Defining hippocampal area CA2 in the fox (Vulpes vulpes) brain

Since 1959, the Russian Farm‐Fox study has bred foxes to be either tame or, more recently, aggressive, and scientists have used them to gain insight into the brain structures associated with these behavioral features. In mice, hippocampal area CA2 has emerged as one of the essential regulators of social aggression, and so to eventually determine whether we could identify differences in CA2 between tame and aggressive foxes, we first sought to identify CA2 in foxes (Vulpes vulpes). As no clearly defined area of CA2 has been described in species such as cats, dogs, or pigs, it was not at all clear whether CA2 could be identified in foxes. In this study, we cut sections of temporal lobes from male and female red foxes, perpendicular to the long axis of the hippocampus, and stained them with markers of CA2 pyramidal cells commonly used in tissue from rats and mice. We observed that antibodies against Purkinje cell protein 4 best stained the pyramidal cells in the area spanning the end of the mossy fibers and the beginning of the pyramidal cells lacking mossy fibers, resembling the pattern seen in rats and mice. Our findings indicate that foxes do have a “molecularly defined” CA2, and further, they suggest that other carnivores like dogs and cats might as well. With this being the case, these foxes could be useful in future studies looking at CA2 as it relates to aggression.

gene expression differences among the populations (Gulevich et al., 2004;Hekman et al., 2018;Popova et al., 1997;Rosenfeld et al., 2020;Wang et al., 2018). The genetic mapping studies identified genetic loci and gene networks likely underlying behavioral differences among tame, aggressive and conventional foxes, and the recently generated fox sequence assembly allowed direct comparisons between the fox and dog genomes (Kukekova et al., 2011;Kukekova et al., 2018;Nelson et al., 2017;Wang et al., 2018). The Farm-Fox experiment has also offered a unique opportunity to isolate specific changes in brain structure related to changes in agonistic and antagonistic behaviors (Hecht et al., 2021;Huang et al., 2015). The brain differences identified using high-resolution MRI included greater gray matter volume in both tame and aggressive animals compared with conventionally bred farm foxes. Among the several portions of the brain that were larger in both tame and aggressive animals was the hippocampus (Hecht et al., 2021).
Early studies on the neural basis of aggressive behavior had shown that damage to or removal of the hippocampus can suppress aggressive behaviors in cat and rodent models (Eichelman Jr., 1971;Ely et al., 1977;Siegel & Flynn, 1968;Watson Jr. et al., 1983), but the regulation is apparently complex and appears to rely on the location along the dorsal-ventral axis. For example, Watson, et al. demon-strated that stimulation of locations within the dorsal hippocampus can decrease aggressive behaviors in cats, whereas stimulation of the ventral hippocampus had the opposite effect (Watson Jr. et al., 1983).
Area CA2, in particular though, has been shown to play a pivotal role in enabling social aggression in mice (Leroy et al., 2018;Pagani et al., 2015). Thus, further study of CA2 in foxes bred for aggressive or tame behavior could be informative as to whether CA2's role in regulating aggression in mice generalizes across species.
As a first step toward this goal, we sought to determine whether CA2 could even be identified in foxes, as area CA2 has yet to be convincingly delineated in carnivores such as cats and dogs. In cats, the mossy fiber (MF) projections appear to extend all the way into the regio superior, and as far as the area of small pyramidal cells defined as CA1, suggesting that that cats lack a defined CA2 (Hirama et al., 1997;Laurberg & Zimmer, 1980). In the case of dogs, one study showed a small area likely to be CA2 in the dorsal hippocampus, but not in the ventral hippocampus (Amayasu et al., 1999). In another study in dogs, Hof et al. used Lorente de No's definition of CA2 as the area just beyond the ending of the MF axons, but they noted no specific defining features other than the emergence of calbindin positive CA1 neurons (Hof et al., 1996;Lorente de N o, 1934). In addition, studies on the banded mongoose and the domestic ferret failed to find evidence of a distinct CA2 (Pillay et al., 2021). Thus, the previous studies were inconclusive regarding the presence of CA2 in carnivores. Similarly in a study describing the hippocampus in pigs, the authors described no features discerning CA2, but rather an abrupt transition from the large, densely packed pyramidal cells in the regio inferior (CA3) to the small, loosely packed pyramidal cells in the regio superior (CA1) (Holm & West, 1994). We conclude that in many species, this transition (from CA2) to CA1 can be evident in a Nissl stain, but the differences between CA2 and CA3 areas are less clear, and defining the transition is hampered by the classic definition by Lorente de N o which does not align with many of the molecular markers of CA2 used recently Lorente de N o, 1934); (Radzicki et al., 2023).
