What is the Thalamus?

Published

The thalamus is a centrally-located brain structure that controls the flow of all information to the cortex. It is a paired structure joined at the midline and sitting very near the center of the brain. In the human, each half is roughly the size and shape of a walnut. There are two major components.

First is the dorsal thalamus, which is comprised of roughly 15 nuclei with relay cells that project to the cerebral cortex. (By “cortex” in this account, we mean “neocortex.”) Second is the ventral thalamus, the major portion of which is the thalamic reticular nucleus, which sits like a shield flush against the lateral surface of the dorsal thalamus; reticular cells are GABAergic and project into the dorsal thalamus to inhibit relay cells. The figure schematically shows the major thalamic nuclei.

The other cellular component of thalamus, in addition to relay and reticular cells, is interneurons, which are also GABAergic, sit amongst the relay cells, and inhibit them. Generally, the relay cell to interneuron ratio is between 3 and 4 to one. An exception is found the mouse and rat, in which interneurons are essentially missing from all thalamic nuclei except the lateral geniculate nucleus (Arcelli et al, 1997).

Most of the relay nuclei topographically innervate the middle layers of cortex, but a few along the midline and extended between other nuclei project rather diffusely to upper cortical layers, including layer 1; rather little is known of these latter, diffusely-projecting nuclei, and they are not further considered in this account (for further details, see Sherman and Guillery, 2006; Jones, 2006).

The remaining thalamic relay nuclei each innervates one or a small number of cortical areas. Indeed, all information reaching cortex passes through thalamus, and thus thalamus sits in a strategic position for brain processing.

The major role of thalamus is to gate and otherwise modulate the flow of information to cortex. For example, visual information from the retina is not sent directly to visual cortex but instead is relayed through the lateral geniculate nucleus of the thalamus.

In the macaque monkey, there are roughly 1×106 geniculate relay cells (Williams and Rakic, 1988), but in primary visual cortex there are roughly 1.6×108 neurons (O’Kusky and Colonnier, 1982), which is typical of thalamocortical relationships. Thus thalamus represents the final bottleneck of information flow before it gets into cortex.

In other words, to modify information flow for processes of attention and other behavioral requirements, it is more efficient to do this at the level of thalamus before it reaches cortex. While there is still much to learn about the cell and circuit properties of thalamus in this role, what we do know supports this general view of thalamic function. For further details of thalamus, see Jones (2006) and Sherman and Guillery (2006).

This entry will focus on the role of the thalamus in normal, behaving animals. It should be noted that there is a vast and interesting literature on thalamic function during sleep and certain pathological conditions, such as epilepsy, but this is beyond the scope of this account, and the reader is instead directed to other reviews (Steriade and Llinás, 1988; Steriade et al., 1993; McCormick and Bal, 1997).

Cell Properties

Thalamic relays cells, like neurons everywhere in the central nervous system, display a wide range of ionic conductances in their soma and dendritic membranes; most of these conductances depend for their expression on voltage being held above or below a certain level for a sufficient amount of time.

Examples exist of such relatively common conductances for Na+ (e.g., the action potential), K+ (e.g., the A current, or IA), Ca2+ (e.g., the T current or IT) and general cation currents (e.g., the h current or Ih). Details of these ionic conductances can be found elsewhere (Sherman and Guillery, 2006), but the point here is that their expression greatly alters the way in which relay cells respond to their inputs and thereby relay information to cortex.

Of particular interest is the Ca2+ conductance based on T type Ca2+ channels that gives rise to IT (Jahnsen and Llinas, 1984; Sherman, 2001; Sherman and Guillery, 2006). IT is an inward current that produces a transient depolarization of the cell.

While T channels are common to neurons everywhere, providing for depolarization, what sets the thalamic relay cell somewhat apart is that it exhibits a sufficiently dense distribution of T channels in its soma and dendrites that initiation of IT generally leads to an all-or- none Ca2+ spike that propagates throughout the soma and dendritic arbor. Every relay cell in every thalamic nucleus of every species studied so far shows this behavior, and this is seen only rarely in other neuronal types in the central nervous system.

The voltage and time dependency for the T channels (Jahnsen and Llinas, 1984; Sherman, 2001; Sherman and Guillery, 2006). These channels can be both activated and inactivated by voltage, and thus have two voltage gates. At rest (e.g., roughly -70 mV or so), the inactivation gate is open but the activation gate is closed, and so the channel is both de-inactivated and deactivated.

