© 2017 Nature America, Inc., part of Springer Nature. All rights reserved.
news and views existence of two distinct AT1a circuits that drive water or sodium intake. Matsuda et al.6 demonstrated that one AT1a neuron population receives inhibitory drive from cholecystokinin-sensitive interneurons, projects to the organum vasculosum of the lamina terminalis and is necessary and sufficient for thirst. A separate AT1a population in the SFO is inhibited by local sodium-channelexpressing neurons, projects to the ventral bed nucleus of the stria terminalis and is necessary and sufficient for sodium appetite (Fig. 1b). This study defines the anatomical organization, physiological properties and molecular mechanisms of AT1a neuron populations in the SFO, substantially broadening our knowledge of circuit connectivity and effectively dissociating neural circuits that promote water versus sodium intake. These formative studies define independent circuits that are required for the full expression of sodium appetite. The circuits described are distinct not only with respect to the molecular phenotypes of the respective neural populations but also with regard to anatomical location. Jarvie and Palmiter5 point toward functional projections residing primarily in brainstem regions, whereas
Matsuda et al.6 outline circuits with nodes in the forebrain. Together, these studies point to the redundant nature of the circuitry that drives sodium intake, perhaps akin to the redundant circuitry mediating energy balance control14,15. Determining the interaction between these distinct circuit nodes will be an important step forward in understanding how the brain integrates signals of need to drive sodium intake. Do several distinct, non-overlapping circuits exist to ensure adequate sodium intake? Alternatively, do these circuits converge on a master regulator of fluid homeostasis or interact with a distributed and elaborate network that regulates osmolarity? Further, what behavioral and motivational mechanisms are recruited by activity in these neurons, and are they similar to other biological survival drives? Of course, osmoregulation is a complex physiological process that likely requires the coordination of need sensing with a motivated response in order to resolve the appropriate need, perhaps indicating the necessity of multiple circuits that influence discrete and identifiable aspects of behavior. Since a central goal of neuroscience is to understand how the brain orchestrates behavioral responses to
physiological needs, the identification of neural circuits that coordinate sodium appetite represents an important step toward unraveling this complex and multifaceted problem. COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests. 1. Geerling, J.C. & Loewy, A.D. Exp. Physiol. 93, 177–209 (2008). 2. Richter, C.F. Am. J. Physiol. 115, 155–161 (1936). 3. McEwen, B.S., Lambdin, L.T., Rainbow, T.C. & De Nicola, A.F. Neuroendocrinology 43, 38–43 (1986). 4. Buggy, J. & Fisher, A.E. Nature 250, 733–735 (1974). 5. Jarvie, B.C. & Palmiter, R.D. Nat. Neurosci. 20, 167–169 (2017). 6. Matsuda, T. et al. Nat. Neurosci. 20, 230–241 (2017). 7. Formenti, S. et al. Am. J. Physiol. Regul. Integr. Comp. Physiol. 304, R252–R259 (2013). 8. Geerling, J.C., Engeland, W.C., Kawata, M. & Loewy, A.D. J. Neurosci. 26, 411–417 (2006). 9. Geerling, J.C. & Loewy, A.D. J. Comp. Neurol. 497, 223–250 (2006). 10. Oka, Y., Ye, M. & Zuker, C.S. Nature 520, 349–352 (2015). 11. Johnson, R.F., Beltz, T.G., Thunhorst, R.L. & Johnson, A.K. Am. J. Physiol. Regul. Integr. Comp. Physiol. 285, R394–R403 (2003). 12. Nation, H.L., Nicoleau, M., Kinsman, B.J., Browning, K.N. & Stocker, S.D. J. Neurophysiol. 115, 3123–3129 (2016). 13. Betley, J.N. et al. Nature 521, 180–185 (2015). 14. Betley, J.N., Cao, Z.F., Ritola, K.D. & Sternson, S.M. Cell 155, 1337–1350 (2013). 15. Grill, H.J. Obesity (Silver Spring) 14 (Suppl. 5), 216S–221S (2006).
New building blocks for navigation Jeffrey S Taube Many spatial correlates have been identified that form the neural basis for navigation. Two studies have now uncovered a new cell type: bidirectional cells, which fire when the head is pointing in one of two opposing directions. Place cells, head direction cells, grid cells, border cells, angular head velocity cells, speed cells, pitch cells and various combinations of these parameters have been identified over the past five decades1,2. Since the first report of place cells in 1971, these cells have formed the building blocks for understanding how the brain forms an awareness of perceived orientation, which allows organisms to navigate successfully. This issue of Nature Neuroscience adds two new spatial cell types to this mix: axis-tuned cells3 and bidirectional head direction (HD) cells4. Both of these new cell types can be considered variants of the well-studied HD cell, first discovered by James Ranck in 1984 (ref. 1).