In this study, we aimed to identify CA2 in foxes by using immunostaining for molecular markers such as Purkinje cell protein 4 (PCP4), RGS14, and STEP, which have been shown to be highly expressed in CA2 pyramidal cells in mice and rats (Kohara et al., 2014;Radzicki et al., 2023;San Antonio et al., 2014). Besides being critical for future experiments comparing aggressive and tame foxes, the results of this study may provide a useful method for identifying CA2 in other species.

| Animals and samples
Brain tissue was collected from farm foxes which were bred and maintained at the experimental farm of the Institute of Cytology and Genetics (ICG) of the Russian Academy of Sciences in Novosibirsk, Russia. All animal procedures at the ICG were performed in accordance with standards for humane care and use of laboratory animals by foreign institutions, the Office of Laboratory Animal Welfare Assurance F16-00180 (A5761-01). Samples were collected from 1.5-year-old, sexually naive male and female foxes, although at this stage of the study, no attempt was made to assess tissue for sex differences or differences in lineage as all sections had roughly similar staining patterns. Images presented in Figures 1-3 were from a male fox (51), bred for aggressive traits. Images in Figure 4  it is possible that we might have had better staining with some antibodies using PFA fixation, we did have excellent preservation as evident by staining of neuronal structures such as dendrites and axons, which were visible in many of our images.
After perfusion, the brain was removed, placed in a 10% solution of neutral buffered formalin and stored for 1-4 days at 4 C. Hippocampi with surrounding temporal lobe tissue were dissected from right hemispheres and each hippocampus cut into three pieces (dorsal, middle, and ventral) prior to postfixing in 4% PFA solution in PBS (Thermo Scientific, Waltham, MA) at 4 C for 7-10 days. Samples were transferred to PBS and stored at 4 C for 10 days with one PBS change before shipping to NIEHS.

| Staining and microscopy
Tissue was dissected away from the hippocampus and blocked perpendicular to the long axis for cutting on a vibratome (Leica VT1200) at 40 μm thick. Hippocampi from dorsal, middle, and ventral levels were sliced and sections stored in PBS with sodium azide until use.
For Nissl staining, sections were mounted on microscope slides and allowed to air dry. Sections were then stained for 5 min in 0.1% Cresyl violet acetate in water, followed by a brief rinse in water. Sections were then dehydrated in 50 and 70% ethanol (EtOH) for 5 min each and then differentiated in five drops of glacial acetic acid in 95% EtOH. Sections were further dehydrated in sequential EtOH solutions (5 min each at 95, 100, and 100%) and 3 Â 5 min in xylenes before cover-slipping with Permount (Fisher Chemical).
For immunostaining, free-floating sections were blocked for 1 h on a shaker at room temperature with 5% Normal Goat Serum (NGS) in PBS containing 0.1% triton X100 (PBST). Sections were incubated overnight at 4 C with primary antibodies or biotinylated Wisteria Floribunda in the same solution. After washing 2 Â 15 min in 0.1% PBST, sections were incubated with secondary antibodies in 5% NGS in 0.1% PBST for 2 h at room temperature, followed by 2 Â 15 min washes in PBST and 2 Â 10 min in PBS Antibodies and concentrations are listed in Table 1. Slides were mounted and cover-slipped with Vectashield Hardset mounting medium with DAPI (Vector, H-1500-10).
For immunostaining in combination with fluorescent Nissl staining, 1% fish gelatin was used instead of NGS for the blocking and for primary and secondary antibody incubations. Following immunostaining, sections were incubated for 20 min in NeuroTrace 500/525 green fluorescent Nissl stain (Invitrogen, N21480), which was diluted in PBS at 1:250. Sections were washed for 10 min in PBST and 2 Â 5 min in PBS, before washing overnight in PBS at 4 C.