If the cell is then depolarized to threshold for the activation gate (to roughly -65 mV), the activation gate pops open, leading to the inward IT, which, in turn, activates the upswing of the all-or-none Ca2+ spike; in this state, the T channel is activated and de-inactivated (panel B). After roughly 100 msec of depolarization, the T channel inactivates (panel C), and this, along with activation of a slower series of K+ conductances, leads to repolarization of the neuron.

Here, the T channel remains inactivated (panel D) for another 100 msec or so, after which time the original state of panel A is returned. The two gates of the T channel have opposite voltage dependencies, but while the activation gate responds quickly to voltage change, the inactivation gate is slower, requiring roughly 100 msec of polarization change to open or close.

Actually, inactivation and de-inactivation are complex functions of voltage and time so that the more the cell is depolarized, the faster IT inactivates, and the more the cell is hyperpolarized, the faster IT de-inactivates. This requirement of roughly 100 msec of hyperpolarization to de-inactivate the channel provides a refractory period that limits low threshold Ca2+ spiking to ≤ 10 Hz.

Note that this behavior of the T channel is qualitatively identical to that of the Na+ channel underlying conventional action potentials, with several quantitative differences. These are:

  • the inactivation kinetics of the T channel are much slower

  • the regime of the voltage dependency of the T channel is 5-10 more hyperpolarized

  • T channels are not found in the axons.

This last point means that it is only conventional action potentials that convey the information relayed to cortex.

At an initial membrane potential of -59 mV, IT is inactivated and thus the cell responds in tonic mode (A). Here, the response to a depolarizing 0.3 nA current injection is a steady stream of unitary action potentials. At an initial membrane potential of -70 mV, IT is de-inactivated and thus the cell responds in burst mode (B).

Now, the very same current injection activates the low threshold Ca2+ spike, which in turn activates, in this case, a burst of 8 conventional action potentials. C: Initial response of cell in A,B to various levels of current injection from different initial membrane potentials.

At levels that inactivate IT and produce tonic firing (-47 mV and -59 mV), a fairly linear relationship ensues. At levels that de-inactivate IT and produce burst firing (-77 mV and -83 mV), a nonlinear relationship in the form of a step function is seen at low stimulus intensities.

The real significance of T channel behavior is shown in the figure. That is, the low threshold Ca2+ spike is sufficiently large (typically 25-40 mV) to activate a transient, high frequency burst of conventional action potentials. If the relay cell is relatively depolarized for ≥100 msec, T channels are inactivated and play no role in neuronal responses. Now, a depolarizing current pulse (or EPSP) directly activates a series of unitary action potentials for as long as the input remains suprathreshold; this is the tonic mode of firing.

However, if the same cell is sufficiently hyperpolarized for ≥100 msec, T channels are de-inactivated and primed for action. Now, the very same depolarizing pulse (or EPSP) activates the low threshold Ca2+ spike, which in turn, activates a burst of 2-10 action potentials; this is the burst mode of firing.

Note that the very same input (or EPSP) results in a very different message relayed to cortex, depending on the recent voltage history of the relay cell. Note also that, because of the slow kinetics of the inactivation gate, a polarization change must be sustained for roughly ≥100 msec to change the inactivation state and affect a switch between firing modes.

Some of the consequences of the tonic versus burst firing modes (Sherman, 1996, 2001; Sherman and Guillery, 2006). Because tonic firing represents a direct link between an input depolarization or EPSP, the larger the depolarization, the greater the response. There is thus a fairly linear input/output relationship during tonic firing.

In contrast, burst firing represents an indirect link between the input EPSP and action potential generation, the link being the low threshold Ca2+ spike; because this is all-or-none, meaning that a larger EPSP does not evoke a larger low threshold Ca2+ spike, this input/output relationship is decidedly nonlinear, approximating a step function.

The linear input/output relationship of tonic firing offers an obvious advantage: for cortex to faithfully reconstruct the information relayed, the linear relay of tonic firing is far superior to the nonlinearity of burst firing. What, then is the advantage of burst firing? There are two (Sherman, 1996, 2001; Sherman and Guillery, 2006):

  • The patterns of action potentials renders bursts more detectable in the spike train, meaning that an evoked burst is more likely to be detected by cortex than is an evoked train of tonic action potentials.

  • Bursts much more powerfully activate cortex than do tonic spikes (Swadlow and Gusev, 2001; Swadlow et al., 2002).

The reason for the greater activation of bursts has to do with the nature of the thalamocortical synapse, which is a depressing synapse. The sustained firing during tonic mode means that the synapse is perpetually depressed, whereas bursts occur only after a requisite silent period of ≥100 msec (due to the hyperpolarization needed to de-inactivate the T channels), meaning that all synaptic depression is relieved.

