Edited by: Trygve B. Leergaard, University of Oslo, Norway
Reviewed by: Rebecca D. Burwell, Brown University, USA; Mihail Bota, University of Southern California, USA; Andreas H. Burkhalter, Washington University School of Medicine, USA
*Correspondence: Natalie L. M. Cappaert, SILS – Center for NeuroScience, University of Amsterdam, Science Park 904, 1098 XH Amsterdam, Netherlands. e-mail:
†Niels M. van Strien and Natalie L. M. Cappaert shared last authorship.
This is an open-access article subject to a non-exclusive license between the authors and Frontiers Media SA, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and other Frontiers conditions are complied with.
A connectome is an indispensable tool for brain researchers, since it quickly provides comprehensive knowledge of the brain's anatomical connections. Such knowledge lies at the basis of understanding network functions. Our first comprehensive and interactive account of brain connections comprised the rat hippocampal–parahippocampal network. We have now added all anatomical connections with the retrosplenial cortex (RSC) as well as the intrinsic connections of this region, because of the interesting functional overlap between these brain regions. The RSC is involved in a variety of cognitive tasks including memory, navigation, and prospective thinking, yet the exact role of the RSC and the functional differences between its subdivisions remain elusive. The connectome presented here may help to define this role by providing an unprecedented interactive and searchable overview of all connections within and between the rat RSC, parahippocampal region and hippocampal formation.
A connectome is a comprehensive description of the network elements and connections that form the brain (Sporns et al.,
The RSC is the most caudal subdivision of the strip of cortex around the corpus callosum that is generally referred to as the cingulate cortex. In primates, the cingulate cortex is subdivided into an anterior and a posterior part and the most caudoventral subdivision of the posterior cingulate cortex is called RSC, whereas in rodents the RSC comprises the entire posterior cingulate cortex (Vogt and Peters,
The cognitive functions in which the RSC is engaged show a striking similarity with those that engage the medial temporal lobe system; a system that comprises the HF–PHR. The loss of cognitive capabilities as seen in patients with RSC lesions is remarkably similar to those seen in patients with HF–PHR damage (Scoville and Milner,
As in humans, the RSC of rats is thought to be important for a variety of cognitive tasks. RSC lesions impair performance in spatial memory tasks (Sutherland et al.,
The functional relevance of the RSC and the striking overlap with the functional connotations attributed to HF–PHR strongly suggests a functional relationship between these areas. Knowledge about the connectome that underlies this relationship is relevant, but presently not available in an accessible format. In this review, all reported anatomical connections within the RSC and between the RSC and the HF–PHR in the rat are presented. The general patterns of connectivity will be presented in a condensed written form and specific connection patterns will be highlighted to evaluate possible functional implications. Additionally, all published connections between the RSC and HF–PHR and the intrinsic connectivity of the RSC were integrated in the already published interactive diagram of all published connections of the rat HF–PHR (Van Strien et al.,
Multiple definitions and nomenclatures for the rat cortical mantle exist. Krieg (
Vogt et al. ( |
Brodmann ( |
Rose ( |
Krieg ( |
Rose and Woolsey ( |
|
A30 | A29d | RSag (rostral and intermediate A30) | 29c | Area cingularis | |
A29c | A29c | RSgβ (caudal A30 and A29c) | 29b | Area cingularis | |
A29ab | RSgα | Retrosplenial area | |||
A29b | A29b | ||||
A29a | A29a | ||||
Vogt et al. ( |
Meibach and Siegel ( |
Krettek and Price ( |
Sripanidkulchai and Wyss ( |
Van Groen and Wyss ( |
Shibata ( |
A30 | RSAG | RsAg | Rag | Rdg | RSA |
A29c | RSGd | RsG | Rgb | Rgb | RSG |
A29ab | RSGv | RsG | Rga | Rga | RSG |
A29b | |||||
A29a | |||||
Vogt et al. ( |
Zilles and Wree ( |
Burwell and Amaral ( |
Jones et al. ( |
Shibata et al. ( |
|
A30 | RSA | RSPd | Rostral and intermediate Rsd | 29d | |
A29c | RSG | RSPv | RSv-b | 29c | |
A29ab | RSG | RSv-a and caudal RSd | |||
A29b | 29b | ||||
A29a | 29a |
†
‡
§
