Introduction
In a series of magisterial papers, Joseph Price and his colleagues characterized the frontal cortex and its projections in various species (Bacon and Smith
1993; Carmichael and Price
1994,
1996; Ongur et al.
2003). In the rat, the OFC was subdivided—medially-to-laterally—into ventral (VO), ventrolateral (VLO) and lateral (LO) regions (Krettek and Price
1977), which are analogous to the multimodal areas 14, 13a and 13 m/l, respectively, in the monkey (Price
2007). Area 13a, the so-called “visceromotor” region, is a constituent of the medial network (Carmichael and Price
1996), receiving inputs from limbic structures and projecting to the hypothalamus and the PAG. Area 13 m/l forms a part of the orbital network (Carmichael and Price
1996). Receiving multimodal information, it is consequently referred to as the “viscerosensory” region. It projects almost no descending axons to either the amygdala, the hypothalamus or the PAG (Ongur et al.
1998).
The LO/VLO cortex in rodents is therefore a homologue of part of the reward network region in the ventromedial prefrontal cortex (VM) (Ongur and Price
2000) that is involved in decision-making and emotional handling in primates (Bechara et al.
2000). In humans, lesioning of the VM interferes with the processing of somatic or emotional signals, thereby leading to an impairment in decision-making (Bechara et al.
2000). An influential hypothesis (“somatic marker”) postulates the VM to be a key link between emotional regions (amygdala) and the autonomic nervous system (Damasio
1996), of which the hypothalamus is the key organizer.
The projections from the OFC in rodents have been exhaustively investigated by various authors by means of antero- and retrograde tracing, The findings have revealed the OFC to be connected with the primary olfactory cortex (Price
1985), the piriform cortex (Illig
2005), the caudatoputamen (Beckstead
1979; Berendse et al.
1992; Gabbott et al.
2005; Groenewegen et al.
1990; Leonard
1969; Schilman et al.
2008), the amygdala (Groenewegen et al.
1990; McDonald et al.
1996), the extended amygdala (Groenewegen et al.
1990; Reynolds and Zahm
2005), the submedial (Coffield et al.
1992; Craig et al.
1982; Price and Slotnick
1983; Yoshida et al.
1992) and the mediodorsal thalamic nuclei (Beckstead
1979; Gabbott et al.
2005; Groenewegen
1988; Guldin et al.
1981; Leonard
1969; Price and Slotnick
1983; Ray and Price
1993; Reep et al.
1996), the parafascicular nucleus (Jones and Leavitt
1974), the claustrum (Zhang et al.
2001), the lateral hypothalamus (Gabbott et al.
2005; Hardy
1994; Hurley et al.
1991; Price et al.
1991), the PAG (Hardy
1986; Wyss and Sripanidkulchai
1984), the ventrolateral PAG (Beckstead
1979; Craig et al.
1982; Leonard
1969) and the oculomotor complex (Leichnetz and Gonzalo-Ruiz
1987b; Leichnetz et al.
1987a). The VLO-cortex projects additionally to the visual cortex (Reep et al.
1996). The targets of OFC projections are distributed throughout the entire brain—the hippocampus and the cerebellum exempted. Although the lateral hypothalamus and the periaqueductal grey matter (PAG) have been shown to receive terminals, their localization has not been identified with precision.
Using viral constructs (Chamberlin et al.
1998; Wickersham et al.
2007a) and genetically modified mice that express Cre-recombinase in parvalbumin- (Parv) (Hippenmeyer et al.
2005) and/or Foxb1-expressing neurons (Zhao et al.
2007), we demonstrate that the LO- and the VLO-cortices heavily project collaterals to three as yet less well-known structures, namely, the parvafox nucleus (Bilella et al.
2014; Celio
1990; Celio et al.
2013; Girard et al.
2011; Meszar et al.
2012), the supraoculomotor nucleus (Su3) (Carrive and Paxinos
1994) and the parvalbumin 2 nucleus (PV2) (Celio et al.
2013) of the PAG. The parvafox nucleus [formerly called PV1 nucleus (Celio
1990)], is located amongst the fibres of the medial forebrain bundle in the ventrolateral hypothalamus. The Su3 and the PV2 nuclei are two longitudinally oriented columns of neurons, which are located ventral to the aqueduct at the border of the PAG in the mesencephalic tegmentum.
