The orbitofrontal cortex of the sheep. Topography, organization, neurochemistry, digital tensor imaging and comparison with the chimpanzee and human

A total of 12 animal brains were analysed (Ovis aries n = 8, Pan troglodytes n = 4). DTI scans of human brains (n = 5) were obtained from the Human Connectome Project (HCP) database (https://​ida.​loni.​usc.​edu/​login.​jsp). Information can be found in Table 1.

Sheep samples originate from a commercial slaughterhouse. The sheep were treated according to the European Community Council directive (86/609/EEC) on animal welfare during the commercial slaughtering process and were constantly monitored under mandatory official veterinary medical care. The definition of “adult” was based on the official documentation available corresponding to each individual ear mark and confirmed evaluation of body size by examination of teeth wear. The chimpanzees were brought to the Department of Comparative Biomedicine and Food Science (BCA) of the University of Padova from their outdoor location for post-mortem examination. All animals were previously housed in zoological parks operating under mandatory governmental CITES permits and consequent vigilance. All animals were found dead by the respective wardens. Causes of death were due to heart or system failure unrelated to the nervous system. Post-mortem macro- and microscopic examination of the brain did not identify lesions or anomalies.

The HCP project (Principal Investigators: Bruce Rosen, M.D., Ph.D., Martinos Center at Massachusetts General Hospital; Arthur W. Toga, Ph.D., University of Southern California, Van J. Weeden, M.D., Martinos Center at Massachusetts General Hospital) is supported by the National Institute of Dental and Craniofacial Research (NIDCR), the National Institute of Mental Health (NIMH) and the National Institute of Neurological Disorders and Stroke (NINDS). HCP is the result of efforts of co-investigators from the University of Southern California, Martinos Center for Biomedical Imaging at Massachusetts General Hospital (MGH), Washington University, and the University of Minnesota.

All the animal brains were extracted and fixed in neutral buffered paraformaldehyde (4%). The time interval between the death of the animal and the extraction of the brain are reported in Table 1. All these processes took place at the department.

After at least 1 month of fixation, three sheep brains were transferred to the Department of Neuroscience, Biomedicine and Movement of the University of Verona to perform the MRI using a 4.7 Tesla, 33-cm bore horizontal magnet (Oxford Ltd., Oxford, UK) equipped in a Bruker tomograph (Bruker, Karlsruhe, Germany). The OFCs of the remaining sheep were collected following landmarks available in (Rose 1942; Constantinescu 2018; Barone 2004). Sampling in the chimpanzee brain was performed following directions found in specific atlases (Bailey et al. 1950). In each specimen, we collected an axial section in the middle of the area to include an easily recognizable layer IV.

The samples were processed with conventional techniques and then embedded in paraffin for histology and immunocytochemistry studies. For both the processes, samples were then cut into 8 µm sections through a rotatory microtome (LEICA®). The paraffin was removed through three baths of 5 min each and rehydrated through descendent alcohol solutions (100%, 95%, 70%, 50%) and distilled H₂O for 3 min each.

For histology we used a Nissl protocol. Therefore, sections were immersed in a 0.4% cresyl violet solution for 4 – 20 min depending on the intensity of the staining. Subsequently, they were rinsed in tap water, the result of the staining was rapidly checked under an optic microscope and, when necessary, some sections were immersed in cresyl violet twice or more until they reached a satisfactory result. Finally, they were dehydrated in an increasing alcohol solution series (70%, 95%, 100%) for a few seconds, immersed in xylene (four baths of 3 min each) and cover slipped.

