Experimental procedures were performed with male Wistar (N = 16) and female Hooded Lister (N = 2) rats (230–260 g). Animals were obtained from local breeding facilities of Ribeirão Preto and Sheffield laboratories respectively. The rats were kept under environmentally controlled conditions: temperature (23 ± 2°C), 12-h light/dark cycle (lights on at 7 AM), and had free access to water and standard laboratory diet. All animal husbandry and experimental procedures were performed in accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals (NIH Publications No. 80-23, 1996). United Kingdom experiments were conducted in compliance with UK Govt. Home Office approval under section 5 (4) of the Animals (Scientific Procedures) Act 1986. Experimental protocols also received prior approval of the relevant Institutional Ethics Committees. Effort was made to minimize the number of animals used and their suffering.

Rats were anesthetized with an intraperitoneal injection of a mixture of ketamine, xylazine, and acepromazine (100:20:3 mg/kg; União Farmacêutica Nacional S/A, Embu-Guaçu, São Paulo, Brazil; Laboratórios Calier S/A, Barcelon, Spain; Rhobifarma Indústria Farmacêutica Ltda, São Paulo, São Paulo, , Brazil) or a mixture of ketamine (Ketaset^®, 0.765 ml/kg) and xylazine (Rompun^®, 1.1 ml/kg). The sensitive retrograde tracer Fluoro-Gold (solution of 2.5 or 4% – Fluorochrome, Denver, CO, USA) was injected unilaterally into either the medial (AP 6.0 mm caudal to bregma; ML 0.7 mm; DV 3.2 mm from the brain surface) or lateral (AP 6.0 mm; ML 2.4 mm; DV 4.5 mm) sub-region of the SC (Paxinos and Watson, 2007). Deposits were made iontophoretically through a glass micropipette (tip diameter, 25 μm) using a constant-current generator (model CS3, Midgard Electronics, Canton, MA, USA) to deliver +2.5 μA, with 7-s on-off duty cycle for 7 min.

After a survival period of 7–12 days, the animals were deeply anesthetized with pentobarbitone sodium (Pentoject^®, 0.6–0.8 ml/kg) or a mixture of ketamine and xylazine (100:20 mg/kg, ip.) and perfused transcardially with a brief saline rinse followed by an ice-cold solution of 4% paraformaldehyde in a 0.1-M phosphate buffer at pH 7.4. Two hours later the brains were removed from the skull and cryoprotected with 20% sucrose in 0.1 M phosphate buffer at 4°C. Brains were frozen and coronal sections (30 μm) were obtained using a microtome (Leika Microsystems, Germany). For two cases, sections were collected directly onto slides, allowed to dry in a light-protected box and coverslipped with DPX mountant. An adjacent series was stained with cresyl violet to serve as an anatomical reference. The location of transported Fluoro-Gold in the tissue of these animals was determined by fluorescence microscopy.

Fluoro-Gold in all other cases was determined by immunohistochemistry performed according to published protocols (Schmued and Fallon, 1986; Pieribone and Aston-Jones, 1988). Briefly, sections were incubated with rabbit anti-Fluoro-Gold antibody (1:5,000; Chemicon International, Temecula, CA, USA, lot # 0606034308; 72 h) and washed in potassium phosphate-buffered saline (KPBS; pH 7.4). Next, sections were incubated in a solution of biotinylated goat anti-rabbit IgG (Vector Laboratories, Burlingame, CA, USA). After being washed, sections were incubated in ABC reagent (ABC Elite Kit; Vector Laboratories). The peroxidase reaction–product was visualized using the glucose oxidase procedure (Itoh et al., 1979) and the metal free 3,3′ diaminobenzidine tetrahydrochloride (DAB; Sigma, St Louis, MO, USA) as a chromogen. Sections were exposed for 10 min to a 0.02% chromogen solution containing 0.3% nickel-ammonium sulfate in 0.05 M phosphate buffer (pH 7.4) and glucose oxidase, followed by the addition of β-d-glucose to produce a blue–black product. The reaction was terminated by washing in KPBS, pH 7.4.

After processing, sections were mounted on gelatinized slides, dehydrated, and coverslipped with DPX (Sigma). An adjacent series was stained with thionin to serve as an anatomical reference for the distribution of Fluoro-Gold labeling.

Excluding the olfactory bulb, sections throughout the brain were analyzed for the presence of retrograde labeling. Non-processed sections were observed with a fluorescent microscope equipped with episcopic illumination (Nikon Eclipse E800M, Kingston-upon-Thames, UK) and UV excitation filter (330–380 nm). Sections subjected to immunohistochemistry were examined under brightfield illumination (Leika DM 2500). Fluoro-Gold deposits in the injection site and retrogradely labeled perikarya were plotted with the aid of a camera Lucida onto maps from adjacent thionin-stained sections. Brain structures were identified with reference to the stereotaxic atlas of Swanson (2004). Photomicrographs were obtained with microscope-mounted digital cameras and adjusted for contrast and brightness with Adobe Photoshop CS3 (version 10.0.1; Adobe Systems, Mountain View, CA, USA). Line drawings were prepared with Adobe Illustrator CS3 (version 13.0.1; Adobe Systems).

