Introduction
Much of our knowledge about the human brain is based on knowledge obtained in other species. While numerous species have been used to model the human brain, the mouse has emerged as the most prominent of these, due to its rapid life cycle, straightforward husbandry, and amenability to genetic engineering (Dietrich et al.
2014). The overall assumption in this work is that the knowledge obtained in the mouse ‘model species’ is translatable to the human, due to overall similarities in biological properties of the two species. However, the success rate of such translations have sometimes been disappointing, especially in the case of neuropsychopharmacology (Hay et al.
2014). This is due, in part, to assumptions of between-species similarities not holding (Striedter
2022). As such, it is becoming increasingly apparent that these assumption should be subjected to explicit empirical validation.
The various divisions of cingulate cortex have repeatedly been shown to be important in many aspects of emotional processing, decision making, and cognitive control (Behrens et al.
2013; Leech and Sharp
2014; Kolling et al.
2016) and alterations in cingulate morphology (Goodkind et al.
2015; Opel et al.
2020) and functional connectivity (Marusak et al.
2016) are a common observation across a range of psychiatric disorders. Cingulate cortex is thought to be an evolutionary conserved region in mammals. Indeed, an analysis of common areas across six major mammalian clades suggests that cingulate cortex is present in all and that it could have been part of a limited set of neocortical regions present in early mammals (Kaas
2011). The combination of common alterations in disease and apparent evolutionary conservation make cingulate cortex an important target for translational neuroscience research.
However, the similarity of rodent and human cingulate has been called into question on a number of grounds. First, it has been argued by some authors that rodent cingulate cortex has organizational features similar to those of primate dorsolateral prefrontal cortex or that at least it performs homologous functions (Brown and Bowman
2002; Uylings et al.
2003; Carlén,
2017). Second, among researchers who reject these claims, there still is some debate about how rodent cingulate should be subdivided and how its organization relates to that of the primate (Laubach et al.
2018; van Heukelum et al.
2020). Third, even if cingulate cortex were found to be fully homologous in mouse and human it would be embedded within the larger prefrontal network within the human brain compared to other species (Schaeffer et al.
2020). These arguments continue to be reassessed with the appearance of new data types that enable better and more complete comparisons across species.
One way to explore similarities and differences in brain organization across species is by studying connectivity. The connections of brain areas constitute a unique ‘fingerprint’ and provide information about the area’s incoming information and the influence it exerts on other parts of the brain (Passingham et al.
2002; Mars et al.
2018). We have previously employed functional connectivity as assessed using resting state fMRI to compare connectivity across humans and non-human primates (Mars et al.
2011,
2016) and humans and mice (Balsters et al.
2020). Cingulate connectivity has been studied using neuroimaging in a number of studies using both diffusion MRI tractography (Beckmann et al.
2009; Smith et al.
2018) and functional connectivity (Margulies et al.
2007; Hutchison et al.
2012; Schaeffer et al.
2020). Even if the species studied have diverged such a long time ago that assessing homology purely by means of connectivity is likely to be difficult, studying the patterns of connectivity across cingulate cortex is likely to provide insight in the similarities and differences in cortical organization (cf. Van Heukelum et al. (
2020)). Here, we study mouse cingulate functional connectivity, assessing it against the ‘gold standard’ of tracer-based structural connectivity, and compare it with similar data from the human. The goal of the study is to assess to what extent the general organizational principles of cingulate organization are comparable across the two species.
Discussion
We set out to investigate whether the mouse and human cingulate cortex are organized according to similar principles in terms of their connectivity to other parts of the brain. Overall, we show that the two species’ cingulate cortices follow broadly similar principles, with anterior areas mostly interacting with amygdala, nucleus accumbens, and orbitofrontal cortex; a midcingulate territory interacting with premotor and posterior parietal cortex; and a retrosplenial zone interacting with hippocampus. The similarity of these patterns is inconsistent with theories ascribing homology of rodent anterior cingulate with primate granular prefrontal cortex (Krettek and Price
1977; Eichenbaum et al.
1983) or that suggest rodent cingulate contains a mixture of primate cingulate and granular prefrontal features (Uylings et al.
2003). Rather, it is consistent with notions that infralimbic cortex in the mouse is similar to primate area 25 (Alexander et al.
2019), that there is a midcingulate zone with parietal and premotor connections in both species (Vogt
2016), and a generally similar anterior–posterior organization in both species (cf. Van Heukelum et al. (
2020)).
These similarities between species notwithstanding, some differences are apparent. Overall, many human cingulate areas have connectivity profiles that seem more distinct from one another that do those of the mouse cingulate cortex. Earlier work suggested parietal connectivity in the mouse is with both midcingulate and retrosplenial cortex (Zingg et al.
2014) and we indeed see rather widespread parietal connectivity in the mouse. Premotor and parietal connectivity are more restricted to the mid-cingulate cortex in the human. Human midcingulate cortex is thought to contain distinct anterior and posterior subdivisions (Vogt
2016), the first of which is not present in the mouse (Vogt and Paxinos
2014). In the human, parietal connectivity is stronger in the posterior part of midcingulate.
Mouse connectivity as assessed using tracers and using resting state functional MRI were overall in agreement, but some differences were noticeable. Hypothalamic connectivity as assessed using tracers was very strong in the most anterior parts of the cingulate, consistent with earlier reports (van Heukelum et al.
2020), but functional connectivity was more broadly distributed in both species. Human hippocampal functional connectivity resembled that of mouse tracers, but not mouse functional connectivity. This could be due to the effect of anaesthesia on the resting state fMRI of the mouse, but this awaits systematic comparison. Hippocampal and hypothalamic connectivity to posterior seeds was much stronger in the human than in the mouse. This is potentially due to the presence of a large posterior cingulate in the human (Bzdok et al.
