|Home | About | Journals | Submit | Contact Us | Français|
The medial geniculate body (MGB) has three major subdivisions - ventral (MGV), dorsal (MGD) and medial (MGM). MGM is linked with paralaminar nuclei that are situated medial and ventral to MGV/MGD. Paralaminar nuclei have unique inputs and outputs when compared with MGV and MGD and have been linked to circuitry underlying some important functional roles. We recorded intracellularly from cells in the paralaminar nuclei in vitro. We found that they possess an unusual combination of anatomical and physiological features when compared to those reported for “standard” thalamic neurons seen in the MGV/MGD and elsewhere in the thalamus. Compared to MGV/MGD neurons, anatomically, 1) paralaminar cell dendrites can be long, branch sparingly and encompass a much larger area. 2) their dendrites may be smooth but can have well defined spines and 3) their axons can have collaterals that branch locally within the same or nearby paralaminar nuclei. When compared to MGV/MGD neurons physiologically 1) their spikes are larger in amplitude and can be shorter in duration and 2) can have dual afterhyperpolarizations with fast and slow components and 3) they can have a reduction or complete absence of the low threshold, voltage-sensitive calcium conductance that reduces or eliminates the voltage-dependent burst response. We also recorded from cells in the parafascicular nucleus, a nucleus of the posterior intralaminar nuclear group, because they have unusual anatomical features that are similar to some of our paralaminar cells. Like the labeled paralaminar cells, parafascicular cells had physiological features distinguishing them from typical thalamic neurons.
Since the early brain slice work of Jahnsen and Llinas (1984a,b) studies have shown that thalamocortical (TC) cells function in two different response modes, “burst” and “tonic”, that depend on the activation of a low threshold voltage sensitive calcium (Ca) conductance which generates a transient current known as IT. At a membrane potential of around −70 mV, synaptic inputs to TC cells activate the Ca conductance, resulting in a large, rapid depolarization that can elicit 3-5 high frequency (>200 Hz) spikes with variable first spike timing. If depolarized to −50 or −60 mV, the Ca conductance is inactivated and the same synaptic input will instead elicit a shorter latency, well timed spike. Thus, a TC cell’s firing mode alters its spike response to ascending synaptic messages (see Sherman, 2001) and subsequently the temporal information being sent to cortex.
Reports of recordings from cells in the auditory thalamus, or medial geniculate body, (MGB) (Hu, 1995; Peruzzi et al., 1997; Tennigkeit et al., 1997; Bartlett and Smith, 1999) confirmed that virtually all TC cells in the two main MGB subdivisions, the dorsal and ventral MGB (MGD and MGV), show membrane potential dependent burst and tonic modes. MGV and MGD are the two primary areas of the auditory thalamus. MGV is part of the tonotopically organized primary or “lemniscal” pathway while MGD is a component of the “extra- or nonlemniscal” division (Clerici and Coleman, 1990; Clerici et al., 1990; Winer et al., 1999). In rats, almost all neurons in these regions are thalamocortical projecting to intermediate layers of the auditory cortex (Roger and Arnault, 1989; Arnault and Roger, 1990; Clerici and Coleman, 1990; McMullen and de Venecia, 1993; Khazaria and Weinberg, 1994; Winer et al., 1999). In MGV, the TC cells have been called tufted or bushy based on highly intertwined dendritic trees. MGD is less homogeneous but many TC cells here are classified as either tufted or stellate.
A third subdivision of the auditory thalamic complex, the medial division (MGM), has been described. MGM is one of a group of nuclei referred to as paralaminar (Herkenham, 1980) that are situated medial and ventromedial to the MGB. This group also includes the posterior interlaminar nucleus (PIN), suprageniculate (SG) and peripeduncular (PP) nuclei (Winer and Morest, 1983; LeDoux et al., 1987; Winer and Larue, 1988). Several unique features have been described regarding these areas. First, anatomical studies showed that, compared to MGV and MGD, cells here have the least uniformity of soma size (Morest, 1964; Clerici and Coleman, 1990; Clerici et al., 1990; Winer et al., 1999). Some dendritic trees may have tufted or stellate morphology like MGV and MGD cells, but many others can show long sparsely branching dendrites. Second, the inputs to these areas are multimodal, arriving not only from the inferior colliculus (IC) but from visual, somatosensory and other centers as well (LeDoux et al., 1984; Peschanski, 1984; Hicks et al., 1986; Arnault and Roger, 1987; LeDoux et al., 1987; Bordi and LeDoux, 1994a,b; Benedek et al., 1997; Linke et al., 1999). Third, the outputs of cells here are diverse. Like MGV and MGD neurons, some cells in MGM and the other paralaminar nuclei are thalamocortical (Scheel, 1988; Roger and Arnault, 1989; Arnault and Roger, 1990; Clerici and Coleman, 1990; Brett et al., 1994; Linke, 1999; Winer et al., 1999; Doron and LeDoux, 2000) but, unlike MGV and MGD cells they often project to layer 1 (Niimi and Matsuoka, 1979; Mitani et al., 1984, 1987; Linke and Schwegler, 2000; Kimura et al., 2003). Jones (Jones, 1998a,b) speculated that the this layer 1 projection may engage multiple cortical areas at times when it is necessary to “bind” the many aspects of a sensory experience together. This projection has also been implicated in the modulation of the cortical high-frequency 40 Hz gamma oscillations (Barth and McDonald, 1996; Sukov and Barth, 2001) and the expression of long term cortical plasticity (Weinberger et al., 1995; Weinberger, 1998). Another major projection of paralaminar cells is to the amygdala. Work from LeDoux and others have shown this connection to be a vital part of the pathway responsible for the behavioral and autonomic responses elicited by the conditioned fear response (eg. LeDoux et al., 1986, 1988; LeDoux, 1995; LeDoux and Muller, 1997; Linke et al., 2000, 2004). Paralaminar cells can also project to the basal ganglia (Moriizumi and Hattori, 1992; Shammah-Lagnodo et al., 1996; LeDoux et al., 1985) and the inferior colliculus (Senatorov and Hu, 2002; Winer et al., 2002).