This has led to the following hypothesis (Sherman, 1996, 2001; Sherman and Guillery, 2006). As a result of better detectability and cortical activation, bursts could serve as a “wake-up call” to cortex that something has changed in the environment, and this is very useful for unattended scenes (e.g., during drowsiness or other episodes of reduced attention). Once attention is properly directed, tonic firing is employed to maximize fidelity of information relayed by thalamus.

It is worth noting that other thalamic cells also show similar T channel activity and a burst/tonic duality in firing modes. This is clearly the case for cells of the thalamic reticular nucleus, although the details of T channel voltage and time dependencies vary slightly from those of relay cells {Huguenard and Prince, 1992). The situation with thalamic interneurons is more complicated, because it is often difficult to demonstrate bursting in these cells. However, an analysis by Pape et al. (1994) suggests that these cells do have an IT that is usually obscured by the presence of a voltage gated K+ current.

Thalamic Circuitry

The lateral geniculate nucleus serves as a useful model for thalamus, because the basic circuit features found there are found throughout thalamus. There are, of course, some differences among nuclei and species, but we shall start with a concentration on common features. The figure summarizes the basic circuit.

Briefly, geniculate relay cells are innervated by many inputs in addition to the retinal inputs that represent the main input to be relayed. These nonretinal inputs, which account for approximately 90%-95% of synaptic inputs onto the relay cells, arise from local GABAergic neurons, reticular cells, and interneurons plus two main sources of extrinsic input, which are a feedback projection from layer 6 of cortex and an ascending projection from various scattered cell groups in the brainstem reticular formation. These cortical and brainstem inputs also innervate reticular cells and interneurons.

All of these inputs to relay cells operate via conventional chemical synapses, meaning that they affect relay cells by releasing neurotransmitters that activate specific postsynaptic receptors. These come in two flavors: ionotropic and metabotropic.

There are many differences between them (for details, see Nicoll et al., 1990; Mott and Lewis, 1994; Pin and Duvoisin, 1995; Conn and Pin, 1997), but an important one for the purposes of this subject is time course. Activation of ionotropic receptors leads to postsynaptic potentials that last for tens of msec, whereas postsynaptic potentials related to metabotropic receptors last for hundreds of msec to several seconds.

Ionotropic receptors include

  • AMPA receptors for glutamate

  • GABAA receptors (obviously for GABA)

  • Nicotinic receptors for acetylcholine

Metabotropic receptors include

  • multiple metabotropic glutamate receptors for glutamate

  • GABAB receptors (again, for GABA)

  • multiple muscarinic receptors for acetylcholine.

Note the important distinction here: retinal inputs activate only ionotropic receptors, whereas all of the nonretinal inputs activate metabotropic receptors in addition to ionotropic receptors.

Recall that control of the T channel requires membrane potential changes lasting ≥100 msec. Therefore, the nonretinal inputs that activate metabotropic receptors are ideally designed to control T channels and thus the operative burst or tonic firing mode.

For example, activation of direct cortical or brainstem inputs leads to a prolonged EPSP that inactivates T channels and promotes tonic firing.

In contrast, activation of GABAergic inputs produces long duration IPSPs that de-inactivate the T channels, promoting burst firing. Since the cortical and brainstem inputs innervate both relay cells and GABAergic circuits, these nonretinal inputs are able to exert powerful control over the firing mode of relay cells (for details of this, see Sherman, 2001; Sherman and Guillery, 2006).

Drivers and Modulators

In terms of circuitry of the lateral geniculate nucleus, retinal inputs are distinguished by a long list of features (Sherman and Guillery, 2006). Among them are thick axons producing very large terminals and synapses.

These synapses are quite powerful, activate only ionotropic receptors, and show activity dependent depression (i.e., paired-pulse depression associated with a depressing synapse). Where information is available, nonretinal inputs differ on all of these criteria.

Finally, the basic information in the form of receptive field properties is provided by retinal inputs, since these inputs have the same center/surround receptive fields as do relay cell receptive fields, which is clearly not the case of the cortical or brainstem inputs.

This has led to the idea that these inputs can be functionally divided into drivers and modulators (Sherman and Guillery, 1998, 2006). The drivers represent the main information to be relayed, and the modulators modify the thalamocortical relay. One such modification is the burst/tonic transition.