The RSC is a neocortical structure situated in the midline of the cerebrum. It arches around the dorsocaudal half of the corpus callosum in the rat, where it is bordered rostrally by the anterior cingulate cortex, caudoventrally by the PHR and laterally by the parietal and visual cortices. The coordinate system that defines position within the RSC is explained in Figure
The HF is a C-shaped structure situated bilaterally in the caudal part of the brain. It is subdivided into the dentate gyrus (DG), the Cornu Ammonis (subdivided into CA3, CA2, and CA1), and the subiculum (Sub). The HF consists of three layers, a deep polymorph layer, a more superficial cell layer and on the outside a molecular layer that is almost devoid of neurons. The deep layer is called hilus in the DG and stratum oriens in CA and is not really differentiated in Sub. In DG the cell layer consists of granule cell bodies. In CA and Sub, the cell layer contains pyramidal cells. The superficially positioned molecular layer in DG and Sub is not further subdivided, whereas in CA3, it is divided into three sub-layers: stratum lucidum, stratum radiatum, and stratum lacunosum-moleculare. The lamination of CA2 and CA1 is the same, with the exception that the stratum lucidum is missing.
The PHR borders HF caudally and medially. It is subdivided into the presubiculum (PrS), the parasubiculum (PaS), the entorhinal cortex (EC), further subdivided into the medial and lateral entorhinal area (MEA and LEA respectively), the perirhinal cortex (PER; divided into Brodmann's areas 35 and 36) and the postrhinal cortex (POR). The PHR is generally described as having six layers. The delineation and the HF–PHR connections are extensively described in earlier publications (Witter and Amaral,
A search was performed on publications reporting tract-tracing studies on intrinsic RSC and RSC – HF–PHR connections in PubMed
Next, results from independent retrograde and anterograde experiments were combined, such that both the layers of origin and termination could be determined. The connections were added to the existing HF–PHR connectome (Van Strien et al.,
In the following section, the intrinsic connections between RSC subdivisions and connections between RSC subdivisions and the HF and PHR are summarized in a condensed written form (for a condensed overview see Table
Intrinsic projections are those confined to a defined cytoarchitectonic subarea. In case of A29a and A29b, reports either describe those separately or the two areas have been combined into one A29ab. We therefore deal with the two areas in this section together. The intrinsic projections of A29a originate in layer III–VI and terminate in layers I, II, III, and V. Those of A29b follow a rostral-to-rostral and a caudal-to-caudal pattern. Rostral projections originate in layers II, III, V, and VI and terminate rostrally in all layers. Caudal projections terminate caudally in layers I, II, III, and V (Shibata et al.,
Both A29a and A29b project to the entire rostrocaudal extent of A29c (Vogt and Miller,
Neurons in layers V and VI of the caudal part of A29a project to caudal levels of A30, terminating in layers I, II, III, and V (Shibata et al.,
The intrinsic connections of A29c arise from the entire rostrocaudal extent and project to the entire rostrocaudal extent (Vogt and Miller,
The intrinsic connections of A30 arise from the entire rostrocaudal extent and project to the entire rostrocaudal extent, whereby layers II–VI project to all layers (Vogt and Miller,
A29ab projects to all subdivisions of PHR. Dense projections exist from A29ab to PrS (Van Groen and Wyss,
A29c also projects to all subdivisions of PHR. The projection to PrS is topographically organized along the rostrocaudal axis, such that caudal A29c projects to the entire septotemporal extent of PrS, intermediate A29c projects to intermediate and septal PrS, whereas rostral A29c only projects to septal PrS (Meibach and Siegel,
Similar to A29ab and A29c, A30 also projects to all subdivisions of the PHR. The projections to PrS originate in layers II and V and terminate in layers I, II, III, V, and VI (Vogt and Miller,
Some PHR sub-regions send return projections to A29ab. Septal and temporal PrS layer V neurons project to A29ab layers I–V (Finch et al.,
Neurons in PrS layer V project to layers I and III of A29c. Septal PrS projections terminate in the whole rostrocaudal extent of A29c (Vogt and Miller,
Neurons in layers V and VI of septal PrS project to layers I, III, IV, and V of A30 (Vogt and Miller,
