of mean peak eIPSC amplitude and τD,W in β3 +/+ and β3 −/− mice; *** = P < 0.001, * = P < 0.05.

These data show that inhibitory efficacy is specifically reduced for the recurrent connections between RTN neurons of β3knockout mice, thus rendering part of the thalamic circuit defective. The thalamic network responsible for generating sleep and epilepsy-related oscillatory activity requires intact excitatory and inhibitory connections between RTN and relay cells (3–5). These connections appear to be normal in β3 knockout mice as judged by the aforementioned properties of EPSCs in RTN cells and IPSCs in relay neurons. These mice then provided a unique opportunity to directly test for the functional role of RTN inhibitory collaterals (4, 7). Oscillatory activity was evoked by stimulation of the internal capsule in vitro (18) and, as previously shown (5), was GABAA receptor–dependent. In control slices, oscillatory responses were characterized by an initial fixed-latency burst (asterisks; Fig. 3A, left), followed by two to six repetitive bursts with variable latencies (arrowheads; Fig. 3A, left). In contrast, highly synchronous activity (for example, Fig. 3A, right) was obtained in the majority of slices (39 of 48, ∼81%) from β3 knockouts. Similar synchrony was rarely observed in wild-type slices (4 of 33, ∼12%; P < 0.0001, Fisher's Exact Test). Interestingly, synchronous oscillations occurred spontaneously in a few knockout slices. The timing of burst occurrence was particularly notable in the knockouts—oscillations were both highly synchronous and phase-locked throughout their duration (diamonds; Fig. 3A, right).

Figure 3

Simultaneous multiunit recordings in VB and RTN during evoked oscillatory activity in acute thalamic slices in β3 +/+ and β3 −/− mice. (A) Consecutive raw traces of thalamic oscillations in β3 +/+slices reveal repetitive burst activity (asterisks, arrows) at variable latencies (left panel; note considerable trial-to-trial variability). In contrast, highly synchronous, phase-locked oscillations were obtained in β3 −/− mouse slices (right panel; little trial-to-trial variability); • = onset of stimulus. Vertical bars and diamonds represent fixed intervals of 121 ms. (B) Autocorrelograms of thalamic oscillations represented in (A). Note the small oscillatory components adjacent to the central peak in β3 +/+ mice (arrowheads indicate satellite peaks) versus the multiple large peaks in β3 −/− mice. (C) Histogram of the oscillatory index computed from autocorrelograms from slices of β3 +/+ and β3 −/−mice, *** = P < 0.001. (D) Phase difference in VB as a function of distance during oscillations generated in slices from β3 +/+ and β3 −/− mice; □ = β3 +/+, ▪ = β3 −/−, *** = P < 0.001, * = P < 0.05, n = two to seven slices for each point.

Autocorellograms derived from control slices typically had distinctive large central peaks with much smaller and irregular satellite peaks (arrowheads; Fig. 3B, left). In contrast, autocorrelograms from knockout mice showed numerous satellite peaks of gradually decaying amplitude, with a less distinct central peak (Fig. 3B, right). In control slices, 29 ± 6% (n = 9) of the neuronal activity in RTN was deemed oscillatory (19), compared to 70 ± 6% (n = 11, P < 0.001) in β3 knockout slices (Fig. 3C). Comparable oscillatory indices were obtained from VB recordings (β3 +/+, 27 ± 4%, n = 10 versus β3 −/−, 59 ± 5%,n = 12; P < 0.001). Therefore, knockout of β3 resulted in highly synchronous and oscillatory activity throughout the RTN-VB network. Further support for synchrony was indicated by large-amplitude maximal extracellular field potentials (β3 −/−, 86 ± 7.7 μV,n = 36 versus β3 +/+, 13.6 ± 2.9 μV, n = 30; P < 0.0001), suggesting that local groups of neurons fire nearly simultaneously during the oscillatory response. Not only was local synchrony high, but oscillatory activity throughout the slice was tightly phase-locked. Cross-correlation analysis of dual recordings in VB complex demonstrated that phase differences in the knockouts were negligible (<4 ms) at distances of up to 500 μm (Fig. 3D). In contrast, oscillatory activity in control slices was characterized by phase lags that increased linearly with distance up to ∼70 ms at 500 μm separation (Fig. 3D).

These results demonstrate that GABAAreceptor–mediated IPSCs in RTN are impaired in β3knockouts. The resulting reduced inhibitory efficacy between RTN neurons thus leads to hypersynchrony that is detrimental to the normal function of the thalamic circuit. Consistent with this hypothesis, local application of GABAA antagonists within RTN of wild-type mouse slices enhanced oscillatory activity (20), in accordance with results previously reported in rat (7). These results indicate that reduced intra-RTN inhibition is sufficient to produce thalamic hypersynchrony.

The powerful effects of β3 ablation could be explained by the dependence of GABAA responses in RTN neurons on relatively few receptor isoforms (11). GABAA receptor function is less impaired in hippocampal neurons of β3 knockouts (21), which may reflect a lower dependence on this β subunit (11). The shortening of IPSC duration in β3 knockouts potentially relates IPSC properties to neural circuit activity in the thalamus and may show how the specific functional deletion of intra-RTN connections affects phasic oscillations (7, 8). The role of intra-RTN connections in thalamic oscillations is controversial. These connections might either facilitate (22) or dampen oscillations (7, 9, 23). Our results, showing highly synchronous oscillations in animals lacking functional GABAA receptors in RTN, indicate that intra-RTN inhibition desynchronizes thalamic activity. Further, these data show how inactivation of a postsynaptic receptor gene can result in functional deletion of a specific neuroanatomical circuit and provide information regarding mechanisms of human disease (13).

  • * To whom correspondence should be addressed. E-mail: John.Huguenard{at}


Stay Connected to Science

Navigate This Article