Surprisingly, in terms of numbers of synapses on geniculate relay cells, retinal inputs provide only about 5% (Van Horn et al., 2000). The remaining 90%-95% of synaptic inputs are roughly equally divided among local GABAergic inputs (i.e., thalamic reticular neurons and interneurons), layer 6 cortical feedback, and brainstem inputs.

Apparently, relatively few but very powerful synapses are needed to get the basic visual information to relay cells, but having many, weak modulatory synapses that can be combined in numerous ways allows for many forms of subtle and not so subtle modulation.

Where sufficient information is available, such as the primary somatosensory and auditory relays and the pulvinar (reviewed in Sherman and Guillery, 2006), these different properties of drivers and modulators are evident, including the small number of driver synapses.

A very important point worth emphasizing is that this recognition of driver and modulator inputs means that, at least in thalamus, not every input to relay cells can be treated equally as if some sort of anatomical democracy prevailed. Instead, it is important to recognized which of the many inputs is the driver.

As an example, if one had only anatomical data showing that roughly 1/3 of all inputs to relay cells derived from the brainstem with only 5% from retina, one might well conclude that the lateral geniculate nucleus relayed brainstem information with retinal input providing some obscure, relatively unimportant input. Only knowledge of functional properties and this concept of drivers and modulators prevents such an embarrassing conclusion.

First and Higher Order Thalamic Relays

Obviously, much of the function of a thalamic relay is defined by its driver input. Thus geniculate relay cells are largely defined by their retinal inputs, those of the ventral posterior nucleus, by their medial lemniscal inputs, etc. Thus defining the driver input to a thalamic relay is an important step. The result of appreciating driver inputs to much of thalamus has led to the concept of first order and higher order thalamic relays.

First order relays are the ones we know best, like the lateral geniculate and ventral posterior nuclei; they represent the first relay to cortex of a particular type of subcortical information, like vision or somesthesia. Higher order relays instead relay information already in cortex via driver input from layer 5 of one cortical area to middle layers of another cortical area (Guillery, 1995; Sherman and Guillery, 2006).

All thalamic relays receive a feedback from layer 6 of cortex (as well as local GABAergic and brainstem inputs; not all are shown in the figure for simplicity), but the higher order relays receive an additional input from layer 5 of cortex, and this is in a feedforward configuration.

The layer 5 input is the driver input and is equivalent to the retinal input to the lateral geniculate nucleus with regard to many morphological and functional features (reviewed in Sherman and Guillery, 2006).

If we consider the primary sensory systems, they have both first and higher order thalamic components. Thus,

  • for the visual system, the lateral geniculate nucleus is first order, and the pulvinar is higher order

  • for the somatosensory system, the ventral posterior nucleus is first order, and the posterior nucleus is higher order

  • for the auditory system, the ventral division of the medial geniculate nucleus is first order, and the dorsal (and perhaps medial) division is higher order

One messy proviso is that parts of these higher order nuclei may contain first order components as well, since the pulvinar contains some input from the midbrain that appears to be driver (Kelly et al., 2003), the posterior nucleus receives some spinothalamic input that may be driver, and the dorsal part of the medial geniculate nucleus receives input from the inferior colliculus that may be driver.

These details need further study. Non-sensory parts of thalamus can also be divided into first and higher order, and it is now clear that most of thalamus is comprised of higher order relays (Sherman and Guillery, 2006).

The main point, which is illustrated in figure 5, is that the higher order thalamic relays represent a key link in cortico-thalamo-cortical routes of information processing, routes that until recently were undetected.

That is, just as retinal information must be relayed by thalamus before reaching cortex, these information routes must also have a thalamic relay. What, then, of the huge networks of direct corticocortical connections?

One unlikely possibility that is mentioned to crystallize the issue is that all of these corticocortical pathways are modulators, and that the major information that passes between cortical areas is relayed by higher order thalamic nuclei.

It seems more likely that information is processed in parallel between direct corticocortical and indirect cortico-thalamo-cortical pathways, but even here much work needs to be done to establish which subset of corticocortical pathways are truly driver in properties and also to understand why two such different routes of information processing may exist. For further speculation of this issue, see Sherman and Guillery (2006).

Conclusions

A consideration of the complexity of thalamic cell and circuit properties puts a lie to the old notion that thalamus represents a simple, machine-like relay of information to cortex. We can now be certain of two major concepts to replace this.

The first is that the thalamus represents a last bottleneck of information flow, providing a convenient substrate to influence that flow. This is achieved by the many modulatory pathways that innervate relay cells to influence relay function in numerous ways. One detailed above is the burst/tonic transition in the firing mode of relay cells, but this is just the tip of the iceberg. We need much more information about the many ways thalamic circuitry controls information flow to cortex.