Despite the participation of cells in the MGM and other paralaminar nuclei in the important functions listed above and their unusual inputs and outputs, no one has taken advantage of the slice preparation to compare their physiological properties with those of their “standard” TC counterparts. We have made sharp electrode recordings from MGM cells and cells of the adjacent paralaminar nuclei using thalamic slices. We found that in vitro data from cells here reveal several unique and potentially important features that distinguish them from “standard” TC cells (see Sherman and Guillery, 2001).
All methods were approved by the University of Wisconsin Institutional Animal Care and Use Committee. Animals were maintained in an American Associations for Accreditation of Laboratory Animal Care-approved facility. Our experimental methods are similar to those described previously (Peruzzi et al., 1997; Bartlett and Smith, 1999, 2002). Brain slices from 3-6 week old Long-Evans hooded rats of either gender were used. Rats were given an anesthetic overdose then perfused transcardially with chilled, oxygenated sucrose saline (described below). The portion of the brain containing the thalamus was removed and 500 μm horizontal or coronal slices were taken through the MGB. These two slice planes were used so that we could stimulate two major inputs to the MGM, the IC inputs in horizontal slices and the superior collicular (SC) inputs in coronal slices. Details of these synaptic inputs will be reported in a subsequent paper. Slices were placed in a submersion-style holding chamber containing oxygenated artificial cerebrospinal fluid (ACSF) at room temperature. After equilibrating, one slice was transferred to the recording chamber and perfused with normal, oxygenated, ACSF at 33-34°C which contained the following (in mM): NaCl, 124; KCl, 5; KH2PO4, 1.2; CaCl2, 2.4; MgSO4, 1.3; NaHCO3, 26; and glucose, 10. The sucrose ACSF contained sucrose instead of NaCl (Aghajanian and Rasmussen, 1989). For low calcium saline, the calcium concentration was reduced from 2.4 mM to 0.4 mM. Nickel (Sigma-Aldrich Co.), 4-Aminopyridine (4AP, Sigma-Aldrich Co.) and apamin (Vector Labs) were all mixed in ACSF at the stated concentrations on the day of the experiment and bath applied. Recording began at least 30 minutes after the slices were placed in the recording chamber.
Intracellular recordings of responses to injected current were made with glass microelectrodes of 100-150 MΩ resistance when filled with a solution of 2M potassium acetate and 2% Neurobiotin (Vector Labs). Cells were considered viable if their initial resting potential was more negative than −50 mV and their action potentials overshot 0 mV. Intracellular current and voltage records were digitized with a Digidata 1322A (Axon Instruments) and saved using pClamp software for subsequent analysis.
During intracellular recording, Neurobiotin was injected into the recorded cell with a 0.3 to 0.5 nA current for 2-10 minutes. After the experiment, the slice was fixed in fresh 4% paraformaldehyde, cryoprotected, and 60-70 μm frozen sections were cut and collected in 0.1 M phosphate buffer, pH 7.4. The sections were then incubated in H202, rinsed in phosphate buffer then refrigerated and incubated overnight in the avidin-biotin-HRP complex (ABC Kit, Vector Labs). The following day, the sections were rinsed in phosphate buffer and the Neurobiotin reacted using the diaminobenzidine (DAB)-nickel/cobalt intensification method, mounted, counterstained with cresyl violet, and coverslipped.
To determine a cell’s location and make morphological measurements, camera lucida drawings were made. A low power, X40 drawing of the MGB and surrounding structures was made and the location of the labeled cell noted. The location of the cell body relative to the divisions of the rat MGB was determined by published reports of the differences in Nissl staining between MGM and SG (LeDoux et al., 1985) and by other landmarks in the vicinity that distinguish the location of PP and PIN. We also compared our sections/drawings with sections in the atlas of Paxinos and Watson (1986), illustrations in the cytoarchitectural study of Clerici and Coleman (1990) and the illustration of paralaminar subdivisions by LeDoux et al. (1985) and Doron and LeDoux (2000). A high power X1250 drawing of the cell body, dendritic tree and initial portion of the axon was also made.
Several measurements of the cell body and dendritic tree of our intracellularly labeled cells were made so they could be compared with similar data from tufted and stellate cells in the MGV and MGD that we previously reported (Bartlett and Smith, 1999). The measurements described here were taken from the high power drawing of the cell. Measurements were made only from cells where it was apparent that the majority of the dendrites were intact as indicated by the fact that they did not end abruptly at the top or bottom of the slice. Cell body areas were determined by drawing the outline of the cell body using Neurolucida software (Microbrightfield). Dendritic tree area was determined using the same software by drawing a line between the ends of adjacent dendrites and measuring the enclosed area. Another measure of the extent of the dendritic tree was dendritic projection distance where we measured and averaged the length of a cell’s three longest dendrites. As a measure of dendritic branching we used scaled circles of 50 and 100 μm, centered on the cell body, and counted the number of dendrites extending beyond these distances. This value was termed the number of dendritic intersections.
Several variables of the basic cell physiology were measured using Clampfit software (Axon Instruments) so the data could be compared with our previous data from cells in the MGV and MGD (Bartlett and Smith, 1999). Resting potential was measured as the difference between the voltage measured extracellularly and intracellularly during recording. Input resistance was measured from the maximum voltage deflection to a −0.1 nA current pulse. Spike amplitude was measured as the voltage difference between rest and spike peak. Spike threshold was measured as the least amount of positive current required to generate a spike. Spike duration was measured as the duration of the spike at half amplitude. The amplitude of the low threshold Ca conductance was made by measuring the area under the rebound burst that occurred when the cell was hyperpolarized and then returned to rest. The range of DC levels over which a burst occurred was measured by injecting depolarizing or hyperpolarizing current to move the membrane potential around rest and noting whether a −0.3 – 0.5 nA current pulse elicited a burst.