Although the largest contingent of axon terminals derives from pyramidal cells in layer V–VI of the LO/VLO-cortex and is excitatory, parvalbumin-expressing and a few GABAergic neurons also contribute to the projections. By virtue of its dual inhibitory and excitatory projections, the LO/VLO-cortex may modulate the activity of the parvafox nucleus and its PAG targets.
Materials and methods
This study was conducted in accordance with the regulations of the Swiss Federal Animal Protection Law and under the supervision of the Veterinary Authority of the Canton of Fribourg (permissions 2013-04-Fr; 2013-05-FR, 2016-36-Fr).
Experiments were performed on 36 adult C57/Bl6 mice and 13 Wistar albino rats (Janvier, Lyon, France) of both genders, weighing 26–39 g and 240–325 g, respectively (Table
1). Other strains of mice that were used were either homozygous for the
Pvalb-Cre genotype [129P2-
Pvalb < tm1(cre)Arbr>/J] (Hippenmeyer et al.
2005) or heterozygous for the
Foxb1-Cre one [Foxb1
tm1cre−EGFPGabo] (Alvarez-Bolado et al.
2000) (Table
1).
TVA-floxed mice [B6;129P2-
Gt(ROSA)26Sortm1(CAG−RABVgp4,−TVA)Arenk/J, strain 024708] were used in the rabies experiments and
VGAT-Cre mice [
Slc32a1tm2(cre)Low l, strain 016962 (Jackson Laboratory, Bar Harbor, Maine, USA)] for anterograde tracing with Cre-dependent constructs. The mice that were used for the trans-synaptic rabies injections were of the TVA-
Pvalb/
Foxb1 genotype; they were bred in house (Table
1).
Table 1
Strains of rats and mice that were utilized to study the projections from the OFC
Wild type | Wistar | | Janvier, Lyon, France |
Wild type | C57/Bl6 | | Janvier, Lyon, France |
Pvalb-Cre | 129P2-Pvalbtm1(cre)Arbr/J | 008069 | Dr. Silvia Arber, Basel (Switzerland) and Jackson Laboratory |
Foxb1-Cre | Foxb1tm1cre−EGFPGabo | | Dr. Gonzalo Alvarez-Bolado Heidelberg, Germany |
TVA-floxed mice | B6;129P2-Gt(ROSA)26Sortm1(CAG−RABVgp4,−TVA)Arenk/J | 024708 | Jackson Laboratory |
VGAT-Cre | Slc32a1 tm2(cre)Low l | 016962 | Jackson Laboratory |
TVA-Pvalb-Cre/Foxb1-Cre | TVA-Pvalb-Cre/Foxb1-Cre | | Bred in house |
The animals were anaesthetized with a mixture of ketamine (40–60 mg/kg of body weight) and xylazine (10–15 mg/kg of body weight) which was diluted with physiological (0.9%) saline. If necessary, supplementary, lower (1/4–1/3) doses of the anaesthetic were administered during the stereotactic procedure, if any signs of awakening became manifest.
Anterograde tracing experiments (Table 2a)
Table 2
List of the antero- and retrograde tracers that were used for the experiments.
Viral, non-Cre-dependent |
1 | AAV2/1.hSynapsin.EGFP.WPRE.bGH | Rats and mice | Vector Core, University of North Carolina, USA |
2 | AAV1.hSynapsin.TurboRFP.WPRE.rBG | Rat | Vector Core, University of North Carolina, USA |
3 | AAV9.hSynapsin.TurboRFP.WPRE.rBG | Rat | Vector Core, University of Pennsylvania, USA |
Viral, Cre-dependent |
4 | AAV2/1.CAG.FLEX.EGFP.WPRE.bGH | Mouse | Vector Core, University of Pennsylvania, USA |
5 | AAV1.CAG.flex.tdTomato.WPRE.bGH | | Vector Core, University of Pennsylvania, USA |
Non-viral |
6 | Biotinylated dextran (MW 10′000) anterograde tracer | Mouse | Invitrogen (D1956 Lot 1148353) |
Trans-synaptic transport (rabies) |
7 | rAAV8/CA-Flex-RG.ape | Vector Core, University of North Carolina (USA) |
8 | rAAV5/EF1-Flex-TVA-Cherry.ape | Vector Core, University of North Carolina (USA) |
| Env-A Δ-G rabies-EGFP | Friedrich Miescher Institute-Basel (Switzerland) |
Others |
10 | Biotinylated dextran (BDA, MW 3000) retrograde tracer | Invitrogen (D7135), Waltham, USA |
11 | Fluorogold Anti-fluorogold antiserum | Fluorogold, Denver, USA Millipore AB153 Lot 2161122, USA |
The head of the animal was secured in the stereotaxic apparatus (Kopf Model 5000) and a craniotomy was performed over the target region in the orbital cortex. Tracers 1–6 in Table
2a were used as anterograde tracers in these experiments.