For visualization of the Nissl stain in brightfield, images were captured on a Zeiss AxioObserver Z1 microscope (Carl Zeiss Inc, Oberkochen, Germany) using an EC Plan-Neofluar 10Â/0.3 Ph1 objective F I G U R E 1 Possible location of CA2 in hippocampus as assessed by Nissl stain. (a) Example of a section from a dorsal hippocampus from a male fox (51) stained with Cresyl Violet to visualize the pyramidal cell layer. Higher power images from the areas indicated with dashed boxes are shown on the right. Note that the smaller, lighter stained neurons can be identified as being in CA1, but that the borders between CA1 and CA2 and between CA2 and CA3 are unclear. Scale bars are 1000 and 100 μm. (b) Example of a section from the mid-hippocampus (also from fox 51), stained with a rabbit antibody raised against PCP4, left (red), and NeuroTrace green fluoresence Nissl stain, center (pseudo-colored magenta). The merged image is on the right. Scale bar is 100 μm.
with a halogen light source. Transmitted light was collected with a Zeiss AxioCam 305 color camera with an exposure of 4.8 ms. Epifluorescence images were taken on a Zeiss AxioObserver.Z1 microscope using a HXP 120 V metal halide light source for excitation coupled with an AxioCam 705 camera and a Plan-Apochromat 10Â/0.45 M27 objective. Multiple fluorescence channels were taken using the following Zeiss filtersets for fluorescence excitation/emission: Filterset 49 for UV/blue, 38 HE for green, 63 HE for red. Confocal images were taken with a Zeiss LSM880 using the 488 nm, and 561 nm laser lines for excitation paired with emission filter settings of 499-553 nm and 562-647 nm, respectively. A C-Plan-Apochromat 40X/1.3 Oil DIC objective was used for image collection. Image brightness and contrast and pseudo-coloring were adjusted for illustrative value and are not to be considered valid for comparisons between sections.

| Nissl staining of fox hippocampus
Most previous work describing the hippocampal cytoarchitecture in carnivores has been performed using staining for Nissl bodies or a comparable method (Amayasu et al., 1999;Amrein & Slomianka, 2010;Hirama et al., 1997;Pillay et al., 2021). Similar to what has been observed in other carnivores, we found that Nissl staining alone was not definitive in delineating the cornu ammonis Thus, although the location of a possible CA2 in the fox was inferred from previous studies examining MFs in hippocampi from dogs and cats (Amayasu et al., 1999;Gall, 1990), one could reasonably place the borders elsewhere if examining only cell morphology and density.

| PCP4 labels a population of neurons likely representing CA2
Antibodies raised against several different proteins are now known to reliably label CA2 in mice (Radzicki et al., 2023). In fact, antibodies against RGS14 label CA2 in a number of species including mice and rats and human and nonhuman primates ( Lee et al., 2010), and Montanez-Miranda et al., 2022). Mouse monoclonal antibodies against RGS14, however, showed no evidence of labeling CA2 or elsewhere in the fox hippocampus (not shown). Similarly, staining with a monoclonal antibody against the mineralocorticoid receptor, which does label mouse CA2 nuclei, did not label CA2 neurons in a way that could be distinguished from autofluorescence emanating from the same neurons (not shown). This finding was not surprising, as antibodies, particularly monoclonal antibodies, are often targeted against only one region of a protein, and protein epitopes that differ substantially between different species may or may not bind to the antibody. Therefore, we took advantage of a polyclonal antibody from rabbits raised against the protein PCP4, which labels CA2 pyramidal neurons in rats and mice (Kohara et al., 2014;San Antonio et al., 2014). Antibodies raised against PCP4 also label dentate gyrus granule neurons and hilar mossy cells, but to a lesser degree. We found that antibodies recognizing PCP4 clearly labeled a population of pyramidal neurons in an area that spanned the end of the dense Nissl stain, as well as staining a few isolated neurons in CA1 ( Figure 1b). However, the PCP4-labeled neurons did not reveal any structural differences between the presumed CA2 and the CA3 neurons visualized with Nissl staining.

| Staining for calbindin and perineuronal nets
For the purpose of defining CA2 in the fox hippocampus by any definition, classical or by molecular characteristics, the precise location of the MF axons from dentate gyrus granule neurons would be desirable.