Statistical analysis was performed using Origin Pro7 (OriginLab Northampton, MA) and Microscoft Excel (Microscoft Corp., Seattle, WA). Data are presented as mean ± SD. A p value of < 0.05 or less was considered to represent a significant difference.
We labeled 91 cells in 67 slices taken from 53 rats of either sex that were medial or ventral to the MGV/MGD in one of the adjacent paralaminar nuclei. No differences were noted in the basic anatomy or physiology of these cells between subdivisions so they will be described as a single population. The terms MGM and paralaminar will be used interchangeably from this point on.
Clerici and Coleman (1990) and Winer et al. (1999) reported that rat MGM cell size is more diverse than MGV/MGD cell size with the largest of MGM cell bodies classified as magnocellular. This was also the case with our intracellularly labeled population (Fig. 1). Paralaminar cell bodies showed a larger range of sizes (from 118 to 451 μm2, mean = 259 ± 81.5) than MGV/MGD cells (from 122 to 226 μm2 mean = 175 ± 35.8). In addition, one way ANOVA indicated a significant difference between the means of MGM cell, MGV/MGD tufted cell and MGD stellate cell sizes (F(2,37) = 5.26, p = 0.01). Post-hoc tests for differences in means showed that although the mean for paralaminar cell size (259 ± 81.5 μm2) was greater than stellate (182 ± 44.5 μm2) and tufted (171 ± 33.1 μm2) significance was only reached for the paralaminar/tufted cell difference (t-test, P < 0.01).
Using Golgi staining methods Clerici and Coleman (1990) and Winer et al. (1999) reported on the dendritic tree configurations of rat MGM cells. Their sections were 100 -150 μm thick while our slices were 500 μm thick so we were able to get a more complete picture of the tree configuration.
We noted significant differences in dendritic tree features of cells in the MGM and other paralaminar nuclei when compared with those of tufted and stellate cells previously reported in MGV and MGD (Bartlett and Smith, 1999). For cells labeled in slices taken in either the horizontal and coronal plane, some of our labeled paralaminar cells had multipolar or stellate dendritic configurations. For these cells it was apparent that all or the large majority of the dendritic tree was present in the slice (Fig. 2). A number of the paralaminar cells labeled in coronal slices displayed elongated dendritic trees that extended for considerable distances primarily in the dorsoventral direction (Fig. 3). We saw cells labeled in our horizontal slices with several of their major dendrites cut off at the top and bottom of the slice and assume that these are probably the same elongate cells. Such cells from horizontal slices with incomplete dendritic trees were not included in the analysis. In the coronal plane these cells with elongated, oriented trees had a small number of long, sparsely branching dendrites that could extend for over 1 mm. The terminal branches often encroached on a paralaminar nucleus adjacent to the one containing the cell body (Figs. (Figs.4,4, ,5).5). Despite their dorsoventral extent, dendrites of these cells usually did not venture far enough laterally to significantly encroach on the MGV or MGD. There was no clear correlation between soma size and the area of the dendritic arbor, except that the smallest MGM cells (areas less than 175 μm2) had the smallest arbor areas, which were still larger than the MGV arbor areas. Thus, cell bodies that were intermediate in size (200-300 μm2) and cell bodies that could be classified as magnocellular (soma areas greater than 300 μm2) could have large elongate dendritic trees with large areas but could also have multipolar/stellate dendritic trees with smaller areas.
As might be expected from their extensive dendritic lengths, the dendritic tree areas of these elongate cells were significantly larger than those of tufted or stellate cells of the MGV and MGD. A one way ANOVA on the dendritic trees of MGM cells labeled in the coronal slices and stellate and tufted cell dendrites in the MGV and MGD showed significant differences in the means (F (2,34) = 15.88, P < 0.0001). Post-hoc t-tests for the differences in the means showed the dendritic tree areas of MGM cells measured in this plane (122,260 ± 64,701 μm2) were significantly larger than both the MGV/MGD tufted (area = 29,333 ± 5,777 μm2, p < 0.0001) and MGD stellate (62,830 ± 12,723 μm2, p < 0.05) dendritic trees. The larger examples of elongate dendritic trees could measure over 4 times that of the average MGD stellate cell and over 8-10 times that of the average MGV/MGD tufted cells. Likewise, the length of the three longest MGM dendrites of these elongate cells in the coronal plane could be over 500 μm. The results of a one way ANOVA showed significant differences between the means of the groups (F(2,29) = 27.0, p <0.0001). Post-hoc t-tests revealed that the MGM cell dendrites in the coronal plane were significantly longer (372 ± 111 μm) than both tufted cells in MGV/MGD (143 ± 19.2 μm, p < 0.0001) and the stellate cells in MGD (204.7 ± 19.2 μm, p < 0.002).