Injections in the OFC
The tracers were injected via a fine-bored needle (external diameter: 0.14 mm, GA: 34), which was connected to a 2.5-µl Hamilton syringe that was mounted on a manual microinjection unit (Kopf, model 5000). In rats, the injections were made at different sites of the LO around central coordinates of anteroposterior (AP): + 4.2, mediolateral (ML): ± 2.4, dorsoventral (DV): − 3.5 (in mm, with respect to the bregma level and the brain surface). If not otherwise specified elsewhere, the injections in mice were usually made around the stereotaxic coordinates of AP: + 2.8, ML: ± 1.3, DV: − 1.8. 20–80 nl of the tracer (Table
2a) was injected into rats and 15–20 nl into mice during an interval of 0.5–1 min. After the injection, the needle was left in place for 3–5 min to allow the tracer to diffuse at the injection site. The needle was then withdrawn, the skin over the skull was sutured and the animals were left to recover. In most of the experiments with rats, bilateral injections of the same or different tracers were made at various AP and ML coordinates. The position of the needle was varied such that deposits of label involved in entirety the medial, lateral and insular portions of the dorsal sulcal cortex (MO, VO, VLO, LO, DLO, AIV). Due to the intricacy and the small size of the various orbitofrontal regions, injections were almost never confined to one region alone and tracers often suffused adjacent cortical areas.
A note on the OFC tracing data found in the Allen Brain Atlas (ABA)
After its first appearance in 2013, subsequent editions of the Allen Brain Atlas (ABA) have reported an ever-increasing body of data that have been garnered from the stereotactic injection of viral tracers into the OFC of murine brains (suppl. Table. 1). We drew on some of the self-same viral tracers in our own experiments. The data that are presented in the ABA are based upon the implementation of an iontophoretic technique, which involves the injection of a viral tracer at two levels in the cortex, whence it attains all layers. The volume of the injected tissue is then calculated. In our experiments, a defined volume of the viral tracer was delivered in a single injection via a Hamilton syringe to the targeted site. Since our data accord well with those that are presented in the ABA, we presume that the different modes of delivery of the viral tracers (iontophoresis versus microinjection) had no impact on their accessibility to nerve cells.
Injections in the parvafox nucleus
For the injections into the parvafox nucleus of mice, the needle was positioned at bregma level: − 1.5 mm, ML: 1.3 mm, DV: 4.9–5.1 mm (Bilella et al.
2016).
Double injection of anterograde tracers in the OFC and in the parvafox nucleus
A red anterograde tracer was injected in the OFC and a green one in the parvafox nucleus (or vice-versa). In these specimens, we investigated the spatial relationship between the OFC endings and the endings from the parvafox nucleus in the Su3 and the PV2 regions of the PAG. A Cre-dependent red-construct was injected into the parvafox nucleus and a Cre-dependent (or non-Cre-dependent) GFP-tracer into the LO/-VLO-cortex. Eight PV-Cre (552-12, 553-12, 138-13, 222-13, 223-13, 357-14, 394-14, 395-14), three Foxb1-Cre (209-14, 223-14, 390-14) mice, and one PV-Cre/Foxb1-Cre (223-14) mouse were investigated.
Are subpopulations of the OFC-cells projecting to the parvafox nucleus inhibitory or Parv-expressing (Lee et al. 2014)?
A Cre-dependent tracer was injected in the OFC of VGAT-Cre or PValb-Cre mice to detect terminals on neurons of the parvafox nucleus and in the PAG. The animals used for these experiments were six Pvalb-Cre mice (356-14; 357-14; 394-14; 306-15; 307-15; 308-15) and three VGAT-Cre mice (234-16, 235-16, 236-16).