Jeffrey S. Taube is in the Department of Psychological & Brain Sciences, Dartmouth College, Hanover, New Hampshire, USA. e-mail:
[email protected]
HD cells discharge as a function of the animal’s perceived heading in the horizontal plane, independently of the animal’s location and behavior (Fig. 1a). Each HD cell is tuned to a single direction, referred to as the cell’s preferred firing direction, and the cell’s firing rate decreases linearly as the head deviates from this direction. A population of HD cells is similar to a collection of compasses with each one tuned to a different direction. HD cells were originally identified in the dorsal presubiculum, often referred to as the postsubiculum. The presubiculum is one of the three primary areas of the subicular complex, which also includes the subiculum and parasubiculum. Each of these three areas has its own connectivity and contains both afferent and efferent connections with the hippocampus and entorhinal cortex. Thus, the subicular complex is an important node integrating information with the hippocampus. HD cells are found in many brain areas throughout
nature neuroscience volume 20 | number 2 | February 2017
the limbic system, and several brainstem sites are responsible for generating the signal1. Self-movement cues are integrated in these brainstem areas, and the processed HD signal propagates rostrally through a network that includes the lateral mammillary nuclei, anterodorsal thalamus, postsubiculum and entorhinal cortex. Visual landmark information is integrated into this circuit via direct projections from visual cortex to the postsubiculum, but the retrosplenial cortex (RSC) also plays a role—a point that we will return to below. Until now, all studies on HD cells had found them to be unidirectional, with each cell tuned to only one direction. The two studies in this issue report cells that are tuned to two different head directions, even in the same environment. In the first study, Olson, Tongprasearth and Nitz3 recorded from cells in the subiculum proper while rats traversed a complex maze (Fig. 1b). The researchers found cells that fired in two different directions, always ~180° apart 131
a
Classic HD cell
c
Classic HD cell
b
d
Axis-tuned cell
Local HD cell
e
Bidirectional HD cell
Debbie Maizels / Springer Nature
© 2017 Nature America, Inc., part of Springer Nature. All rights reserved.
news and views
Figure 1 Axis-tuned, unidirectional and bidirectional HD cells. (a) Dual-chamber apparatus. A rat travels from the familiar cylinder to the novel rectangular chamber via the attached alleyway. A ‘classic’ HD cell fires in all three chambers in the direction of the arrows, with similar peak firing rates in each chamber6. (b) The complex maze used by Olson et al.3. Arrows indicate the location and direction of travel where the axis-tuned cell fired. Peak firing rates were similar in each branch. (c–e) Apparatus used by Jacob et al.4. (c) A classic HD cell was dependent on the global reference frame. (d) A second type of HD cell firing was associated with the local cues in both chambers. (e) A third type, bidirectional HD cells, fired in two directions 180° apart in both chambers and was associated with both the global and local reference frames. The sizes of the arrows in c–e correspond to the cell’s peak firing rate. Note the difference in peak firing rates between the two chambers for the bidirectional cell. Green lines indicate visual cues (local landmarks) attached to the inside wall of the enclosure. Lemon and vanilla symbols denote the scents in each chamber.
from each other. They called these cells ‘axistuned’ because they fired consistently whenever the animal moved along a path that contained one of the two directional headings. Like that of classic HD cells, axis-tuned cell firing was independent of the animal’s location. Different cells were tuned to different directions, and the distribution of orientations across all cells appeared to be uniform across 360°. The orientations were not necessarily tied to the perpendicular orientation of the maze walls. This pattern of firing persisted in darkness and remained attached to the global reference frame, as rotation of the apparatus within the recording room did not change the firing orientation of these cells with respect to the room. That is, the cell’s firing orientation shifted to different tracks within the maze when it was rotated. This travel axis pattern of firing was not observed when the rat was recorded in a cylindrical open-field environment. In this case, responses generally resembled patterns of activity more closely aligned to other established spatial cell types: place cells, border cells or HD cells. Several questions come to mind. Is bodyaxis orientation or HD the critical factor in 132
determining cell firing? Further, because the animal had to move at >3 cm per s for the data to be included in the analyses, how do the cells respond when the animal moves slowly or is motionless? Information addressing this point would help determine whether the cells are conveying information about trajectory or more about body orientation. Because the axis-tuned cell firing is dependent on the presence of linear tracks, it would be interesting to determine how these cells fire in an open field constrained by rectilinear walls. Would the rectangle’s wall constrain the distribution of firing orientations to those aligned with the walls? Similarly, would travel-axis firing be present in a lesscomplex but still linear environment, such as a single linear track? What function might these axis-tuned cells serve? They could be useful for encoding the general trajectory of travel during navigation, although why it is operative in two opposing directions is not clear. As the authors point out, human subjects tend to orient themselves with natural boundaries in the environment, such as a street or a rectangular building. Axistuned cells could act in this process.