One simple way that we quantified and compared dendritic branching patterns between paralaminar and MGV/MGD neurons was to count the number of branches located at 50 μm or 100 μm radial to the cell body. One way ANOVA showed a significant difference in the means when comparing these features for MGM cells, tufted cells in MGV/MGD and stellate cells in MGD (at 50 μm, F(2,32) = 87.8, p < 0.0001: at 100 μm, F(2,32) = 38.2, p <0.0001). Post hoc analysis of the means revealed that the paralaminar elongate cells (as are illustrated in Fig. 3) had significantly fewer branches at both 50 (8 ± 3.73) and 100 μm (10.9 ± 4.4) than both the tufted (40.2 ± 9.2, p < 0.0001: 22.4 ± 7.9, p < 0.001) and stellate cells (29.7 ± 5.89, p < 0.001: 32.8 ± 5.4, p < 0.001) in MGV/MGD. ANOVA tests also revealed that the dendritic branching pattern of the stellate/multipolar population of MGM cells were significantly different than their counterpart stellate cells in the MGD at both 50 (F(2,27) = 63.1, p < 0.0001) and 100 μm (F(2,27) = 19.7, p < 0.0001). Post-hoc t-test analysis indicated that the MGM stellate/multipolar cells had significantly (p < 0.0001) fewer branches at 50 (8.75 ± 2.6) and 100 μm (12.9 ± 4.3) than their counterpart stellate cells in MGD (at 50μm, 29.7 ± 5.89; at 100 μm, 32.8 ± 5.4). However, despite having fewer branches, the dendritic tree areas of the MGM stellate cells (59,808 ± 26,895 μm2) did not differ significantly from the MGD stellate cells (62,830 ± 12,723 sq. μm2)
Another unique feature of some (20%) of the MGM/paralaminar cells was the presence of numerous dendritic spines along the length of the dendrite (Fig. 3, cell 2 inset; Fig. 4D). Although we (Bartlett and Smith, 1999) and others (Winer et al., 1999) have seen some dendritic specializations (knob-like bumps, short dendritic appendages, some spines) on MGV or MGD TC cells those on MGM cells could be more numerous and more consistently resembled spines (a distinct spine head and shaft). These spiny dendrites could belong to either the multipolar/stellate or the elongate cell types. In contrast, the remainder of the stellate and elongate cells had dendritic trees with very few spines (Fig. 3, cell 4 inset). These cells were darkly labeled so it was not simply a problem of being unable to visualize appendages.
Thus, the paralaminar cell bodies had a wider range of sizes and tended to be larger than tufted MGV/MGD cells. The dendritic trees of cells in paralaminar nuclei could be multipolar/stellate or could be elongate extending for very long distances primarily in the coronal plane. MGM elongate cell dendritic trees encompassed a significantly larger area than those of stellate cells in MGD or tufted cells in MGV/MGD. Stellate/multipolar cells in MGM had dendritic trees that were significantly larger than tufted MGV/MGD cells but no different than stellates in MGD. Both elongate and multipolar/stellate cell types in MGM tended to have fewer dendritic branches than the MGV and MGD cell types and could bear numerous dendritic spines. Cell body size was not a good predictor of dendritic size except that smaller cell bodies tended to have smaller dendritic trees. We only labeled a small number of cells in the paralaminar nuclei that might be considered tufted and these were usually close to the borders between the paralaminar and dorsal or ventral nuclei. They were not included in the analysis.
Another difference between TC cells in MGV and MGD and cells in adjacent paralaminar nuclei was the presence of local axon collaterals. In our previous study (Bartlett and Smith, 1999) none of the TC cells in MGV and MGD showed axon collaterals within or near the MGB. In contrast, some (10%) of our MGM cells displayed one or more collaterals (Figs. (Figs.44,,5)5) with en passant and en terminaux swellings (Figs 4C, arrows). This percentage is likely to be an underestimate because these collaterals could arise several hundred microns down the main axon (Fig. 4A) and, in slice preparations, main axons are often cut off close to their origin on the cell body or dendritic tree. The collaterals typically arose from the axons of cells that also had spiny dendritic trees. They could branch and display swellings near the parent cell (Fig. 5) or could come off the main axon at more distant sites to innervate other paralaminar regions (Fig. 4). None of the collaterals were seen entering MGV or MGD.
Stable intracellular recordings from cells in paralaminar nuclei could last for several hours and resting membrane potentials (−65 ± 6.7 mV) were not significantly different from those of cells in MGV and MGD that we previously reported (−62.2 ± 8.8 mV). Action potential amplitudes of paralaminar cells (72.1 ± 10.9 mV) were significantly larger than those of the MGV/MGD cell population (61.3 ± 10.3 mV, p < 0.0001) as were their input resistances (82.6 ± 49.6 mΩ vs. 55.2 ± 31 mΩ, p < 0.005).
The most striking response feature of many of the paralaminar cells was the absence or reduction of the low-threshold, voltage-sensitive calcium conductance that elicits the burst firing response which is so prominent in all thalamocortical cells in the MGV, MGD and elsewhere in the sensory thalamus. A typical bimodal spike response of an MGV or MGD cell to depolarizing current pulses at various membrane potentials is illustrated in Figure 6A. At depolarized levels (Fig. 6A, top trace) the cell fires in a sustained irregular fashion. At more negative levels the same depolarization activates the calcium burst over a large range of membrane potentials (Fig. 6A, 3rd −5th traces). Responses in Figure 6B-C show examples from 2 different paralaminar cells. For some paralaminar cells (25%) the cell fired in a regular, sustained fashion at all membrane potentials where the current pulse was suprathreshold (Fig. 6B, top 4 traces). At more hyperpolarized levels, these cells stopped firing with no indication of a calcium burst (Fig. 6B, 5th trace). Nor could these bursts be elicited at higher current levels at these membrane potentials (not shown). Other paralaminar cells (24%) showed persistent firing at depolarized levels, but with some spike frequency adaptation (Fig. 6C, top 4 traces). Again, these cells would show little or no indication of a burst at more hyperpolarized levels. Some of these cells with little or no burst response could fire at high sustained rates, sometimes over 400 Hz, when depolarized (Fig. 7A). Paralaminar cells like those shown in Figure 6B-C also did not display typical rebound burst firing after being hyperpolarized. Figure 7B illustrates the standard response of a cell in MGV or MGD to a hyperpolarizing current pulse as the membrane potential is moved around rest. At a fairly wide range of potentials, the hyperpolarization deinactivates the Ca conductance such that when the membrane is repolarized the activated Ca conductance elicits an offset burst. Figure 7C illustrates the response of a paralaminar neuron to a similar set of stimuli. No rebound burst is elicited at any membrane potential.