Do experiments with classical tracers confirm the results found with viral tracers?
In eight mice (250-13, 399-15, 400-15, 401-15, 402-15; 28-16, 29-16), biotinylated dextran (BDA 10,000 MW, Invitrogen, USA) was injected at various medio-lateral coordinates into the OFC. The distribution of the terminals in the parvafox nucleus and in the PAG corresponded exactly to the picture that was revealed after the injection of the viral tracers. Only injections that involved the LO/VLO-cortex disclosed the typical “pony-tail-like” terminal field in the parvafox nucleus.
The cortices at the medial and lateral boundaries of the orbital cortex were selectively targeted in various experiments (126-13 [prelimbic (PrL), medial orbital (MO)]; 185-13 [MO]; 186-13 [MO]; 130-15 [PrL]; 132-15 [infralimbic (IL)]; 552-12 [DLO]; 553-12 [DLO]; 556-12 [DLO]). None revealed targeted terminals in the parvafox- or in the Su3 and PV2-nuclei, only a diffuse innervation of the lateral hypothalamus and of other columns in the PAG. As demonstrated in the ABA-database, tracer injection in the AID, AIV and AIP did not show selective innervation neither of the parvafox nor of Su3 and PV2-nuclei (
http://connectivity.brain-map.org/).
In a few experiments, the injection needles pierced the OFC and penetrated the olfactory bulb (394-13, 395-13), the anterior olfactory nucleus (132-15; 309-15), the olfactory tract (386-13) and the olfactory tuberculum (386-13).
Retrograde tracing experiments (Table 2b)
To confirm the existence of the projections that were observed in the anterograde tracing experiments and to ascertain whether the projections to the different sites originate from the same or different populations of neurons, retrograde (including double) labelling experiments were performed (tracers 7–11 in Table
2b). These included also some double-labelling experiments targeting in the same animal the PAG and the parvafox (both targets of the OFC as the present study reveals). Two classical retrograde tracers, viz. Fluorogold [2% in physiological (0.9%) saline; 12–30 nl/1930s] and biotinylated dextran amine [(BDA) 10% in physiological (0.9%) saline; 40–100 nl/1–2 min] were injected into targeted regions of the LO/VLO-projections in mice, namely, into the PAG (coordinates of AP: − 4.1, ML:± 0.5, DV: − 2.7) and/or the parvafox nucleus (coordinates of AP: − 1.5, ML: ± 1.3, DV: − 4.9) using the same tools and procedures that are described above for the injection of the anterograde tracers.
In addition, trans-synaptic labelling of the parvafox nucleus was executed in
Pvalb-Cre/
Foxb1-Cre mice that had been bred with TVA-floxed (
Pvalb-
Foxb1-
TVA) mice. To this end, a Cre-dependent glycoprotein-deleted mutant strain of the SAD B19 rabies vaccine strain bearing an EGFP-insert (Env-A Δ-G rabies-EGFP) was concomitantly injected with AAV-Flex-G into the parvafox nucleus (Table
2b) (Wickersham et al.
2007a).
Histological procedures
After 6–8 days for retrograde, and 3–4 weeks for anterograde tracing experiments, deep anaesthesia was induced in the animals by the administration of a lethal dose of pentobarbital (150–200 mg/kg of body weight). They were then transcardially perfused, first with physiological (0.9%) saline and then with paraformaldehyde (4% in 0.1 M phosphate buffer, pH 7.4) the brains were excised and postfixed overnight in 4% paraformaldehyde. They were then submerged in a 30% solution of sucrose for cryo-protection. Using a cryomobile (Reichert-Jung), the brains were serially sliced into 40/80-µm-thick sections, which were collected in 0.1 M phosphate buffer containing 0.01% sodium azide. The sections were usually cut in the coronal (frontal) plane. However, in four brains (393-13, 142-14, 132-15, 307-15) the sections were prepared in the sagittal plane, and in three (386-13, 394-13, 141-14) they were cut in the horizontal direction. The sections were subsequently incubated with the appropriate antibodies to confirm the precision of the injection and the presence of neuronal endings in the region of interest, particularly around the cells of the parvafox nucleus. Series of sections that were derived from untreated mice and rats were stained with Nissl and exposed to an antibody against non-phosphorylated neurofilaments (SMI-32), which served as a neuronal marker (Franklin and Chudasama
2012) to define the subdivision of the prefrontal cortex. The sections were mounted on glass slides for histological inspection in either a Leica 6000 epifluorescence microscope [equipped with a Hamamatsu C4742-95 camera], a digital slide-scanner (Nanozoomer, Hamamatsu), or a Leica TCS SP5 confocal laser microscope. Staining with GFP, RFP, Tomato or Fluorogold was detected by virtue of the intrinsic fluorescence; that with BDA was revealed after exposure to streptavidin-Cy3.