The second point is that the role of thalamus is not limited to getting information to cortex in the first place, which is the role of first order relays, but also continues to function in the higher order cortico-thalamo-cortical pathways, thereby providing an essential, ongoing function for cortical processing.

This dramatically alters long-standing views of cortical processing, and we need to know much more about the different roles of cortico-thalamo-cortical versus direct corticocortical pathways in cortical functioning.

References

Arcelli P., Frassoni C., Regondi M.C., De Biasi S., and Spreafico R. (1997) GABAergic neurons in mammalian thalamus: A marker of thalamic complexity? Brain Res. Bull. 42, 27-37.

Conn PJ, Pin JP (1997) Pharmacology and functions of metabotropic glutamate receptors. Annual Review of Pharmacology & Toxicology 37: 205-237.

Guillery RW (1995) Anatomical evidence concerning the role of the thalamus in corticocortical communication: A brief review. J Anat 187: 583-592.

Huguenard J.R. and Prince D.A. (1992) A novel T-type current underlies prolonged Ca2+-dependent burst firing in GABAergic neurons of rat thalamic reticular nucleus. J. Neurosci. 12, 3804-3817.

Jahnsen H, Llin s R (1984) Electrophysiological properties of guinea-pig thalamic neurones: an in vitro study. J Physiol (Lond ) 349: 205-226.

Jones EG (2006) The Thalamus. Cambridge, U.K.: Cambridge University Press.

Kelly LR, Li J, Carden WB, Bickford ME (2003) Ultrastructure and synaptic targets of tectothalamic terminals in the cat lateral posterior nucleus. J Comp Neurol 464: 472-486.

McCormick D.A. and Bal T. (1997) Sleep and arousal: Thalamocortical mechanisms. Annual Review of Neuroscience 20, 185-215.

Mott DD, Lewis DV (1994) The pharmacology and function of central GABAB receptors. International Review of Neurobiology 36: 97-223.

Nicoll RA, Malenka RC, Kauer JA (1990) Functional comparison of neurotransmitter receptor subtypes in mammalian central nervous system. Physiol Rev 70: 513-565.

O’Kusky JR, Colonnier M (1982) A laminar analysis of the number of neurons, glia, and synapses in the visual cortex (area 17) of adult macaque monkeys. J Comp Neurol 210: 278-290.

Pape H.-C., Budde T., Mager R., and Kisvárday Z.F. (1994) Prevention of Ca2+-mediated action potentials in GABAergic local circuit neurones of rat thalamus by a transient K+ current. J. Physiol. (Lond. ) 478, 403-422.

Pin JP, Duvoisin R (1995) The metabotropic glutamate receptors: structure and functions. Neuropharmacol 34: 1-26.

Sherman SM (1996) Dual response modes in lateral geniculate neurons: mechanisms and functions. Visual Neurosci 13: 205-213.

Sherman SM (2001) Tonic and burst firing: dual modes of thalamocortical relay. Trends in Neurosci 24: 122-126.

Sherman SM, Guillery RW (1998) On the actions that one nerve cell can have on another: Distinguishing “drivers” from “modulators”. Proc Natl Acad Sci USA 95: 7121-7126.

Sherman SM, Guillery RW (2006) Exploring the Thalamus and Its Role in Cortical Function. Cambridge, MA: MIT Press.

Steriade M. and Llinás R. (1988) The functional states of the thalamus and the associated neuronal interplay. Physiol. Rev. 68, 649-742.

Steriade M., McCormick D.A., and Sejnowski T.J. (1993) Thalamocortical oscillations in the sleeping and aroused brain. Science. 262, 679-685.

Swadlow HA, Gusev AG (2001) The impact of ‘bursting’ thalamic impulses at a neocortical synapse. Nat Neurosci 4: 402-408.

Swadlow HA, Gusev AG, Bezdudnaya T (2002) Activation of a cortical column by a thalamocortical impulse. J Neurosci 22: 7766-7773.

Van Horn SC, Erisir A, Sherman SM (2000) The relative distribution of synapses in the A-laminae of the lateral geniculate nucleus of the cat. J Comp Neurol 416: 509-520.

Williams RW, Rakic P (1988) Elimination of neurons from the rhesus monkey’s lateral geniculate nucleus during development. J Comp Neurol 272: 424-436.

“Thalamus” ( S. Murray Sherman ) / CC BY-NC-SA 3.0