For immunocytochemistry sections were immersed in a citrate solution (pH 6) or Tris HCl (Trizma base, pH 9, Sigma-Aldrich®), heated up to 90 °C in a microwave for 10 min and left to cool down at room temperature for 30 – 50 min. They were then washed in three baths of 3 min each of Tris HCl Buffered Saline (TBS). After this, they were moved into a wet chamber and a perimeter with the ImmEdge pen (Vector Laboratories, Inc., Burlingame, CA, USA) was drawn to maintain the volume of the subsequent solutions (150 µL) on the sections and were treated with H₂O₂ at 3% for 10 min. After three rinses in TBS, they were incubated 1 h at room temperature then overnight at 4 °C with the primary antibody (Table 2) in a dilution solution (SuMi, 0,5 mL Triton X-100 and 0,25 g of gelatin in 100 mL of TBS). Three rinses in TBS followed and then, the sections were incubated 2 h with the secondary antibody (Table 2) in SuMi or 20 min in HRP (EnVision FLEX system, Dako, USA) solution. After three washes in TBS, samples in HRP were treated with a diaminobenzidine-chromogen Substrate Solution (EnVision Flex) while others were incubated 30 min in Avidine Biotin Complex kit (Vectastain, Vector Labs, Burlingame, UK), rinsed in TBS and incubated in diaminobenzidine-tetrahydrochloride kit (DAB, Merck). Both last incubations lasted 5 – 20 min depending on the intensity of the staining. Consequently, sections were rinsed in distilled H₂O, counterstained with the Nissl’s protocol described in the previous paragraph and finally cover slipped.

Final slides were scanned with the D-Sight v2 (Menarini Diagnostics srl, Italy), a semi-automated microscope, at 20 X magnification, fast mode and automatic focusing. Images acquired were in Jpeg2000 format and were used, through the program Navi Viewer (Menarini Diagnostics srl, Italy), to calculate the thickness of the cortical layers. These are expressed in mean plus standard deviation and the percentage of the single layer on the total thickness. Three measurements were taken per section, two sections per individual. Those images used for this paper were transformed and elaborated in GIMP (GNU Image Manipulation Program 2.10, www.​gimp.​org) and Inkscape 1.0 (The Inkscape Project, www.​inkscape.​org).

In the sheep, layer IV was absent in all the sampled tissue (Fig. 3a) and we did not observe any rostro-caudal development. Layers I, III, V and VI were relatively equally distributed (range 20.2 – 26.2%) except for layer II (range 9.2 – 9.9%).

In the chimpanzee, on the other hand, a layer IV was present (Fig. 3b) and we found percentages of similar thickness for layers III, V and VI, whereas layer II was reduced by 40% and layer I more than doubled (Table 3).

Staining with Ca-BPs revealed immunoreactive (-ir) neurons in the examined species for all the three antibodies (Figs. 4, 5 and 6). Their distribution spanned throughout the cortex with some differences.

In the sheep, PV-ir neurons were mostly present in layers III and V and fewer were generally found in layer II (Fig. 4a). Most of them were interneurons with various shapes (Fig. 4a’ and 4a’’) but few pyramidal neurons were also found, mostly in layer V (Fig. 4a’’’ and 4a’’’’). The chimpanzees showed a distribution of PV-ir neurons more clustered in layer III, followed by layers V, VI, II and IV (Fig. 4b). Only interneurons were marked (Figs. 4b’ – 4b’’’).

In the sheep, CR-ir neurons were mainly present in layer V and to a lesser extent in layers III and VI (Fig. 5a). They were mostly interneurons (Fig. 5a′ – a′′′′); however, pyramidal neurons were also found both in layer III (Fig. 5a’) and V (5a′′′, 5a′′′′). In the chimpanzee, CR-ir neurons were seemingly more present in layer II and less in layers I, III, V and VI (Fig. 5b). They were mostly interneurons (Figs. 5b′, 5b′′) but also pyramidal neurons were found (Fig. 5b′′′).

In the sheep, CB-ir neurons were evenly distributed along the cortex, with relatively more neurons found in layer VI (Fig. 6a). Many of them were interneurons in layers II (Fig. 6a′), III (Fig. 6a′′), V (Fig. 6a”’) and VI. However, pyramidal neurons were mostly found in layer VI (Fig. 6a’’’’). In the chimpanzee, CB-ir neurons were found mainly in layer III but could also be seen in other layers (Fig. 6b). We observed interneurons in layers II (Fig. 6b′), layer III (Fig. 6b′), layer IV (6b′′′) and layer V (Fig. 6b). Pyramidal neurons were present only in layer III (Fig. 6b′′).