Immunohistochemistry
The immunofluorescence technique and the immunoperoxidase reaction were performed as previously described (Gerig and Celio
2007; Meszar et al.
2012). In short, floating sections were incubated in 24-well plates with the primary antisera or antibodies, which were diluted in the range 1:1000–1: 5000 (Table
3). The efficacy of the antisera and the antibodies against Parv had been hitherto established in antigen-pre-adsorption experiments, by immunoblotting and by the absence of immunoreactivity in knock-out mice (see Table
3). Incubation with the biotinylated secondary antibody was followed by exposure to either streptavidin-Cy2 (Alexa460), -Cy3 (Alexa 550) or -Cy5 (Alexa 650).
Table 3
List of the antibodies and the antisera that were used in the experiments
PV235 | Purified carp parvalbumin | Swant Inc., Marly, Switzerland | Mouse monoclonal Lot 10-11F | 1:1000–5000 |
PV25 | Recombinant rat parvalbumin | Swant Inc., Marly, Switzerland | Rabbit polyclonal Lot 5.10 | 1:1000–1:5000 |
GP72 | Recombinant mouse parvalbumin | Swant Inc., Marly, Switzerland | Guinea pig polyclonal | 1:1000–1:5000 |
PVG213/214 | Recombinant rat parvalbumin | Swant Inc., Marly, Switzerland | Goat polyclonal | 1:1000 |
GFP | Recombinant peptide | Molecular Probes, Waltham, (USA) | Rabbit polyclonal | 1:3000 |
5-HT | Serotonin | Immunonuclear | Rabbit polyclonal | 1:2000 |
TH | Tyrosine-hydroxylase | Immunostar, Stillwater (USA) | Rabbit polyclonal | 1:10,000 |
SMI-32 | Non-phosphorylated filaments | Millipore, USA | Mouse monoclonal | 1:1000 |
VGlut 1 | Recombinant peptide | Synaptic System, Germany | Rabbit or mouse | 1:5000/1:20,000 |
VGlut 2 | Recombinant peptide | Synaptic System, Germany | Rabbit | 1:10,000 |
GAD | Recombinant peptide | Millipore, USA | Mouse | 1:2000 |
Electron microscopy
In one case (127-13) five coronal sections of the region of the hypothalamus in which positive terminals were observed around the parvafox nucleus were incubated with antibodies against GFP during 5 days (without Triton-X100). The immunostaining was continued with a biotinylated antiserum against rabbit-IgG (1 day) and followed by avidin-peroxidase (1 day). After washing, the enzymatic activity was revealed by incubating the sections with DAB-H2O2. The sections were further post-fixed with 2.5% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.3 and after washing, were exposed to 1% OsO4 (osmium tetroxide) in phosphate buffer for 2 h. Embedding took place in Epon. Semi- (0.5 µm) and ultrathin sections (60 nm, grey interference colour) were cut with a Reichert Ultramicrotome and mounted on one-hole grids. Uranyl acetate and lead citrate were employed for contrasting purposes. The region of the parvafox was searched for the presence of synapses between immunoreactive terminals and dendrites or cell bodies using a Jeol microscope.
Discussion
The hypothalamic target of the murine LO/VLO cortical regions—corresponding to the Area 13a and 13 m/l in primates—is the newly described parvafox nucleus and its targets Su3 and PV2 in the PAG.
By drawing on self-replicating virus-based tracing tools, which enhance the sensitivity of connectivity studies, we have defined three targets of the LO/VLO cortices that have been hitherto overlooked, namely, the parvafox nucleus which is located in the ventrolateral hypothalamus, and the Su3- and the PV2 nuclei which are serially located ventral to the aqueduct in the periaqueductal grey matter (PAG).