Neurons in layer V of A29ab project to Sub in the HF (Van Groen and Wyss,
Also for A29c, Sub is the only HF target (Meibach and Siegel,
The HF projections to A29ab originate in Sub and CA1. Sub projections to A29ab terminate in layers I, II, and III and mimic the topographical organization of the retrosplenial projections to Sub. The projection from CA1 to A29ab originates from the septal portion (Van Groen and Wyss,
The HF projections to A29c also originate in Sub and CA1. The proximal part of the septal Sub projects to rostral A29c. Intermediate septotemporal and distal Sub project to caudal A29c (Vogt and Miller,
The HF projections to A30 only originate in Sub. The distal portion of the septal Sub, as well as the intermediate proximodistal portion of intermediate septotemporal Sub project to layers I and II of A30 (Vogt and Miller,
There are strong intrinsic connections within the RSC subdivisions. All rostrocaudal levels within both A29c and A30 issue projections to their respective rostrocaudal extents. A29b projections have a strict topography from rostral-to-rostral and from caudal-to-caudal; A29a only has a caudal-to-caudal projection. There are also strong reciprocal connections between the RSC subdivisions. All rostrocaudal levels of one subdivision project to all rostrocaudal levels of all other subdivisions, but there are some exceptions: (1) caudal A29a projects only to caudal A29b, A29c, and A30; (2) caudal A29b does not project to rostral A29c; (3) rostral A29c does not project to caudal A30; (4) rostral and midrostrocaudal A30 only projects to caudal A29b and the return projection from A29b only terminates in caudal A30.
The RSC projects to all PHR subdivisions and Sub. Only the projections of RSC to PrS and Sub show a topographical organization such that the rostrocaudal axis of origin in RSC correlates to a septotemporal terminal distribution in PrS and Sub. The projections to PrS are among the densest of RSC–PHR connections (Jones and Witter,
The connectome of the rat brain should describe all network elements and connections in a clear and comprehensible way. Compared to the comprehensiveness that a connectome implies, current knowledge is in its infancy. When considering the vast number of neurons in the rat nervous system and their connections, together with the currently available technologies to collect and handle information about them, creating a connectome is an expensive, time consuming, and complicated task. Scientists will eventually have a comprehensive map of the rat brain available, and just like the usefulness of an easily accessible map of all the roads in the world, such a connectome will be an indispensible foundation for basic and applied neurobiological research (Sporns et al.,
With this publication, we present the current version of our partial rat brain connectome, which to the best of our knowledge represents all current information on the ipsilateral pathways within and between the HF, PHR, and RSC. A different approach from that of a traditional meta-analysis was taken, to create this connectome. In a traditional meta-analysis, typically only a subset of data is selected, summarized, and organized according to the author's views. The resulting reduction in detail of anatomical networks is useful for creating scientific hypotheses, but contradicts with a fundamental characteristic of a connectome: to be an exhaustive knowledge resource. Therefore, we chose the approach to present the anatomical data of the selected regions in the fullest available detail, which allows scientists to prune this information themselves to match their hypotheses, or design competing anatomical hypotheses.
We realize that the current state of knowledge is not exhaustive and hence one could argue that the connectome presented here is not a real connectome. Nevertheless, the connectome presented here provides the best approximation of a full connectome at the current point in time. With future publications we aim to continually update and expand the database. Still, users of connectomes should always keep a perspective on where the current state of knowledge stands compared to having absolute knowledge. For this reason, this discussion will first touch upon some of the challenges of anatomical connectomes that remain to be resolved, after which the potentials of connectomes will be exemplified using the information presented in this review.