2015), whereas in the mouse, this area only contains a retrosplenial cortex (Vogt and Paxinos
2014).
The clearest dissociation between the mouse and human data was in the connectivity of the insula. This is true even though we seeded in territory commonly described as agranular anterior insula in both species. In the mouse, the insula seed showed connectivity with anterior parts of the cingulate in both tracer and rs-fMRI data, while in the human, the insula seed showed strong functional connectivity with midcingulate areas. The human results are in accordance with models of dorsal anterior cingulate function in cognitive control and the participation of the two regions in a so-called salience network (Seeley et al.
2007; Menon
2011). Previous work has shown that the salience network, although present in both species, has different associations with the serotonergic network across human and mouse (Mandino et al.
2022), suggesting that the area has changed substantially since the last common ancestor of mice and humans. Alternatively, the insula seed areas we selected in human and mouse are not homologous. We have taken a region commonly used in neuroimaging studies as our human anterior insula (Cieslik et al.
2015; Molnar-Szakacs and Uddin
2022), but Öngür and Price (
2000) describe a number of insular regions more anteriorly, on the caudal orbital surface. Whether the anatomical similarity between these human caudal orbital areas and mouse insular areas is greater than that between our human insula seed and the mouse is a topic for further investigation.
It is important to emphasize that we here compare principles of cingulate connectivity across species, rather than matching cingulate areas across the human and mouse brain one by one. We have previously used connectivity fingerprints to make more explicit, quantitative comparisons between regions of the human and macaque monkey brain (Mars et al.
2013; Sallet et al.
2013; Neubert et al.
2014); we developed a formal framework to do so (Mars et al.
2016) which has been used by a number of other groups since (Thompkins et al.
2018; Wang et al.
2019; Schaeffer et al.
2020). However, humans and mice share a common ancestor about 87 millions years ago, which is much more than the 29 million years of humans and macaques (Kumar et al.
2022). This means that changes in how connections relate to other aspects of brain organization, such as gene expression, receptor architecture, and cytoarchitecture, might have occurred (cf. Krubitzer and Kaas
2005). Testing hypotheses of similarity between distinct cortical areas in the two species should, therefore, ideally use a multi-modal approach. The current study is a first test of similarity in principles of connectivity, ongoing and future work will supplement this work by investigations in other modalities, after which a more detailed areal comparisons across species can be achieved.
In general, it is important that the target areas used are homologous when comparing connectivity across species. Here, we have taken care to use regions that are identified as such, but some discussion is in order especially when in case of targets in the neocortex. The approach used here can be used to ascertain the degree of similarity/difference between any areas in human and mouse. With respect to premotor cortex, the human brain contains areas that have no homolog in the rodent (Wise
2006), but the two brains’ premotor cortices do follow largely similar organizational principles (Lazari et al.
2023). The human ventral frontal cortex contains agranular, dysgranular, and granular areas, but rodent prefrontal cortex contains only agranular areas (Wise
2008; Rudebeck and Izquierdo
2022). We here used human area 14m as defined by Neubert et al. (
2015), which is posterior to the granular areas, as our orbitofrontal cortex seed. We do note that similar results could we obtained using targets in granular area 11 m. Posterior parietal cortex is dramatically expanded in primates compared to other mammals (Krubitzer and Padberg
2009), but a mouse parietal association area that is homologous to primate posterior parietal cortex has been identified (Lyamzin and Benucci
2019). The current human parietal results are similar for targets overlapping with human MIP or 7A (Mars et al.
2011).
Apart from these differences described earlier, it should be taken into account that the human cingulate is embedded within a much larger and more elaborate neocortical network than that of the mouse. This means that, even if the overall organization of the two species’ cingulate with homologous areas is comparable, connectivity with non-homologous areas mean that the overall connectivity profile can still be quite distinct. This was previously shown to be the case for the human dorsal caudate, although striatal connectivity follows similar organisational principles in both species, connectivity of the dorsal caudate with the human frontal pole means that its connectivity profile is distinct from any found in the mouse (Balsters et al.
2020). Connectivity between the human medial frontal gyrus and human cingulate is evident in the present data, in particular area 9/46D as defined by Sallet et al. (
2013), and in previous studies (Sallet et al.
2013; Loh et al.
2018). In the striatum, areas with a distinct human connectivity profile were associated with higher order cognitive processes, including executive control and language. It remains to be seen whether functional differences are found between the two species’ cingulate regions.
Model species are an essential part of research in biology and by extension neuroscience (Striedter
2022). Differences between the model and the species of ultimate interest, i.e., the human, are to be expected and do not necessarily present a problem for translational neuroscience, as long as these differences are properly understood. Whole-brain, high-throughput data are now increasingly available and allow us to gain a much more systematic understanding of such differences than ever before (Mars et al.
2014). The present work contributes to this effort by comparing a major target area for clinical research across species by means of connectivity. Future work will focus on comparing these results obtained using comparative connectivity with those obtained using other modalities, such as spatial patterns of gene expression, tissue properties, and receptor densities (Vogt et al.
2013; Beauchamp et al.
2022).
In sum, this work shows the feasibility of extending existing approaches of comparing frontal cortical organization across species using functional MRI to rodent-human comparisons. The results show a generally conserved macro-level organization, although there are important differences in both regional specialization and embedding within larger cortical networks. Such differences are important to take into account when performing between-species translations in the context of clinically relevant research.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.