In the second study, Jacob et al.4 report a second type of HD cell variant that also fires in two opposing directions. It is not well understood where and how the brain represents different reference frames simultaneously. For example, when you are in your office, you have a sense of your orientation within your office (local reference frame). At the same time you are aware of your orientation with respect to your campus or town (global reference frame). Where in the brain these separate reference frames are integrated with one another is still unclear, but the findings of Jacobs et al.4 suggest that the RSC could be a good candidate. Different hippocampal place cell populations in CA1, but not CA3, are capable of encoding either proximal or distal cues simultaneously, even when the spatial information from them is in conflict5. In contrast, HD cells respond coherently to one reference frame or another but never both simultaneously. For example, as an animal moves from a familiar environment to a novel environment that is either a short distance away (Fig. 1a)6 or in another room7, HD cells maintain a preferred direction that is similar between the two environments, indicating that the animal is accurately updating its orientation as it moves between chambers. Even when the HD cell population encodes either the local or global environment on a three-dimensional spatial task, there is coherence across the entire population8. Finally, when HD cells are monitored during periods when place cells are undergoing split responses between two reference frames, thalamic HD cells always remain coherent to one reference frame9. All spatial cell subtypes are controlled by landmark cues. Thus, when a salient visual landmark, whether local or distal, is rotated in a cue-controlled environment, spatial cells rotate their representation by a similar amount. HD cells in these cue rotation experiments have been recorded throughout the traditional HD network, including the RSC1,10. Given this background, the findings by Jacobs et al.4 are all the more surprising. Jacobs et al.4 recorded from HD cells in RSC while rats traveled between two rectangular chambers that were attached along their long sides, with a doorway that could be opened or closed (Fig. 1c). Each chamber had a similar polarizing visual cue attached to one of the short walls, such that the chambers were mirror images of one another. However, the two compartments could be distinguished by scent. Jacobs et al.4 report three types of HD cell responses in RSC. The first are classic HD cells, which are seen throughout the limbic system and fire in relation to the global environment. Thus, when the rat went from one chamber
volume 20 | number 2 | February 2017 nature neuroscience
© 2017 Nature America, Inc., part of Springer Nature. All rights reserved.
news and views to the next the cell’s preferred direction was maintained across both chambers (Fig. 1c), as in the HD cell responses found in the earlier studies (Fig. 1a)6,7. The second cell type fired in response to the local environment: the cell’s preferred direction shifted 180° when the rat entered the adjacent chamber (Fig. 1d). This response indicated that these cells were tuned to a landmark in the local reference frame, despite the short travel distance to the next chamber. While not particularly unusual to find HD cells tied to local cues8, it is surprising to find these cells coexisting with HD cells that remain tied to the global environment. Thus, two different spatial representations are being encoded simultaneously across the two cell populations. These findings are reminiscent of the different place cell responses described above in the hippocampus9 but contrast with the coherent HD cell responses observed when an animal travels between different enclosures6,7. What was truly remarkable was the third type of HD cell response. These cells fired in two directions 180° apart from one another within either chamber (Fig. 1e). Following rotation of the apparatus, the two firing directions rotated an equal amount in both chambers, indicating that the firing remained associated to each chamber’s local cues. Interestingly, these cells did not encode the two directions equally: the cell’s peak firing rate for one direction was larger than the rate for the other direction; this higher firing rate reversed directions between compartments. Thus, these cells appeared to be encoding the reference frame of both chambers simultaneously— precisely the type of representation one needs to encode multiple environments. While the first and second cell types were found in all areas of the RSC (granular and dysgranular), this within-compartment bidirectional cell type was located only in the dysgranular RSC. When the authors repeated their experiments on thalamic and postsubicular HD cells, they found no bid irectional responses. The dysgranular RSC has extensive reciprocal connections with visual cortex, postsubiculum and granular