Timm's stain is most typically used for this purpose, but the technique requires that sodium sulfide be used at the time of perfusion (Sloviter, 1982). Another method of staining MFs is the use of antibodies that specifically recognize proteins enriched in the dentate granule cell axons. Examples of commonly used antibodies include those raised against Znt3 or calbindin (Palmiter et al., 1996;Rami et al., 1987). Attempts to stain fox hippocampus with antibodies recognizing Znt3 were unsuccessful (not shown). On the other hand, immunofluorescence staining using a rabbit polyclonal antibody recognizing calbindin gave a strong signal that labeled the dentate gyrus granule neurons and their axons as well as CA1 pyramidal neurons, similar to what has been reported in mouse, rat, and dog brain ( Figure 2, (Hof et al., 1996;Rami et al., 1987)). Because of the slight overlap of the calbindin positive dentate granule cell axons with calbindin positive neurons in CA1, this antibody was less than optimal for determining the exact extent to which MF axons might be covering. Nevertheless, one could gain an appreciation for the approximate end of these axons in the stratum lucidum.
One of the most convenient stains for CA2 in mice is the lectin from Wisteria Floribunda agglutinin (WFA), which labels the specialized extracellular matrix called perineuronal nets (PNNs) (Bruckner et al., 2003;Celio, 1993;Hartig et al., 1992). Notably, the stratum lucidum occupied by the MFs is devoid of this stain in mice, where WFA does label CA2 pyramidal cell dendrites and cell bodies (Carstens et al., 2016). Upon staining fox hippocampus with biotinylated WFA,  Table 2). Nevertheless, we conclude that most of these PCP4-positive neurons reside in what is likely hippocampal area CA2 and that PCP4 can be used to identify the location of CA2, or at least those homologous to mouse CA2 neurons.
To examine (1) the exact relationship between the MF terminals and the PCP4-positive pyramidal neurons, and (2) determine whether PCP4 is co-expressed with calbindin in presumed CA1 neurons, we used a polyclonal antibody recognizing human PCP4, raised in guinea pigs, together with the rabbit antibody raised against calbindin. Consistent with our observations using the rabbit antibody against PCP4 and WFA to outline the stratum lucidum, this guinea pig antibody against PCP4 recognized the same CA2 neurons and isolated neurons in CA1 in ventral hippocampi, as was labeled with the PCP4 antibody from rabbits (Supplementary Figure 2). Also similar was that the PCP4-positive neurons either overlapped with the area of calbindin positive (mossy) fibers or did not (Figure 4). Additionally, we found that although many of the PCP4-positive neurons intermingled with the calbindin-positive, presumed CA1 pyramidal neurons, the two stains were usually mutually exclusive upon examination at higher magnifications, with some exceptions noted (Figure 4). These results demonstrate that the relationship between PCP4-positive pyramidal F I G U R E 2 Mossy fibers from the dentate gyrus can be identified in a fox (51). Sections from the dorsal hippocampus were stained for calbindin (white, left), Wisteria Floribunda agglutinin (WFA; cyan, center left), and DAPI (magenta, center right). The merged image is shown on the right. A higher power image of the calbindin stain is shown below and better illustrates the mossy fiber axons and the positive pyramidal neurons in CA1. The presumed ending of the mossy fiber axons is indicated with the arrow. Although the neuropil stained with WFA is not particularly enriched in CA2, as in mice, it outlines the calbindin positive mossy fibers, providing a useful option for identifying the location of the dentate gyrus axons. Scale bars are 1000 and 100 μm.
neurons and the MFs in foxes are very similar to that in a number of rodent species such as rats and mice.

| DISCUSSION
In many mammalian species, including those as diverse as rodents and canines, neurons in area CA1 can be identified with antibodies recognizing calbindin (Pillay et al., 2021;Sloviter, 1989), and CA1 neurons in foxes are no exception. That said, several species differences, evident with the use of calbindin antibodies, have been noted regarding the cytoarchitecture of CA1. For example, in dogs, calbindin-positive neurons appear in the form of a double pyramidal cell layer in CA1 (Hof et al., 1996), but in mice, a distinct calbindin-positive pyramidal cell population appears in CA1's stratum pyramidale closest to the stratum radiatum (Dong et al., 2009;Kohara et al., 2014). In any case, staining for calbindin can serve as a useful marker of CA1 pyramidal cells, necessary for distinguishing CA1 (often the regio superior) from CA2 and CA3 (often the regio inferior). Some exceptions might be found in human and nonhuman primates, where several authors have attributed calbindin staining to CA2 neurons (Seress et al., 1991;Seress et al., 1993;Sloviter et al., 1991), but a comparison with markers such as RGS14, which does stain human CA2 Squires et al., 2018), would be informative. These findings would not be inconsistent with our data showing some mixing of calbindin positive and PCP4 positive neurons, however.