The pattern of layering within the rodent SC was identified according to the nomenclature established in previous investigations (Sahibzada et al., 1986; Telford et al., 1986; Coizet et al., 2007; Favaro et al., 2011): superficial layers (SZ, stratum zonale; SGS, stratum griseum superficiale; SO, stratum opticum); the intermediate layers (SGI, stratum griseum intermediale; SAI, stratum album intermediale); and the deep layers (SGP, stratum griseum profundum; and SAP, stratum album profundum; Figure 2). However, in this paper we follow the widely adopted convention of using the term “deep layers” to refer to both intermediate and deep layers. A line approximating the horizontal meridian bisected the SC into medial-upper and lateral-lower field sub-regions (Telford et al., 1986; Figure 2).

To provide an estimate of the potential importance of afferent control exercised over the medial and lateral sub-regions of the SC by different structures, we determined the relative densities of retrogradely labeled cells in afferent structures. Thus for each structure, a standard graticule (0.5 mm × 0.5 mm area = 0.25 mm²) was placed over the area within the structure that contained the maximum density of retrogradely labeled cells. The number of cells contained within the graticule was recorded. The resulting numbers were then converted to a four-point ordinal density measure: low (1–40 cells); low-medium (41–80 cells); high-medium (81–120 cells); and high (>120 cells). For the two illustrated cases these numbers were entered into Table 1.

This section describes the afferent structures that project to the SC deep layers, but make contact only with the medial sub-region. The most obvious general observation was that many structures located throughout the entire neuraxis make contact exclusively with this subdivision of the deep SC. These structures may be classified as follows.

The most obvious difference was that, compared to the medial deep layers of the SC, only a few afferent structures project exclusively to the lateral deep layers. Thus, a few retrogradely labeled neurons were observed in parts of the ipsilateral primary somatosensory cortex representing the upper limb and nose (Figures 5B′ and 6G), also in the gustatory, agranular insular, and visceral areas of the isocortex (GU, AI, Figure 5B′). In the brainstem, a moderate number of retrogradely labeled neurons were found only in the contralateral parvicellular reticular nucleus (PARN; Figures 5N′ and 6H).

In this section we will describe the afferent structures that project to both medial and lateral sub-regions of the deep SC, but that within the parent structure neurons retrogradely labeled from the medial and lateral SC were spatially separated. These topographically organized inputs originated from some isocortical areas, zona incerta, substantia nigra pars reticulata, and some restricted areas in the brainstem.

Only the lateral hypothalamic area and a very few structures in the brainstem contained a similar distribution of retrogradely labeled neurons following injections into medial or lateral sub-regions of the deep SC. Thus, the lateral hypothalamic area had only scattered labeled neurons located adjacent to the cerebellar peduncle at the tuberal level (LHA; Figures 5E,D′,E′). Moreover, moderate amounts of label with a similar spatial distribution were observed in the pedunculopontine nucleus (bilaterally for medial injections and ipsilaterally for lateral injections; PPN; Figures 5K,K′ and 8A,B). Also following both medial and lateral SC injections, moderate numbers of labeled neurons were observed bilaterally in the locus ceruleus (LC; Figures 5L,L′), and contralaterally in the cerebellar dentate nucleus (DN; Figures 5O,O′ and 8C,D). Finally, small numbers of retrograde labeled neurons were found ipsilaterally in the midbrain reticular nucleus including the retrorubral area (MRN, RR; Figures 5H,I,H′,I′), and contralaterally in the spinal nucleus of trigeminal (SPV; Figures 5N,O′, again following injections into both medial and lateral SC.

In this study we exploited the regional distribution of function in the SC (medial-defense; lateral-appetitive/approach) to identify neural structures that may contribute differentially to approach and avoidance responses. Our main findings were: (i) there was very little overlap in the afferent control of medial and lateral sub-regions of the rodent SC. (ii) Significantly more neural structures exercise afferent control over defense compared with approach associated regions of the SC. Thus, numerous sites in different parts of the cerebral cortex, visual thalamus, limbic regions, pre-tectum, and brainstem contained retrogradely labeled neurons only after injections into the medial SC. (iii) In structures that send projections to both medial and lateral SC, the majority (several cortical regions, zona incerta, pre-tectum, periaqueductal gray, substantia nigra pars reticulata, the parabrachial area, and principal sensory trigeminal nucleus) contained topographically organized retrograde labeling with little or no spatial overlap. (iv) Comparatively few structures (LC, retrorubral area, pedunculopontine nucleus, spinal sensory trigeminal nucleus, and cerebellum) contained overlapping distributions of retrogradely labeled neurons following injections into medial and lateral SC. A summary of these results is presented in Figure 9. We will discuss these observations, first in terms of technical issues that may influence their interpretation. We will then relate them to previous investigations of afferent SC anatomy, and conclude by considering their functional implications.