Other paralaminar cells (44%) did show what appeared to be a calcium burst that generated a series of sodium spikes. Figure 6D and E illustrate two examples. At depolarized levels these paralaminar cells fired in a regular fashion. As the cells were stepped to more hyperpolarized levels they could first show an onset response (Fig. 6D, third panel) then a burst (33%, Fig. 6D, fourth panel) or could show a burst followed by a pause and a resumption of sustained activity (11%, Fig. 6E, second panel) then a burst only (Fig. 6E, fourth panel). Rebound bursts could also be generated in these paralaminar cells (Fig. 7D). We used the area under the rebound bursts as a measure of the burst amplitude and compared these values for paralaminar cells that showed a burst with the TC cells in the MGV/MGD. Paralaminar cells with burst responses had significantly smaller bursts (593.9 ± 390 arbitrary units) than MGV/MGD cells (851.8 ± 301, p < 0.002). As might be expected, these smaller paralaminar calcium bursts generated a significantly lower spike rate (206.3 ± 103.2 Hz) than that seen in MGV/MGD cells (264.6 ± 64.2 Hz, p < 0.002).
Another unique feature of the response of many paralaminar cells was its action potential afterhyperpolarization (AHP). 72% of our MGM cells displayed spikes with biphasic AHPs (bAHPs) consisting of a fast and a slow component while the rest showed monophasic AHPs (mAHPs) like MGV/MGD cells. Both paralaminar cells with no apparent calcium burst and those with bursts could display action potentials with either biphasic or monophasic action potentials. Biphasic AHPs are never seen in recordings from MGV/MGD cells (Bartlett and Smith, 1999). The bottom panel in Figure 6B compares the action potential of the MGV TC cell whose responses are seen in Figure 6A (top spike) with one from the paralaminar cell seen in Figure 6B (bottom spike). The bottom panels of Figure 6C-E also illustrate spike waveforms of the other paralaminar cells in these columns. Unlike the MGV cell monophasic spike afterhyperpolarization, 3 of the 4 paralaminar cell spikes (B, D and E) show a slow component of the AHP.
We used the biphasic or monophasic nature of the action potential AHP to divide MGM cells into 2 groups. We do not know if this represents a true functional subdivision of cell types but differences in other features of their physiology led us to make this distinction. First, as described above, a comparison the rebound burst size for paralaminar cells that showed a calcium conductance (regardless of the nature of their spike AHP) with cells in the MGV/MGD demonstrated that the paralaminar cell calcium conductance was significantly smaller. However, if the paralaminar cells were divided into those with mAHPs or bAHPs the burst size generated by the cells with mAHPs was comparable to the burst size seen in MGV and MGD cells while the bursts of cells with bAHP was significantly smaller. ANOVA comparisons of the area under the rebound bursts generated by paralaminar cells with mAHPs, by those with bAHPs and by cells in the MGV/MGD revealed a significant difference (F(2, 93) = 10.9, p < 0.0001). Post-hoc t-tests illustrated that the paralaminar cells with mAHPs had rebound bursts whose amplitudes (823 ± 344 arbitrary units) were not significantly different from those of cells in MGV/MGD (859 ± 301) while both these groups had significantly larger bursts than the paralaminar cells with bAHPs (511 ± 374, p < 0.003 and p < 0.0001). A second feature distinguishing the paralaminar cells with biphasic vs monophasic AHPs was the cells input resistance. As we described above, the paralaminar cell input resistances were significantly different from the MGV/MGD cell population but it was due to the high input resistances of the paralaminar cells with bAHPs. ANOVA evaluation of the means of MGV/MGD, paralaminar bAHP and paralaminar mAHP cell input resistances showed that there were significant differences (F(2,120) = 6.56, p < 0.002) but post-hoc t-tests showed that the paralaminar bAHP cell input resistance (88.4 ± 52.9 mΩ) was significantly different than that of the MGV/MGD cells (55.2 ± 31 mΩ, p < 0.01) while that of the paralaminar mAHP cell was not. A final feature of the paralaminar cell that led us to make this distinction was action potential duration. ANOVA comparisons followed by post-hoc t-tests showed no significant differences in the mean values of the MGV/MGD, paralaminar mAHP or paralaminar bAHP cell action potential half widths but those of the paralaminar bAHPs spanned a much larger range (0.16 to 1.48 ms) than the other groups (0.34 to 0.9). Spikes with very short (< 0.3 ms) half widths exhibited by several paralaminar cells with bAHPs are comparable to those reported for interneurons in the thalamus (Pape and McCormick, 1995).
In summary paralaminar cells can show little or no calcium burst and a sustained regular or adaptive firing pattern over a large voltage range. This firing can reach very high rates. Their resting potentials are comparable to that seen in MGV/MGD cells but their action potentials are larger. Their action potentials can show biphasic or monophasic afterhyperpolarizations. Both paralaminar cells that do and do not show bursting can have either bAHPs or mAHPs. For those that burst the size of the burst is smaller in those cells with bAHPs. Other unique features of the paralaminar bAHP cells are the higher input resistance, the wide range of action potential durations and the rate of spiking generated by the burst.
Figures Figures88 and and99 verify that these bursts in some MGM cells are in fact calcium mediated. Figure 8A shows that reduction of the calcium concentration (lower panel) reduces the amplitude of the bursts (arrows) of a typical TC cell in MGV to depolarizing and hyperpolarizing pulses (n=7 cells). Likewise, a similar reduction in calcium levels reduces the depolarization and hyperpolarization induced bursts (Fig. 8B, arrows) in an MGM cell. For those paralaminar cells with no noticeable burst the sustained firing would persist when the calcium level was lowered (N=5 cells) but at an altered rate (Fig. 8C). Nickel has been shown to block/reduce the low threshold Ca conductance in thalamocortical cells (Hernandez-Cruz and Pape, 1989). Figure 9A illustrates that nickel reduces both the depolarization induced and the rebound Ca burst in a TC cell in MGD. Figure 9B and D are representative responses from two cells in the paralaminar nuclei and in both cases the depolarization induced and rebound bursts are significantly reduced in the presence of nickel (N=7). Paralaminar cells with no apparent Ca burst continued to fire in a sustained fashion but with a slightly altered rate (N=4, Fig. 9C).