The border between area CA2 and area CA3 presents a different problem because the cell sizes in the two areas are very similar (i.e., both typically larger than those in CA1), and until recently, specific molecular markers for either CA2 or CA3 pyramidal neurons had not been particularly available. In addition, the absence or presence of MF axons in the stratum lucidum has been used as the primary way to distinguish CA2 from CA3, by the classical definition (Lorente de N o, 1934). However, recent work using protein expression has been consistent with the drawings of mouse hippocampus by Ram on y Cajal, showing large (CA2) neurons, most of which receive MF input, and yet lack the thorny excrescences characteristic of CA3 neurons in mice (DeFelipe & Jones, 1988;Kohara et al., 2014;Ram on y Cajal, 1909). Thus, if one were to accept a modern definition of CA2 that relies on molecular markers rather than simply the lack of the granule cell axons, which tend to vary substantially in length by species, identification of either CA2 or CA3 markers is critical to distinguishing these two populations of pyramidal cells. In fact, because the MFs appear to reach all the way into the regio superior not only in foxes, but F I G U R E 3 Antibodies to PCP4 stain a population of neurons that likely represents CA2. Sections from (a) dorsal, (b) middle, and (c) ventral hippocampus from a fox (51) stained using a rabbit antibody recognizing PCP4 (red, left) and Wisteria Floribunda agglutinin (WFA) staining PNNs (cyan, middle left). Merged images are shown on the right. As in mice, PCP4 stains a population of neurons that appear to either receive mossy fibers (evident as the lack of WFA stain in the stratum lucidum), or not. Scale bars are 1000 and 100 μm (rightmost images). also in cats and hedgehogs, for example (Laurberg & Zimmer, 1980;Rami et al., 1987), the mere presence or absence of these presynaptic structures (MFs) may not necessarily align with postsynaptic cell morphology (thorny excrescences) (Gaarskjaer, 1986). Here, we describe evidence that indeed in foxes, like in mice, the location of a molecularly defined CA2 spans the end of the stratum lucidum, and is not limited to those cells lacking MF input. As an aside, although we do not see evidence of the "islands" of MF staining like in cats (Hirama et al., 1997), we did observe an apparent splitting of the MF tract in CA2 and CA3 in some sections. Further work will be required to determine whether this divergence is consistent, or whether it is dependent upon plane of section or other factors.
Some of the earlier studies investigating different protein or mRNA localization in rat and mouse brain had noted high expression in what appeared to be CA2. They included neurotrophin-3, Basic Fibroblast Growth Factor (aka FGF2), and alpha-actinin-2 (Emoto et al., 1989;Phillips et al., 1990;Wyszynski et al., 1998).
Studies specifically focused on identifying region-specific hippocampal genes have yielded even more information on the molecular nature of CA2 neurons as a distinct population (Farris et al., 2019;Lein et al., 2005). Recently, the commercial availability of antibodies raised against STEP, RGS14, and PCP4 have made the study of CA2 more feasible (Boulanger et al., 1995;Kohara et al., 2014;Lee et al., 2010;San Antonio et al., 2014), and the expression of some "CA2 genes" like Amigo2, Avpr1b, and Cacng5 has made functional studies possible by using region-specific expression of cre recombinase (Alexander et al., 2018;Boehringer et al., 2017;Hitti & Siegelbaum, 2014;Williams Avram et al., 2019). For example, such studies in mice have strongly implicated CA2 neurons in social memory and social aggression (Leroy et al., 2018;Pagani et al., 2015;Smith et al., 2016). Thus, although enormous strides have been made in addressing function of CA2 in mice and rats, work on CA2 in other animals has lagged, likely in part, due to a lack of agreement on even its existence. Agreement on the location of F I G U R E 4 PCP4 positive neurons are typically calbindin negative. Confocal images of dorsal hippocampus from a female fox (75) stained using antibodies recognizing PCP4 (guinea pig, red) and calbindin (rabbit, white)  in different species of mole rats with different social structures (Stöber & Oosthuizen, 2023).