The retrograde anatomical tracing technique was used to provide a general overview of the sources of input to the separate functionally defined regions of the SC, also, a relative estimate of the potential importance among the range of afferent projections. We used iontophoretic application of the tracer Fluoro-Gold because of its sensitivity as a retrograde tracer. In particular, it provides discrete injection sites with minimal uptake of the tracer by non-terminating fibers of passage (Brog et al., 1993). Our use of this single highly sensitive tracer was to maximize the reliability of between case comparisons. That is, the variability associated with uptake and transport is likely to be more consistent for a single anatomical tracer, as opposed to when a range of different anatomical tracers are used. Moreover, given there was so little overlap in the afferent projections to the medial and lateral SC, we considered the use of double label techniques unnecessary. Our use of a maximum density measure of retrograde tracing provides an estimate of the potential density of terminating projections associated with medial and lateral injection sites (Table 1). Clearly, this measure can be used only as a guide as it takes no account the relative volumes of afferent structures, the overall numbers and distribution of labeled neurons they contain. Neither can the density measure accommodate slight differences in the sizes of the two injections, differences in the patterns of terminating fibers, either in terms of the degree of terminal branching, or the locations where afferent terminals make contact with targeted neurons – distal dendrites or proximal neuronal cell bodies. A finer resolution of these issues awaits further investigation.

With possibly one exception (projections from the postsubicular region), all the afferent structures we identified in the present study have been described by others previously (Wise and Jones, 1977; Killackey and Erzurumlu, 1981; Roldán and Reinoso-Suárez, 1981; Huerta et al., 1983; Cadusseau and Roger, 1985; Tanaka et al., 1985; Vertes et al., 1986; Henkel and Schneiderman, 1988; Rhoades et al., 1989; Sesack et al., 1989; Appell and Behan, 1990; Cornwall et al., 1990; Canteras and Swanson, 1992; Ohtsuki et al., 1992; Redgrave et al., 1992; Wiberg, 1992; Waterhouse et al., 1993; Canteras et al., 1994; Klooster et al., 1995; Krauthamer et al., 1995; Kolmac et al., 1998; Canteras and Goto, 1999; Garcia Del Caño et al., 2000; Moore et al., 2000; Olucha-Bordonau et al., 2003; Wyss and Sripanidkulchai, 2003; Kimura et al., 2004; Sefton et al., 2004; Kelly et al., 2009; Alloway et al., 2010).

The importance of the current investigation is the finding that, with very few exceptions, afferent structures, or sub-regions within afferent structures, project to either the medial, or lateral deep SC, but not to both. In previous anatomical studies in rodents there was a strong, entirely rational tendency to place injections of retrograde anatomical tracer centrally within the SC, both in terms of rostro-caudal and medio-lateral placement. Because these injections encroached into both medial and lateral sectors of the SC, previous investigations were able to identify most afferent structures, but missed the point that the medial-defense and lateral-approach territories of the rodent SC receive almost entirely from segregated populations of afferent neurons. The possibility that SC approach and defense systems receive inputs from anatomically separate systems will have important implications for the regulation of initial approach and avoidance responses elicited by sudden sensory events. However, as noted in the Introduction, functional segregation of the SC is likely to be a particular feature of animals whose predators and prey consistently originate in different parts of the visual field. Consequently, we would expect the segregation of afferents, according to targeted regions in the SC, to be less clear in species whose predators and/or prey frequently appear in the upper or lower visual fields (Swanson et al., 1974; Updyke, 1977; Roldán and Reinoso-Suárez, 1981; Huerta and Harting, 1982; Kawakura et al., 1982; Lynch et al., 1985; Room et al., 1985; Henkel and Schneiderman, 1988; Meredith and Clemo, 1989; Harting et al., 1992; Buchanan et al., 1994; May et al., 1997; Klop et al., 2005; May, 2006; Aronoff et al., 2010; Manger et al., 2010).

To understand the variety of afferent projections to the SC identified in this and other reports (Wise and Jones, 1977; Kawakura et al., 1982; Druga and Syka, 1984; Cadusseau and Roger, 1985; Tanaka et al., 1985; Meredith and Clemo, 1989; Buchanan et al., 1994; Ndiaye et al., 2002; Olucha-Bordonau et al., 2003; Klop et al., 2005; Manger et al., 2010), it is important to consider, briefly, the evolutionary history of the SC and the requirements for organizing effective defense and approach behaviors. The SC is an evolutionarily ancient sensorimotor structure found in the midbrain of the earliest vertebrates (Wurtz and Albano, 1980; Sparks, 1986; Dean et al., 1989; Stein and Meredith, 1993). With relatively minor modifications, it has been retained in a largely conserved form and is present in the dorsal midbrain of all modern mammals (Gaither and Stein, 1979; Stein, 1981, 1984; May, 2006). The need to defend against predators and approach sources of sustenance, mates, and in some cases offspring, are, however, also survival imperatives for primitive vertebrates that lack the flexibility conferred by the more evolutionarily recent expansion of the cerebral cortex. Consequently, there is evidence that the SC plays a prominent role in approach and avoidance functions of primitive species (Saitoh et al., 2007). We will therefore, first consider the sub-cortical afferent connections of the SC which are likely to have enabled the SC to perform effective approach and defensive functions in evolutionarily primitive vertebrates that lack a developed cerebral cortex.