RSC11, areas that contain HD cells and/or process visual information. Thus, this area is ideally situated to integrate visual landmark information with the animal’s perceived spatial orientation. Indeed, hints of these bidirectional responses were evident in an earlier RSC cue rotation study, which reported cell responses reflecting the encoding of both local and global reference frames10. These findings have implications for how the brain may build a spatial representation that encodes both the animal’s current local location and its orientation to the larger surrounding environment. That RSC acts in this process is consistent with human studies, where subjects with RSC lesions have topographic disorientation and are unable to accurately remember the spatial relationships between two distant locations12. Functional imaging studies have shown RSC activation when situating a scene within the broader spatial environment13. Together, these results suggest that the RSC processes and compares information across different reference frames to maintain awareness of the animal’s orientation and is ready to be called upon when needed to navigate to a distant goal. While it is straightforward to understand how bidirectional firing may be constructed from inputs from RSC HD cell types 1 and 2, it will be important to determine where this bidirectional information projects to. Given that juxtacellular recording and labeling is now feasible, garnering this information should be possible. It would also be interesting to know how bidirectional cells respond when an animal is engaged in one environmental setting but needs to maintain an awareness of how it is oriented relative to a larger global environment governed by different landmarks. While axis-tuned cells in the subiculum and RSC bidirectional cells both show bidirectional tuning, the two cell types differ in several ways. First, axis-tuned cells only responded with bidirectionality on the complex linear track, not in an open field, where they were unidirectional, whereas RSC cells were bidirectional in an open-field rectangular environment but lost all directional tuning in an open field
nature neuroscience volume 20 | number 2 | February 2017
cylinder. Second, although both cell types contained preferred directions ~180° apart, the peak firing rates for each direction were different for RSC cells, while in axis-tuned cells the peak firing rates were equivalent for both directions. Third, RSC bidirectional cells shifted their representations following rotation of the chambers, but axis-tuned cells were tied to the global environment and did not shift their preferred directions following rotation of the maze. Both cell types share features with cells recorded in parietal cortex, where cells may encode multiple reference frames simultaneously14. Like axis-tuned cells, other parietal cells encode individual segments or trajectories along a route15. One final question for both cell types is whether their firing patterns are evident on the rat’s first experience with an environment or develop over time. Theoretically, spatial cells can be tuned to the local or global environment. Jacob et al.4 show how neurons in RSC can encode both reference frames simultaneously, while Olson, Tongprasearth and Nitz3 demonstrate how cells encoding the rat’s axis of travel can help guide its sense of orientation, keeping track of its alignment with the overall environment. COMPETING FINANCIAL INTERESTS The author declares no competing financial interests. 1. Taube, J.S. Annu. Rev. Neurosci. 30, 181–207 (2007). 2. Moser, E.I., Kropff, E. & Moser, M.B. Annu. Rev. Neurosci. 31, 69–89 (2008). 3. Olson, J.M., Tongprasearth, K. & Nitz, D.A. Nat. Neurosci. 20, 170–172 (2017). 4. Jacob, P.-Y. et al. Nat. Neurosci 20, 173–175 (2017). 5. Knierim, J.J. J. Neurosci. 22, 6254–6264 (2002). 6. Taube, J.S. & Burton, H.L. J. Neurophysiol. 74, 1953–1971 (1995). 7. Yoder, R.M. et al. J. Neurophysiol. 105, 2989–3001 (2011). 8. Taube, J.S., Wang, S.S., Kim, S.Y. & Frohardt, R.J. J. Neurophysiol. 109, 873–888 (2013). 9. Yoganarasimha, D., Yu, X. & Knierim, J.J. J. Neurosci. 26, 622–631 (2006). 10. Chen, L.L., Lin, L.H., Barnes, C.A. & McNaughton, B.L. Exp. Brain Res. 101, 24–34 (1994). 11. van Groen, T. & Wyss, J.M. J. Comp. Neurol. 315, 200–216 (1992). 12. Takahashi, N., Kawamura, M., Shiota, J., Kasahata, N. & Hirayama, K. Neurology 49, 464–469 (1997). 13. Epstein, R.A. Trends Cogn. Sci. 12, 388–396 (2008). 14. Nitz, D.A. Nat. Neurosci. 15, 1365–1367 (2012). 15. Nitz, D.A. Neuron 49, 747–756 (2006).
133
