The supramammillary area: its organization, functions and relationship to the hippocampus

Abstract

The supramammillary area of the hypothalamus, although small in size, can have profound modulatory effects on the hippocampal formation and related temporal cortex. It can control hippocampal plasticity and also has recently been shown to contain cells that determine the frequency of hippocampal rhythmical slow activity (theta rhythm). We review here its organization and anatomical connections providing an atlas and a new nomenclature. We then review its functions particularly in relation to its links with the hippocampus. Much of its control of behaviour and its differential activation by specific classes of stimuli is consistent with a tight relationship with the hippocampus. However, its ascending connections involve not only caudal areas of the cortex with close links to the hippocampus but also reciprocal connections with more rostral areas such as the infralimbic and anterior cingulate cortices. These latter areas appear to be the most rostral part of a network that, via the medial septum, hippocampus and lateral septum, is topographically mapped into the hypothalamus. The supramammillary area is thus diffusely connected with areas that control emotion and cognition and receives more topographically specific return information from areas that control cognition while also receiving ascending information from brain stem areas involved in emotion. We suggest that it is a key part of a network that recursively transforms information to achieve integration of cognitive and emotional aspects of goal-directed behaviour.

Introduction

The supramammillary area (SuM) is a relatively thin, chemically distinct, layer of cells overlying the mammillary bodies in the hypothalamus. While relatively few in number, the cells of this area have recently become of interest as they appear to exert significant modulatory control over the hippocampus and, directly and via the hippocampus, substantial portions of the telencephalon. Thus, like the raphe serotonergic system and the coerulear noradrenergic system, SuM may have functional importance out of all proportion to its size. Its connection with the hippocampus is made early in primate development and it has been suggested that “the early ingrowth of the excitatory SuM–hippocampal system in human and non-human primates may contribute to the prenatal activity-dependent development of the hippocampal formation”, perhaps as a result of its control of hippocampal theta activity (Berger et al., 2001).

It contrasts with the mammillary bodies in being much more a source of input to archicortex (particularly hippocampus) than a receptor of output from it. It also contrasts, as far as can be told, with many other hypothalamic areas in being a specific controller of hippocampal theta activity. It is, then, located in an area of the brain usually associated with emotion and motivation and not only connects with areas of the brain thought to be involved in predominantly cognitive functions but controls a pattern of brain activity (theta) that has been postulated to be important for control of cognition and particularly memory. We argue, on several different grounds, therefore, that its modulation of telencephalic activity is fundamental to cognitive–emotional interactions.

This claim is subject to two caveats. First, we are claiming only a modulatory role for SuM. It is likely to pace information around circuits but not to encode and transform that information itself. It is likely to determine whether other inputs produce plasticity in target structures but not to induce plasticity solely through its own inputs. Second, as will become clear, it is likely to be only the first discovered of a number of (perhaps topographically organized) structures that function in this specific way.

To both make our claim, and flesh out our caveats, we analyse the anatomy, physiology and contributions to behavioural function of SuM. We focus, in particular, on the extent to which SuM may discharge behaviourally important functions through its afferent and efferent connections with the hippocampal formation. Our conclusions may seem to present SuM as fundamentally controlling hippocampal function—but not only must SuM be only one of several structures that control the hippocampus (each, we argue under different behavioural or cognitive circumstances) but also the hippocampus must be only one of many widely separated structures modulated by SuM. Some of the parallels we identify between SuM and hippocampal function are likely, therefore, to be functions that depend on the joint activity of many “limbic” structures.

Our specific link with the hippocampus is made by the fact that SuM has recently been shown to contain cells that can control plasticity in the hippocampus via monosynaptic input. It also contains other cells that control the frequency of the rhythmic phasic firing of hippocampal cells (theta activity) via a relay in the medial septum (MS). These two types of interaction with the hippocampus are functionally separate and anatomically distinct. But they are of such a kind (controlling general tone and controlling the timing of firing of large populations of cells) that it is likely that SuM neurones play a much more important modulatory role in hippocampal function than their paucity and distance from the hippocampus might suggest. SuM also has similar extensive connections with many other structures and may, then, similarly modulate many areas of the forebrain.

The hippocampus has been postulated to contribute to cognitive functions such as spatial mapping (O’Keefe and Nadel, 1978) and relational memory (Cohen and Eichenbaum, 1993) and emotional functions (Papez, 1937), particularly behavioural inhibition (Gray, 1982, Gray and McNaughton, 2000). All components of the hippocampal formation, including the entorhinal and posterior cingulate cortex as well as hippocampus proper (Gray and McNaughton, 2000, Leung and Borst, 1987), show theta activity. This can be viewed as regular phasic firing controlled by pacemaker impulses relayed to the hippocampal formation from the MS (Brucke et al., 1959, Gogolák et al., 1968, Petsche et al., 1962, Tömböl and Petsche, 1969). Thus, the time at which the population of cells in the hippocampus fires is strongly influenced by afferent theta activity that produces regular changes in membrane potential (Bland, 1986). This control could be viewed as septal input actively eliciting theta activity in the hippocampus. However, it is probably better viewed as a temporal gating of activity by phasic inhibitory input (Leung, 1998). Thus when cells are permitted to fire is determined largely by phasic release from inhibition (Smythe et al., 1992), but it is also influenced by other factors (O’Keefe and Recce, 1993, Bose and Recce, 2001). By contrast, which cells fire during the appointed time is determined by other specifically patterned inputs. Theta activity is likely to play a significant specific role in the cognitive and/or emotional functions of the hippocampal formation (O’Keefe and Nadel, 1978, Gray, 1982, Miller, 1991, Gray and McNaughton, 2000) as well, perhaps, as a more general role in the control of sensori-motor integration (Bland and Oddie, 2002, Oddie and Bland, 1998) or sensory inhibition (Sainsbury, 1998). The postulated power of the control of hippocampus (and other areas like entorhinal, posterior cingulate and possibly perirhinal cortex) by SuM rests in the possibility that a failure of appropriate timing of computations carried out between these structures could result in serious degradation or even total loss of the critical information (Miller, 1991).

The frequency of hippocampal theta activity is controlled in part by medially placed SuM neurons (Kirk, 1993, Kirk, 1997, Kirk, 1998, Kirk and McNaughton, 1991, Kirk and McNaughton, 1993). However, lateral rather than medial areas of SuM project preferentially to the MS (Vertes, 1992, Vertes and Kocsis, 1997, Vertes and McKenna, 2000). There is also monosynaptic input from more laterally placed SuM neurons (but not more medially placed ones) to the hippocampus. The ascending pathways that control theta and other hippocampal activity appear, then, to be topographically organized and to be both direct and relayed (as with the MS). The hippocampus, in turn, has topographically organised, relayed, projections back to the hypothalamus including SuM (Risold and Swanson, 1996). The details of the system, then, need to be understood if its function (and by implication that of much of the forebrain) is to be fully understood.

SuM, at least on occasion, controls the frequency of theta activity. Theta activity, in turn, is most associated with the hippocampus. SuM is, therefore, likely to be important for the behavioural functions of the hippocampus. As noted, different parts of SuM may also contribute differentially to hippocampal function. However, as detailed below, SuM also has extensive connections with many other areas. We would expect it (through similar computational mechanisms) to contribute to a variety of different functions. This paper, therefore, reviews the neuroanatomy, neurophysiology and molecular biology of SuM, and provides a preliminary assessment of the functions of SuM, the relation of these functions to its known parts, and the extent to which those functions are discharged by interaction with the hippocampus and particularly by the control of theta activity by SuM.

Abbreviations used in the remainder of the text are defined in Table 1 to allow ready reference.