We also wanted to get some idea of the nature of the paralaminar cell spike afterhyperpolarization. At millimolar concentrations, 4-aminopyridine (4AP) blocks fast, voltage sensitive K channels (IA) used to repolarize action potentials in TC cells in other areas of the thalamus (Huguenard et al., 1991). Bath application of 4AP also affected the repolarization of the paralaminar cell action potentials (Fig. 10, n=5 cells). This resulted in the abolition of the fast component of the AHP and a widening of the action potential (Fig. 10C). MGM cells also appear to have a potassium conductance not found in the MGV/MGD cell. Apamin blocks the small-conductance Ca2+-activated K+ channel known as the SK channel (see Castle et al., 1989). This drug had no apparent effect on the spike waveform and response features of the MGB TC cells (Fig. 11A) but decreased the fast and slow components of the paralaminar spike AHP and subsequently increased the firing rate while maintaining some spike frequency adaptation (Fig. 11B, n=4 cells). In a paralaminar cell with a burst component the burst response was prolonged by apamin (n=2, Fig. 11C).
Intralaminar nuclei are nuclear structures that lie within the internal medullary lamina of the thalamus. Several previously reported features of cells in the parafasicular nucleus (pf), a nucleus in the posterior intralaminar nuclear group, are unique when compared to other areas of the thalamus but are similar to those of paralaminar cells (see our Discussion). This led us to examine the basic response properties of cells in the intralaminar nuclei to see if they too showed the unusual physiological characteristics in vitro that we observed in paralaminar cells. We recorded from and labeled a small sample (N =10) of cells in coronal or horizontal slices containing intralaminar nuclei including pf. Figure 12 illustrates an examples of labeled cells in the pf nucleus. As we have shown for many of our labeled paralaminar cells, these cells showed long sparsely branching dendrites (Fig. 12A, top left) that may or may not display dendritic spines. Their axons could also give rise to local axon collaterals (Fig. 12A, bottom left). Figure 12B-D illustrates the basic response features of these cells. As described for cells in the paralaminar nuclei, pf cells 1) can have action potentials with biphasic AHPs 2) can fire in a sustained regular fashion that can reach high rates with increasing current strength (Fig. 12B) but can also show spike frequency adaptation (not shown) 3) can have a reduced calcium burst (not shown) or can show very little or no calcium burst to depolarizing (Fig. 12C) or hyperpolarizing (Fig. 12D) current pulses regardless of the membrane potential.
Our labeled cell population confirms previous reports that there is diversity in the pool of neurons in the MGM and other paralaminar nuclei adjacent to the MGB. Cells here had multipolar/stellate or elongate dendritic tree configurations and the dendritic branches had or lacked spines. The axons of some cells, usually those with dendritic spines, gave off local axon collaterals but not all cells with dendritic spines did. Axons of cells labeled in slices can be cut off fairly close to their origin on the cell body or dendritic tree so it is difficult to know the actual percentage of cells with local collaterals. We also occasionally labeled cells that could be classified as tufted but the number that could be unequivocally localized to a paralaminar nucleus was low and they were not included in this report.
An interesting feature of some of our labeled paralaminar cells, perhaps related to their input selection, was their long dendritic branches that could extend into neighboring paralaminar nuclei. Paralaminar nuclei receive inputs from several auditory and non-auditory areas. These projections do no terminate uniformly over the entire region. For example LeDoux et al. (1987) noted that auditory inputs from the inferior colliculus and somatosensory inputs from spinal cord both innervate the paralaminar nuclei but did not have a large area of overlap. Likewise, Linke (1999) has shown that although the paralaminar nuclei are all innervated by inputs from both the inferior and superior colliculi, each paralaminar nucleus has its own unique input pattern. Thus cells with more confined dendrite trees (eg. stellate/multipolar cells) might be more restricted in what inputs could have an influence while elongate paralaminar cells, like the ones illustrated in Figures Figures3,3, ,44 and and5,5, would not be subject to such restrictions.
A second unique dendritic feature was the presence of a fairly high concentration of dendritic spines on some of our paralaminar cells but not on others. Spines have been associated with certain forms of synaptic plasticity (Carlisle and Kennedy, 2005) and it is interesting that early reports (Ryugo and Weinberger, 1978; Gerren and Weinberger, 1983) indicated that the MGM is a unique subdivision of MGB in that it is capable of showing plasticity during classical conditioning paradigms. Both studies indicated that synapses of IC inputs onto MGM cells but not onto MGV and MGD cells could show rapid persistent changes. More recent evidence again has implicated MGM as a part of a learning circuit. McEchron et al. (1996) used a classical conditioning paradigm pairing auditory stimuli with a corneal air puff. After training, stimulation of the brachium of the IC but not superior colliculus generated enhanced MGM responses for over an hour. Thus, synaptic inputs to the MGM that are activated by the path carrying auditory information to MGM neurons increased in strength as the result of associative conditioning with an acoustic stimulus. Future studies can test the hypothesis that spiny, but not aspiny, paralaminar cells can show lasting changes in synaptic strength.
Another potentially interesting feature of some of the paralaminar cells was the presence of axon collaterals that branched locally in the paralaminar nucleus containing the cell’s soma or in a nearby paralaminar nucleus (Figs. (Figs.55 and and6).6). These terminals are presumably excitatory. We have preliminary physiological results showing that electrical stimulation of inputs to the paralaminar nuclei from IC, SC and auditory cortex generate early direct synaptic events followed by longer latency indirect excitatory postsynaptic potentials (EPSPs). These later events may be generated by inputs from other activated MGM cells with local collaterals. Similar stimulation of IC and auditory cortex inputs to cells in MGD or MGV elicit direct synaptic events but not longer latency EPSPs which correlates with the fact that the axons of stellate and tufted cells never give off local collaterals and that the MGM cell collaterals do not appear to stray into MGD or MGV. Thus spike activity of a cell within a paralaminar nucleus can have a direct excitatory effect on neighboring paralaminar cells.