In this study, we found that PCP4 was one protein that could be reliably detected in fox CA2. Also known as PEP-19, PCP4, as its name implies, is notable for its expression in cerebellar Purkinje Cells, but it is also expressed in the dentate gyrus granule neurons, olfactory bulb, and caudate putamen in addition to CA2 pyramidal neurons in rats and mice (Sangameswaran et al., 1989). PCP4 has also been reported to be expressed in the entorhinal cortex of a nonhuman primates (Ohara et al., 2021). Its function in these brain areas is unknown, but the protein has been associated with regulating calcium and calcium dependent signaling through calmodulin in CA2 and elsewhere, generally by limiting calcium and calmodulin activity (Kubota et al., 2008;Putkey et al., 2003;Simons et al., 2009). Thus, it comes as some surprise that PCP4 can positively regulate neurotransmitter release and neurite outgrowth in cell lines (Harashima et al., 2011). Interestingly, PCP4 mRNA and protein levels can be regulated by steroid hormones such as estrogen and steroid hormone receptors, like the mineralocorticoid receptor (Gocz et al., 2022;McCann et al., 2021). Additionally, PCP4 is thought to be one of the genes affected in Down Syndrome (Hubert & Korenberg, 1997). Even though PCP4 is expressed in select regions throughout the brain in a number of mammals, its role in CA2 could be in contributing to CA2 synapses' distinct "plasticity profile" Simons et al., 2009).
Although the two antibodies against PCP4 were the only ones that we found effective at labeling CA2 neurons in fox hippocampus among several tested; two antibodies against RGS14 failed to stain CA2, and one against STEP stained what looked to be CA2, but the high background made it less than convincing (summarized in the staining (Radzicki et al., 2023). In the particular case of PCP4, staining of such cells is evident in CA1, particularly in the ventral hippocampus, with the degree appearing to differ in the different species. Identification of another marker of CA2 and double labeling could help to clarify the issue to some extent, but even the lack of thorny excrescences is not an absolute definition of CA2 pyramidal neurons in mice (Radzicki et al., 2023). In addition, we found evidence of mixing of PCP4 positive and calbindin positive cells at the presumed CA1-CA2 transition, so even there, the borders are apparently not absolute. Our findings suggest that although the molecular profile of neurons in CA2 can be useful in considering how certain neurons may behave in terms of their capacity for and type of synaptic plasticity, or how they may be similar in different species, the profile alone may not be sufficient to define an area. Evidence of functional variation within CA2 in mice, both in the proximodistal axis and between the deep pyramidal cell layer (closest to the stratum oriens) and superficial areas (closest to the stratum radiatum), has been reported (Fernandez-Lamo et al., 2019;Loisy et al., 2022); however, identification of molecules characterizing these differences are lacking as yet (discussed by ). Of note, though, we provide here evidence that a defined CA2 can be identified in the ventral hippocampus of foxes; identification of the same in ventral hippocampi of mice has proved somewhat controversial, but likely depends on the plane of section (Bienkowski, 2023).
A number of differentially expressed genes in prefrontal cortex of aggressive and tame foxes have been reported , and some of those genes are highly expressed in adult human or mouse hippocampus, including CA2 (Farris et al., 2019;Hawrylycz et al., 2012;Lein et al., 2007). These include both the serotonin receptor HTR3A, which is higher in prefrontal cortex of tame foxes, and the serotonin receptor HTR7, which is higher in prefrontal cortex from aggressive foxes. Whether these receptors, or more likely, those genes which are important for brain development, can contribute to the increased volume of the hippocampus and other brain regions in these experimental foxes, is yet to be determined (Hecht et al., 2021).
Determining where CA2 is in foxes, though, is a first step in being able to assess whether CA2 is altered in these animals.

ACKNOWLEDGMENTS
The authors would like to thank the expert staff of the NIEHS Fluores- Nissl stain, and members of the Dudek lab for critical reading of the