Combining data produced by many researchers, over 100 years, using many different techniques in a great number of tract-tracing experiments, leads to a number of challenges on demarcation of brain areas and designation of names in research reports. Such nomenclature issues exist not only for the RSC, but for almost all brain regions. These issues have arisen because different histological techniques produce different definitions of borders in the brain, or simply because researchers disagree on the demarcation.
Krieg (
Even when using technically advanced methods to clarify projection patterns, the usefulness of a scientific report on anatomical connections depends to a great extent on how detailed the authors report their results and the way they present the data in figures to support their observations. When reviewing the projections from RSC to HF and PHR, only in a few studies information was provided on the layers of origin or termination, or specific projection patterns. For example, projections from A29ab, A29c, and A30 to LEA and MEA exist (Van Groen and Wyss,
To obtain information about layer specificity of projections, ideally anterograde and retrograde tract-tracing experiments are combined. For example, an anterograde tracer injection is placed in midrostrocaudal A30, thereby discovering regions of termination including layers I, II, III, and V of rostral A29c. This experiment should be completed by placing a retrograde tracer in rostral A29c, to reveal the layers of origin of this projection in midrostrocaudal A30, i.e., layer II, III, V, and VI (Shibata et al.,
Apart from layer specificity, projections may be topographically organized. For example, the rostral RSC projects weaker to EC than caudal RSC. Often, such topographies show a gradient in projection strength, but one cannot rigorously claim that only caudal RSC projects to EC. In the current version of our connectome, such topographies are not visible since no information on relative density is included. We chose this approach, in view of the risk that by emphasizing some brain connections over others, the less emphasized ones may be erased from the scientific working memory. However, users should not forget that brain connections typically have different strengths and may show topographical gradients that are not apparent in the interactive connectome.
When using our connectome, it is important to keep in mind that two connections symbolized by two similar looking lines may be different from one another for a number of reasons.
Our connectome is based on tract-tracing data and thus comes with certain limitations related to this technique. One important limitation is that tract-tracing does not reveal if a connection is excitatory or inhibitory, whereas this information is of functional relevance. By combining immunohistochemistry with either confocal or electron microscopy, it can be established if a projection is excitatory or inhibitory (Van Haeften et al.,
In our current connectome, known GABA-ergic connections are not specifically included, although some projections are likely GABA-ergic. There is a dense GABA-ergic projection from CA1 to the RSC, starting from all layers except the stratum lacunose-moleculare (Miyashita and Rockland,
Most anatomical reports contain subjective descriptions of the strengths of projections. Such subjective reports are impossible to quantify and therefore such information cannot be incorporated in the connectome. Yet, in reality differences in strengths exist. For instance, the projection from A29ab to MEA is reported to terminate in both superficial and deep layers. When assessing this connection in detail, clear differences between superficial and deep layers exist. There is dense terminal labeling in layer V of MEA, whereas terminal labeling in layer III is very light (Jones and Witter,
Another challenge is how to interpret “negative information.” There are two types of negative information: (1) there is data that indicate that a connection between area A and area B is very unlikely, or (2) a connection between area A and area B was not reported, but may exist. For example, the current literature assessment shows that A29c and A29ab both project to Sub, whereas data on a projection from A30 to Sub are lacking. Should one now conclude that a projection from A30 to Sub is not present in the rat brain, or could it be that this projection is present, but perhaps remained undetected or was detected but not reported? This issue has been dealt with in other databases, where connections that are explicitly reported to not exist have been collated (Stephan et al.,
Although it is assumed that researchers aim to report as accurately as possible the results of their tract-tracing experiments, it does not exclude the possibility that brain connections are reported that do not exist. Tract-tracing techniques were much refined over time, such that injections of modern tracers such as Phaseolus Vulgaris-leucoagglutinin or biotinylated dextran-amine can now be injected region or cell layer selective (Gerfen and Sawchenko,
A bigger problem occurs when a report provides many connection details, but too little evidence is presented to confirm the claims of authors. In such cases, one can only trust the author's interpretation of the data, in the absence of proof against it. However, when sufficient proof against the existence of a connection is available, such information is registered in our connection database, but excluded or removed from the connectome. An example of a connection that was reported but is not in the current connectome is based on an injection in A30 and A29c with terminal labeling in Sub, septal PrS and PaS (Vogt and Miller,
Apart from scientific value, a connectome is useful as an educational tool to get an overview of an ever increasing amount of literature. Instead of having to search through many papers, basic anatomical facts such as the three dimensional organization of a structure, but also complex issues such as differences in nomenclature and perhaps most complex, the numerous connections between brain regions, are neatly organized, such that users can easily get an exhaustive overview of information at the level of detail that they choose. This is useful for novice researchers who just started to learn about the organization of the brain, as well as researchers who wish to expand their research into new brain regions that are charted in the connectome.