Both LO and VLO innervate these three targets, although partly in different portions: in the SU3-nucleus, for example, the VLO-projection is located more medially than the one arising in the LO-cortex. This medial location of the VLO-projection is also observed in other part of the brain like the caudatoputamen, and superior colliculi. In the ventral tegmental area, the projection from the LO-cortex is more important than the one deriving from the VLO-cortex. There are also specific projections for each one of these two subdivisions of the OFC. The VLO-cortex for example, innervates parts of the visual cortex (Reep et al.
1996), the dorsal raphe (DR), the dorsolateral column of the PAG, a region that is also target of the Foxb1-subpopulation of neurons of the parvafox nucleus (Bilella et al.
2016) and has a specific projection to the contralateral, dorsal parabrachial nucleus (LPBrel). The LO-cortex innervates the ethmoid thalamic nucleus, the basolateral amygdala (Groenewegen et al.
1990; McDonald et al.
1996) and the ventral part of the submedius nucleus (Craig et al.
1982). In addition, terminals of the LO/LO cortex were also observed in nuclei not mentioned in the previous literature, namely the Gemini nuclei, the lateral parabrachial nucleus, the pontine nuclei, the reticulotegmental nucleus, the reticular substance of the brainstem and the lateral horn of the spinal cord. No major differences in the projection patterns were detected between rats and mice and in both species the projections to parvafox, Su3 and PV2 were of comparable intensity.
Using a Cre-dependent rabies virus (Wall et al.
2010; Wickersham et al.
2007b) the afferences to the Parv- and the Foxb1-expressing neurons could be studied separately and selectively in the corresponding Cre-mice. The input to the Parv-expressing neurons that represent the core of the parvafox nucleus originates mainly in the LO-cortex, whereas the inputs to the Foxb1-expressing neurons derive from the medial (IL, PrL), the orbitofrontal (MO, VO, VLO, LO) and even the lateral prefrontal cortices. These projection patterns correspond well with those that have been revealed by retrograde tracing with peroxidase from the rat ventrolateral hypothalamus [(Allen and Cechetto
1993) their Fig. 5a].
Our study is the second of its kind in which the anterograde projections of the LO-VLO cortex throughout the entire brain have been mapped. In the other investigation, which appeared four decades ago, autoradiographic techniques were implemented to map the staining profiles of OFC-derived projections after the injection of tritiated amino acids (Beckstead
1979). In parallel to our study, a large body of data appertaining to stereotactic injections into the OFC have been published during the past 3 years in the ABA (
http://www.brain-map.org), some of which reveal patterns of labelling of the terminal fields that are almost identical to those that we observed. Our data and those that are recorded in the ABA largely confirm the previous, precise observations of Leonard (
1969), Beckstead (
1979) and Reep et al. (
1996) in the rat as well as for VLO in the cat (Craig et al.
1982). To the list of known targets, we append three chemically defined sites (the parvafox-, the Su3- and the PV2 nuclei). Our knowledge of its existence (Bilella et al.
2016; Celio et al.
2013; Meszar et al.
2012), guided our attention to recognize the parvafox, a small elongated neuronal aggregates as targets of the projections of the LO- and the VLO-cortices [“C
hance favours the prepared mind” (citation of a statement by Louis Pasteur)]. In the aforementioned autoradiographic study, the presence of terminals in a tiny oval region that lay along the optic tract was actually documented pictorially [injection 8, Fig. 6 (Beckstead
1979)], but not mentioned in the text. From its position, we presume it to be the parvafox nucleus. The cortical projection proceeded caudally to the ventrolateral PAG, namely, to a region that lay dorsal to the nucleus of the third nerve, which was referred to by Beckstead as a “subaqueductal portion”, and which approximately corresponds to the terminal field that we observed in the supraoculomotor region (the Su3 nucleus). A terminal field at precisely the same location has been described in feline brains after the injection of tracers into the VLOβ-region [Fig. 15 in Craig et al. (
1982)]. Beckstead observed the projection to terminate in the laterodorsal tegmental nucleus, probably in the region that we refer to as the PV2 nucleus, owing to its composition of parvalbumin-expressing neurons (Celio et al.
2013). In the publication by Jasmin (Jasmin et al.