A whole tractographic image showing tracts from the OFC is in Fig. 7.

As expected, Nissl staining revealed the complete absence of layer IV in the sheep over the entire length of the region, whereas this layer was found in the chimpanzee (Fig. 3). Notably, layer I in sheep was more developed in proportion compared to the chimpanzee (21% vs. 8.7%) (Table 3). This is consistent with the literature and could be consistent with a higher reliance on layer I for cortico-cortical connections. Other studies confirmed the absence of layer IV in Artiodactyla (Hof et al. 1999) and, specifically in the sheep, the “area orbitalis” of Rose (1942) lacked the same layer as well. The OFC in the chimpanzees was indeed sampled to find the layer IV, due to its rostro-caudal development (Petrides and Pandya 2012).

The immunocytochemistry in both species showed higher concentration of PV-ir neurons in layer III and V. Higher concentration of CR-ir neurons were found in layer V in the sheep and layer II in the chimpanzee. Finally, a uniform distribution of CB-ir neurons along the cortex was evident for both species. As observed by Hof and collegues (1999, 2000) and Hof and Sherwood (2005), CaBP-ir neurons had a similar distribution between rats and primates but both differed from artiodactyls. Studies in rats and primates reported that CB-ir and CR-ir neurons were mostly distributed in layers II and III even though other groups present in the other layers were identified. In both cases, they were interneurons and weakly stained pyramidal neurons. On the other hand, PV-ir neurons were found more frequently in layers III-V and another small group in layers II and III (Hof et al. 1999, 2000; Hof and Sherwood 2005). Some pyramidal neurons were also marked in few functional areas in some but not all species (DeFelipe et al. 1997). This was similarly consistent with our results. Indeed, CB-ir and CR-ir were mainly interneurons located primarily in layer III followed by layers I, II and V. PV-ir neurons, present in layer III then V, VI were only interneurons. Furthermore, the organization and neurochemical organization of the orbitofrontal cortex of the chimpanzee has previously been reported in the literature to be largely similar to the human OFC (Petrides and Pandya 1994; Öngür and Price 2000).

Literature concerning artiodactyls described that, while CB-ir neurons were weakly stained and fewer compared to CR-ir neurons, both of them were large multipolar, bipolar or fusiform cells in layers I, II, and superficial III, and a few in layers V and VI. Furthermore, other ungulates such as camels, llamas and giraffes showed more CR-ir neurons than the sheep. In dolphins, in layer IIIc/V some neurons were weakly stained CR-ir and CB-ir pyramidal cells. On the other hand, PV-ir interneurons were limitedly distributed (5%) related to CR-ir and CB-ir cells (40%). In whales and dolphins, these marked interneurons were located in layers IIIc/V, whilst few pyramidal neurons also in layer III (Glezer et al. 1993; Hof et al. 1999, 2000; Hof and Sherwood 2005). In the present study, while our data corresponds to the literature for CB (Fig. 6), we did not find CR-ir cells in all layers, as we did for CB (Fig. 5). Additionally, PV-ir cells were clearly distributed mostly along layer III/V but pyramidal cells were found mostly in layer V rather than III (Fig. 4). PV-ir pyramidal neurons likely play an inhibitory neuromodulatory role in layer V, although studies in other areas indicate the possibility of an excitatory function at least partially (Meskenaite, 1997; Liu et al. 2014). The greater concentration of CR-ir neurons in layer V, compared to all other layers in the literature, could indicate a modified inhibitory function at that level.

The lack of layer IV and the diverse distribution of CaBPs might imply a different integrative architecture in the minicolumn of the OFC and potentially more broadly another organization of neuronal circuits in sheep compared to humans.