Numerous retrograde labeling studies (eg. Ottersen and Ben-Ari, 1979; LeDoux et al., 1985, 1987, 1990; Winer and Larue, 1987, Scheel, 1988, Roger and Arnault, 1989; Clerici and Coleman, 1990; Moriizumi amd Hattori, 1992; Brett et al., 1994; Shammah-Lagnado et al., 1996; Doron and LeDoux, 1999; Linke, 1999; Linke et al., 1999; Winer et al., 1999, 2002; Kimura et al., 2003) have shown that cells in the paralaminar nuclei project to several different sites including the cerebral cortex, amygdala, basal ganglia, and IC. Unfortunately because of the incomplete filling of cellular processes none of these reports described features of the dendritic trees or local axonal branching pattern of the backfilled paralaminar cells so we do not know whether any of the cell types reported here (stellate/elongate, spines/no spines, axon collaterals/no collaterals) project to a unique target.
We have shown that cells in the MGM and adjacent paralaminar nuclei have unusual basic response features when compared to reports of cells elsewhere in the thalamus. The most striking feature was the reduction or apparent lack of a low threshold voltage sensitive Ca conductance in many cells. The vast majority of reports describing recordings from thalamic cells include this conductance as a regular feature of all thalamocortical neurons and a great deal of significance has been assigned to its influence on the output of the cell. The only region of the thalamus where an apparent lack of this Ca conductance has been reported is the GABAergic thalamic reticular nucleus (TRN). In vivo intracellular recordings from cells here (Contreras et al., 1992) were classified as type I or II based on the ability (type 1) or lack thereof (type II) to generate a high frequency burst of spikes. Subsequent in vitro recordings (Brunton and Charpak, 1997) showed a lack or reduction of the low threshold Ca conductance in the type II TRN neurons and, like our paralaminar cells, sustained regular firing to suprathreshold currents at all membrane potentials. The only other report of cells in the thalamus whose basic intrinsic physiology differs somewhat from the norm is a report from Li et al. (2003) on cells in the rat lateral posterior nucleus (LPN) a “higher order” visual thalamic nucleus. Unlike our paralaminar cells, LPN cells showed the typical thalamocortical cell morphology and all displayed the low threshold calcium conductance. However unlike the standard TC cell in “first order thalamic nuclei, some cells had a “regular spiking” mode and short duration action potentials with biphasic AHPs similar to those seen in many of our paralaminar cells. They also noted some cells that displayed a “clustered spiking” defined as a firing mode where depolarizing current elicited epochs of high frequency bursts of firing not caused by the low threshold Ca conductance. We have also seen cases of this “clustered spiking” response type in the paralaminar nuclei but do not yet have a large enough sample of labeled, positively identified cells to report any details with confidence.
Intralaminar nuclei are nuclear structures that lie within the internal medullary lamina of the thalamus. It has been suggested that cells in one of the posterior intralaminar nuclei known as the parafascicular nucleus (pf) supplies the striatum with information about behaviorally significant sensory events which it uses to help modulate the animal’s motor response (e.g. Van der Werf et al., 2002; Smith et al., 2004). We recorded from and labeled a small population of pf cells. This was motivated by previous reports describing features of cells in the rat pf and paralaminar nuclei that were similar. First, cells in both the pf and paralaminar nuclei can project to the striatum (Herkeham, 1980; LeDoux et al., 1985; Feger et al., 1994; Deschenes et al., 1996; Vercelli et al., 2003; Castle et al. 2005). LeDoux et al. (1985) noted that retrogradely labeled cells from the caudate-putamen formed a continuous stream of cells running from paralaminar nuclei rostromedially into the posterior intralaminar nuclei. Second, in vivo recordings from pf and paralaminar nuclei showed that cells can be multimodal responding to auditory and somatosensory stimuli with responses that rapidly adapt to repeated auditory stimuli (Bordi and LeDoux, 1994a,b; Matsumoto et al., 2001). Third, work by Scheibel and Scheibel (1967) and Deschenes et al. (1996) described a set of “non-specific” neurons in the intralaminar nuclei that were “reticular-like” with long, scarcely ramifying spine bearing dendrites and axons that gave off locally ramifying axon collaterals. Deschênes et al. (1996) speculated that these reticular-like intralaminar cells with an anatomy that differed so drastically from the stereotypical TC cell in relay nuclei could also differ in their physiology. All of these observations motivated us to record from pf cells in vitro where we noted similarities in our pf and paralaminar cells. Like paralaminar cells, pf cells could lack or have a reduced low threshold calcium conductance and could show regular sustained firing and action potentials with biphasic AHPs. In agreement with previous descriptions they also resembled the elongate paralaminar cells with their long sparcely branching dendrites which may or may not be spiny and their axons that may display local collateral branches. Based on these similarities we propose that at least some of our elongate paralaminar cells could be thalamostriatal or at least integrate and transmit synaptic inputs in a similar fashion.
To begin to try to interpret why bursts are reduced or absent in paralaminar cell in might be helpful to first understand their function in areas of the thalamus where they are ubiquitous. In the visual thalamus, thalamocortical neuron bursts have been postulated to be involved in the generation of synchronized sleep rhythms (Livingstone and Hubel, 1981; Steriade et al., 1993). More recent findings that cells in the visual thalamus of the alert, awake cat also response in a burst fashion (Guido and Weyand, 1995; Reinagel et al., 1999; Weyland et al., 2001; Lesica and Stanley, 2004) have dampened enthusiasm for this idea. Current thinking is that cells in burst mode are better able to detect near threshold visual stimuli (see Sherman, 2001). Another observation that may be relevant to thalamic burst function in the visual system was made by Swadlow and Gusev (2003) who noted that a spike burst was more efficient at eliciting a spike in a cortical neuron receiving this input.