2004), tracers were injected into a region which, according to Price’s group is probably still the LO-cortex but which the former authors defined as the AIV [images 1a and b in Jasmin et al. (
2004)]. The distribution of the terminal fields accorded with our own observations, although those in the lateral hypothalamic parvafox nucleus were more sparse (their Fig.
3F). The terminals that were detected in the “ventrolateral wings of the dorsal raphe nucleus” probably correspond to those in the region of the Su3-region (their Fig.
3G).
Using the mouse connectivity programme of the Allen Database (ABA;
http://www.brain-map.org) and entering as source structures various areas of the prefrontal cortex without any filter for mouse line or tracer type, the results corresponded well with our own data. For the purpose of our study, the most relevant injections in the ABA were the numbers 112423392 and 112306316, since they were performed in mice of the same wild-type strain (C57/BI6). But also injections into the genetically modified lines A930038C07Rik-Tg1-Cr (Cre-recombinase confined to layer V; no. 168164972) and Rbp4-Cre_KL100 (Cre-recombinase confined to layer V; no. 287769286) revealed patterns of projections that accorded with the observations in our own.
The precise targets of the OFC-derived projection in the PAG are, with a high degree of axial and radial specificity, the Su3 nucleus, which is located slightly dorsolateral to the oculomotor nucleus, and the parvalbumin-expressing ones in the PV2 nucleus, which is located in the posterodorsal raphe nucleus (PDR) and rostral portion of the dorsolateral tegmental nucleus (LDTg). With a view to mapping the frontal eye field, Leichnetz (Leichnetz et al.
1987b) injected retrograde tracers into the oculomotor nucleus, to which end, the needle was oriented slightly obliquely (their experiments OMR2 and OMR3), thereby perforating the supraoculomotor region (harbouring the Su3 nucleus). Retrogradely labelled neurons occurred primarily in the cortex of the dorsomedial shoulder (the putative rodent homologue of the primate frontal eye field), scatterings of stained cells were nevertheless observed throughout the entire OFC (Leichnetz et al.
1987b).
The Su3 nucleus is known to project to the contralateral rostral- and the caudal ventrolateral medulla, which together constitute the sympathetic cardiovascular control centre (Chen and Aston-Jones
1996; Van Bockstaele et al.
1989). The Su3 nucleus receives inputs from the medial cerebellar nucleus (fastigium) (Gonzalo-Ruiz and Leichnetz
1987; Gonzalo-Ruiz et al.
1990) and from the parvafox nucleus (Bilella et al.
2016; Celio et al.
2013) and is activated during predation [of insects as well as by cats (Comoli et al.
2012)]. The connections and functions of the PV2-nucleus are yet unknown.
The hypothalamus and the PAG, two regions of the brain that harbour the parvafox-, and the Su3- and the PV2 nuclei, are themselves interconnected. This is a general organization, which has been described for the lateral hypothalamus broadly and for the ventrolateral PAG in particular (Floyd et al.
2000,
2001). The parvafox nucleus, does not reciprocate the projections from the OFC (Bilella et al.
2016; Celio et al.
2013), either directly or indirectly, via an innervation of the mediodorsal thalamic nucleus. The findings of the double retrograde tracing experiments that were conducted by Gabbott and his colleagues (Gabbott et al.
2005), as well as by ourselves, indicate that the projection from the LO/VLO-orbitofrontal cortex is—at least in part—serial, with the same axon successively innervating multiple subcortical targets via collaterals.
Part of the projection from the LO/VLO-cortex stems from parvalbumin-expressing neurons, which represent a sub-population of cortical GABAergic cells (Celio
1986). Long-range-projecting neurons expressing NO-synthase have been observed to connect cortical areas (Tamamaki and Tomioka
2010) and parvalbumin-expressing GABAergic ones are known to project from the medial prefrontal cortex to the nucleus accumbens (Lee et al.
2014). These parvalbumin-expressing neurons in the OFC are well positioned for a “top-down” inhibitory control of subcortical processes (Fuster
2008). In our study, their GABAergic nature was suggested by the presence of GAD-immunoreactive axonal endings on neurons in the parvafox nucleus. However, injections of Cre-dependent tracers into
VGAT-ires-cre mice (in which Cre-recombinase is expressed in the bodies of GABAergic neurons), revealed the presence of only a few projections outside the cortex (not shown). In the ABA, the injection of a tracer into the VLO of a
Slc32a1-IRES-Cre mouse (VGAT) revealed no projecting axons (injection no. 309580102).