The DTI revealed both similarities and differences between human and sheep white matter tracts. Indeed, while in the human brain the areas corresponding to the OFC are four, in the sheep only one region identified based on current models (Ella et al. 2017). Notably, some tracts were remarkably analogous to human albeit some other tracts were "novel" in the sheep. In the human brain, area 47/12 had connections with more areas than areas 11, 13 and 14 (Fig. 11). Its fibers participated mostly to the uf, ifof, which linked it to the visual and acoustic areas and then the acc. Other connections were to the thalamus, other prefrontal areas such as 9, 10, 11, 9/46 and within the same area (U-shaped). Area 11 contributed to uf, ifof, reached to areas 10 and 47/12 and presented U-shaped fibers (Fig. 10). Area 13 had intra prefrontal connections with areas 10, 11 and 47/12 and joined in the uf (Fig. 12). Finally, area 14 was related to area 10, 11 and its streams to acc and cingulate cortex (Fig. 13).

General agreement on the OFC connections derived from studies, using tracing and audiography, mostly in rhesus monkey (Macaca mulatta), are resumed by reviews of Kringelbach and Rolls (2004) and Petrides and Pandya (2012) (see Table 4 below).

Based on the data files acquired from the HCP, we compared our results with the literature. The analyzed DTI lacked specific tracts and connections to some regions, but we did not find any “extra” tract previously undescribed.

In the sheep, the associations we found were with the caudate nucleus, fornix/hippocampus, claustrum, piriform cortex and acc. Interestingly, we found a large amount of corticofugal fibers and a marked right asymmetry of cortico-cortical fibers (ifof) which connected the OFC primarily to visual areas (Figs. 8 and 9).

In the sheep, tracing related to this area has been performed but data are scarce. The first study dates back to 1948 when Rose and Woolsey attempted to demonstrate whether an area homologous to the primate OFC existed in the cat, the sheep and the rabbit. In their work, they considered that frontal fields receive projections from the mediodorsal nucleus of the thalamus. However, the authors did not distinguish the various components of the mediodorsal nucleus of the thalamus, with their respective projections, and considered the OFC to be the entire PFC. Retrograde tracing was performed in just one sheep, but the results showed a much larger lesion area than the OFC they had previously found (Rose 1942). Most likely the tracer had extended slightly excessively; however, the general results should not be discarded. A subsequent study was carried out by Dinopoulos and colleagues (1985) who inoculated, on the contrary, horseradish peroxidase into various areas of the rostral prorean gyrus in ten sheep. It resulted in lesions not only to the MD nucleus of the thalamus but also to the mediodorsal part of the ventromedial nucleus, and to a lesser extent to the posterior lateral nucleus and the midline nuclei. The authors concluded that the prefrontal cortex in sheep corresponded to the OFC of Rose and Woolsey (1948) and the "regio frontalis" of Rose (1942), although they did not analyze the ventral aspect of the gyrus. These studies focused only on the connections with the nuclei of the thalamus and ignored connections with other areas of the brain. A recent research by Meurisse and colleagues (2008) demonstrated the existence of a relationship between the cortical nuclei of the amygdala with the OFC in sheep, pointing out its implication in olfactory stimuli elaboration. Despite the paucity of data on the OFC, we trust that our results are coherent with the characteristics of the OFC. Indeed, some fibers connected to the thalamus MD and certain similarities in the rest of the connections with that of the human indicate a certain robustness of the method and results.

From our tractography results, the lack of the aforementioned tracts in both species could be related to software or acquisition biases. In the sheep, the large amount of corticofugal fibers could be caused by an artifact. In effect, fibers reaching the hypothalamus as an input/output was not unexpected (Rempel-Clower and barbas 1998; Kringelbach and Rolls 2004); however, their continuation to the spinal cord in a unique bundle is quite doubtful. The hypothalamus is an integration center of visceral/autonomic mono- and polysynaptic inputs (Cameron 2001; Saper 2012). It is possible that, at this level, these fibers aligned with other ones and, having the voxels the same anisotropy, these were mixed by the algorithm and resulted as a single stream.