There are several scientific uses of connectomes. A connectome can help detect knowledge gaps, it will improve the interpretation of experimental results and may facilitate the design of new experiments. Depending on the level of detail of the available information, the scientific value varies between these uses. If little information is available, a connectome will mainly help detect knowledge gaps. When a reasonable amount of details are known, it will help better understand completed experiments. The true potential of a connectome becomes apparent only when the level of detail of information goes beyond the point where one can easily keep track of all the known facts.
Even though the current connectome may look overwhelming already, a close look immediately points to gaps in our knowledge. For example, there are no experiments evaluating the topographical organization of the connections between A29a or A29b and HF–PHR. Anterograde or retrograde injections in A29a or A29b and the topography of terminating patterns or labeling of neurons in PHR have not yet been described. Similarly, injections in HF–PHR and descriptions of the terminating patterns in A29a or A29b are missing. Another example concerns the topography of the projection from PrS to RSC. It is known that temporal PrS projects specifically to A29ab, whereas septal PrS projects to all RSC subdivisions. It is likely that a more specified topography exists, for example along the rostrocaudal axis of RSC, but such information is presently unknown.
Another useful scientific application is the graphical representation of hypotheses. Sometimes it is easier to understand a hypothesis by looking at a graphical representation of the connections. For example, it was hypothesized that information transfer between the RSC and PHR–HF is crucial for adequate navigation and spatial memory. This is supported by the observation that extensive lesions, including A29ab and A29c disrupt spatial memory tasks (Pothuizen et al.,
Another example of an inferred experiment, based on the connectome follows from the topography observed in the RSC – HF–PHR projections. The traditional emphasis on the topography of RSC – HF–PHR projections relies on the observation that there is a gradient in the RSC to HF–PHR projections, such that rostral RSC (A29c) projects primarily to septal parts of HF–PHR (PrS) and caudal RSC (A29c/A30) projects both to septal and temporal HF–PHR (PrS). Upon close inspection, a more striking topography is apparent in the projections of Sub to A29c. The transverse axis of Sub relates to the rostrocaudal axis in A29c such that proximal Sub projects to the rostral part of A29c and distal Sub projects to the caudal of A29c. This topography could be of functional relevance, since LEA projects to proximal parts of Sub, whereas MEA projects distally (Tamamaki and Nojyo,
The current connectome, as presented here, is far from complete and differences exist in the amount of information that is available about anatomical demarcation and connectivity of different brain areas. The HF connectivity is relatively well covered, followed by the PHR – HF connections and least is known about HF – RSC connectivity. As indicated, the amount of available information is largely decisive for the type of questions a connectome can assist with. We do not see these issues as shortcomings undermining the value of this approach. Connectomes can continuously be updated and extended and therefore will always provide an exhaustive and up to date account of the current knowledge. These two factors precisely define the value of connectomes. Moreover, anatomical connectivity characterizes the brain at an intermediate information level, allowing to easily link this to, e.g., functional properties of individual cells or effective connectivity (Friston,
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The Supplementary Material for this article can be found online at
We like to thank Ingrid Riphagen for designing and performing the initial literature searches. Niels M. van Strien was supported by the Norwegian Research Council, Independent projects – Biology and Biomedicine grant number 197245. Menno P. Witter and Jørgen Sugar received support from the Kavli Foundation and the Norwegian Research Council, centre of excellence grant number 145993.
1
2