The rat LO-cortex has been hitherto regarded as a constituent of the orbital network (Krettek and Price
1977) and to be more of a “sensory” than a visceromotor region. In addition to many other cerebral sites that have been reported by various authors, we have demonstrated the LO-cortex to project to circumscribed horizontal columns of neurons in the parvafox nucleus of the ventrolateral hypothalamus in the SU3 and the PV2 nuclei of the PAG. These findings suggest that the subdivision of the OCF-cortex into sensory and visceromotoric regions may not be as absolute in rodents as it is in monkeys (Price
2007).
The lateral hypothalamic region in which the parvafox nucleus is located receives inputs from various olfactory regions and from the amygdala (Price et al.
1991). Each of these areas has reciprocal connections with the OFC, which the findings of our study have revealed to target, non-reciprocally, the neurons of the parvafox nucleus. Furthermore, the olfactory tubercle projects to the Gemini nuclei, which are also targeted by the LO/VLO-cortex-derived projection and by the parvafox. It remains to be established whether the parvafox- or the Gemini nuclei have any distinct olfactory functions. The olfactory projection to the ventrolateral hypothalamus may regulate autonomic or neuroendocrine functions or related behaviours (Price et al.
1991), or it may simply contribute olfactory information to be integrated with other influences.
In addition to its involvement in the processing of olfactory and gustatory information, the OFC also controls the cardiovascular and the respiratory systems (Fuster
2008). The pioneering work of electrophysiologists revealed the most prominent consequences of stimulating Area 13 of the OFC in primates to be manifested in the cardiovascular and the respiratory systems (Fuster
2008). The effects include changes in blood pressure, heart rate, cardiac dynamics, respiratory rate and skin temperature (Bailey and Sweet
1940; Chapman et al.
1948; Delgado et al.
1960; Hall and Cornish
1977; Kaada et al.
1953; Sachs et al.
1949; Spencer
1894), which can even lead to cardiac histopathology (Hall and Cornish
1977). An exploration of various parts of the brain with electrodes whilst stimulating the OFC with strychnine (neuronography) has permitted a mapping of the connections of Area 13 with the lateral hypothalamus, particularly with the ventrolateral region in which the median forebrain bundle resides (Sachs et al.
1949; Ward and McCulloch
1947). Interestingly, some of the autonomic effects that are evoked by stimulation of the OFC may be likewise elicited by stimulating the lateral hypothalamic region (where the parvafox is located) and the ventrolateral PAG (where the Su3 and PV2 nuclei are found), which receive these cortical afferences. (Allen and Cechetto
1992,
1993; Bailey and Sweet
1940; Chapman et al.
1948; Delgado et al.
1960; Fernandez De Molina and; Hunsperger
1962,
1956; Gelsema et al.
1989; Hall and Cornish
1977; Hess
1957; Kaada et al.
1953; Loewy
1991; Ruggiero et al.
1987; Sachs et al.
1949; Verberne
1996; Verberne and Owens
1998; Yasui et al.
1991).
In addition to its role in olfactory processing and in visceromotor activity, the OFC is best known for its involvement in the expression of emotion and in reward-driven decision-making (Bechara et al.
2000; Rolls
2000; Schultz et al.
2000). This co-habitation of sensory, autonomic and behavioural networks in the OFC permits the “integration of primitive autonomic mechanisms (such as are associated with instinctive urges or emotional reactions) with neural activities at the highest functional level of the brain” (Clark Le Gros and Meyer
1950).
Patients with lesions in the OFC [specifically of the VM], manifest an impaired ability to generate anticipatory “skin conductance responses” (SCR) to a conceived outcome of an action (Bechara et al.
2005; Damasio
1996). SCRs are emotional signals (somatic marker) that are generated by the activity of the autonomic nervous system, of which the hypothalamus is the main organizer (Hess
1947,
1957).
Future studies will reveal whether the OFC → parvafox → PAG network that we have delineated in our study is the scaffold on which the first-named region engages the autonomic nervous system. In analogy to the SCRs, the OFC → parvafox → PAG network could also affect the physiology of the cardiovascular and the respiratory systems (Rainville et al.
2006) and be involved in pathologies thereof that relate to disturbances in high mental activities (Pickering et al.
1996).