To summarize, while there are extensive data on OFC in primates, more concerning tracing than DTI, little is known about sheep and other artiodactyls. Given the lack of existing studies on neurofiber tracing in sheep, we could not establish accurate correspondence between the two species. However, our digital results and homologies established from literature data in this animal suggested some degree of similarity compared to the human brain. Further tracing studies could be carried out, dividing parts of the MD nucleus of the thalamus to establish precise areas.

The OFC is considered to be the cortical representation of the limbic system (Mtui et al. 2020). It is involved in reward-punishment evaluation after primary and secondary reinforcements (Hof et al. 1995; Kringelbach and Rolls 2004; Rudebeck and Rich 2018). A leucotomy of the OFC in fact produced uninhibited and aggressive behavior, hence it was originally thought to have an inhibitory role on emotions (Kringelbach and Rolls 2004; Mtui et al. 2020; Rudebeck and Rich 2018). Other types of lesions showed that the OFC was more important for decision making rather than inhibition. This has been explained by the fact that leucotomies most probably also affect areas close to the OFC thereby causing more extensive reactions while excitotoxic lesions were more targeted. Currently, most authors accept the involvement of the OFC in both functions (Kringelbach and Rolls 2004; Rudebeck and Rich 2018).

To date, no frontal leucotomy has been performed in the sheep, therefore we cannot assume or predict what effect such an experiment would have on the specific behavior in this species. However, there have been several recent non-invasive surveys using functional near infrared spectroscopy (fNIRS) to study brain emotions and functions in large mammal brains including sheep (Min et al. 2012; Gygax et al. 2013; Vögeli et al. 2014, 2015; Guldimann et al. 2015; Gygax and Vögeli 2016; Chincarini et al. 2020). Most of them did not point out the possibility that the fNIRS probe could not reach the frontal fields given the shape of the skull in the species used. While in the study of Vögeli and colleagues (2014) the sensor was too rostral to reach orbital cortices, Chincarini and collegues’ (2020) results were related to the motor areas instead of the prefrontal ones due to this anatomical feature. Nonetheless, based on our findings, the presence of connections with the fornix/hippocampus, claustrum, piriform cortex, hypothalamus, visual areas and the absence of U-shaped fibers, cingulum and fibers with other parts of the prorean gyrus (remaining part of the prefrontal cortex), we can infer that in the sheep the area responsible for reward and punishment elaboration and emotional control, the “orbitofrontal cortex” in primates, putatively partially share the same function.

So far, brain lateralization has been well established in humans as well as in some other species (Güntürkün et al. 2020). Although relatively weak, there seems to exist a certain lateralization of function within the orbitofrontal cortex (Lopez-Persem et al. 2020). Actually, reward and positive emotions seem to be more represented on the left hemisphere, whilst the punishment and negative emotions in the right hemisphere (O’Doherty et al. 2001; Alves et al. 2008; Leliveld et al. 2013; Mtui et al. 2020; Güntürkün et al. 2020).

Similarly, in the sheep, a differentiation between the right and left brain hemisphere has been demonstrated through behavioral studies. In fact, it is generally accepted that sheep face recognition, with its possible negative behavioral and emotional consequences (Da Costa et al. 2004; Knolle et al. 2017), are predominant in the right hemisphere (Peirce et al. 2000, 2001; Peirce and Kendrick 2002; Da Costa et al. 2004; Versace et al. 2007). Although from DTI we could not determine quantitatively the cortico-cortical fibers reaching the visual areas were conveying signal in or out from the OFC, the strong relationship between the two areas suggested a high specialization and the importance of one hemisphere’s function over the other. Being phylogenetically a prey, the sheep has an existential incentive to recognize other members of its flock from a dangerous situation, such as the presence of a wolf or another predator. The elaboration of this assessment might have brought this species to lateralize this function to the right hemisphere.