There is less information regarding the function of the bursting response in the auditory thalamus in vivo. Yu et al. (2004) recorded intracellularly from cells in the guinea pig MGB. Using current injection they changed the membrane potential of cells and showed that the driven response to an auditory stimulus could change from burst to tonic firing modes. The burst firing response occurred at hyperpolarized levels and was almost certainly due to the low threshold calcium conductance. Unfortunately they did not report the MGN location of the cells that illustrated this membrane potential-dependent response change. Other extracellular recording studies (He and Hu, 2002; Massaux et al., 2004) found that the proportion of burst firing varied depending on the animal’s state and the electrode location and that bursts were detected in awake, anesthetized or sleeping animals, during spontaneous or sound driven responses and in all areas of the MGB (including MGM). Massaux (Massaux et al., 2004) noted that in awake guinea pigs the burst component of responses to tones occurred preferentially at or around the cells best frequency and that response latency and variability of the response latency were both reduced for the bursts. This would indicate that the burst response might give a more selective and accurate representation of sound frequency to post synaptic targets. All of this would provide very indirect evidence that paralaminar cells lacking bursts might be less likely to respond synchronously as a group or respond to near threshold stimuli and less likely to have narrow more refined response areas or to activate their postsynaptic target with a high degree of reliability.
Another way of attempting to determine why bursts are reduced or absent in MGM cells would be to try to relate how such cells might be better able to perform their proposed function. One set of paralaminar cells, those that project to the amygdala, are part of the pathway responsible for the behavioral and autonomic events elicited by the conditioned fear response (LeDoux et al., 1988; LeDoux, 1995; LeDoux and Muller, 1997). LeDoux and colleagues noted that within this pathway LTP-like associative processes occur during fear conditioning, and may underlie the long-term associative plasticity that constitutes memory of the conditioning experience. Lesion, tracer and recording experiments have shown that paralaminar nuclei are the interface between anterolateral amygdala (AL) and the ascending auditory system (LeDoux et al., 1986; Clugnet et al., 1990; LeDoux et al., 1990a,b). Further experiments documented that amygdalar responses to auditory inputs is enhanced by electrical stimulation specifically applied to the medial MGB or by prior conditioning of the auditory stimulus (Clugnet and LeDoux, 1990; Rogan and LeDoux, 1995; Quirk et al., 1995). It should be noted that the electrical stimuli used by Clugnet and LeDoux to induce LTP were trains of 30 pulses at 400 Hz. Cells in a calcium burst mode could not follow such a rapid stimulus given the long refractory period (around 200 ms) of the underlying conductance (Jahnsen and Llinas, 1984a). Cells with no burst and an ability to respond steadily at high rates (see our Figure 7A) could follow such a stimulus.
Another function of the paralaminar nuclei is to activate 40 Hz oscillations in the auditory cortex. Human studies (Llinas and Ribary, 1993; Joliot et al., 1994) showed that 40 Hz oscillations could reflect cognitive processing of auditory stimuli and the “temporal binding” of sensory stimuli into a single experience. Temporal binding refers to the conscious binding together of sensory events that occur during a particular time as a single experience and the 40 Hz oscillation is “required” for this to occur. Barth and McDonald (1996) reported that gamma oscillations occurred spontaneously or could be evoked in auditory cortex by auditory stimuli. They further showed that auditory thalamus could modulate these oscillations. Electrically stimulating of MGV or MGD inhibited the 40 Hz cortical response while stimulation of the medial aspect of the auditory thalamus elicited it. The finding suggests a functional role for MGM in cognition. Again, in these experiments the electrical stimulation used to evoke the cortical 40 Hz oscillations was at a high rate (500 ms duration at 500 Hz) which could not be followed cells in bursting mode but might be by cells capable of high sustained firing.
We have shown that some paralaminar cells possess the apamin sensitive SK type calcium-activated K channel while TC cells in the MGV and MGD do not. We also showed that the SK channel is controlling some aspects of the firing properties of the paralaminar cells. Blocking the channel had little or no effect on the MGV/MGD spike response (see Fig. 11A) but prolonged the spike response of paralaminar cells that did possess a Ca burst (Fig. 11B) and increased firing frequency of a paralaminar cell that did not show a calcium burst (Fig. 11C). One of the SK channel proteins is highly expressed in the thalamic reticular nucleus (Sailer et al., 2002) where channel blockade by apamin affected neuronal excitability (Bal and McCormick, 1993, Debarbieux et al., 1998) so both TRN and paralaminar cells seem to possess this channel. Another observation on the TRN SK channel is that it was affected by the intravenous anesthetic propofol (Ying and Goldstein, 2005). Propofol also blocks auditory evoked focal gamma band oscillations (40 Hz oscillatory electrical activity) in auditory cortex and stimulation of either the TRN (Macdonald et al. 1998) or the medial division of the MGB (Barth and MacDonald 1996) elicits this 40 Hz cortical oscillation. Perhaps the auditory TRN, its connection to the MGM and the MGM projection to cortex are important components of the 40 Hz activation circuitry and the propofol effect on the SK channel at one or both locations explains its effect on the oscillation. Recently the SK channel has also been implicated in long term potentiation (LTP). Faber et al. (2005) showed that activation of glutamatergic synapses in the amygdala caused calcium influx through NMDA channels that activated SK channels and shunted the epsp. Apamin blockade of this channel allowed synaptic events to summate and generate an enhanced LTP. It is interesting to note that at least some MGM cells express the SK channel and that this is the only part of the MGB where potentiation of synaptic events has been reported ((Ryugo and Weinberger, 1978, Gerren and Weinberger, 1983; McEchron et al., 1996). Thus, just as low-voltage Ca channels can act as “sensory gates” in first order TC neurons, the SK channels in some paralaminar cells may act as a “learning gate”.
In conclusion some cells in MGM and surrounding paralaminar nuclei have been shown to possess unique features which distinguish them from the majority of other cells in the dorsal thalamus but link them with cells in intralaminar thalamic nuclei (and the thalamic reticular nucleus). It will be of great interest to determine which of the various cells in the paralaminar nuclei surrounding the medial geniculate nuclei project to each of their target structures and how these unique anatomical and physiological features that we have described here aid them in the completion of their tasks.
Supported by NIH Grant R01 DC006212 to Philip Smith