Honeycomb-like structure of the intermediate layers of the rat superior colliculus, with additional observations in several other mammals: AChE patterning

The concept of modularity in the brain arose from the seminal work of Woolsey and Van der Loos (1970), Asanuma and Rosen (1972), Hubel and Wiesel (1977), and Kaas et al. (1981), based on the existence of cytoarchitectonic and functional subunits in the cerebral cortex. Modules were designated for decomposition of both the proximal and distal sensory world into physical information such as the location, color, orientation, and movement of stimuli on the one hand and on the other, for fine muscular command.

During the last two decades, physiological and anatomical studies have provided convergent evidence for a modular arrangement of the deep superior colliculus (SC) in a variety of mammals (including primates, carnivores, and rodents). The SC or tectum is a laminated structure in the roof of the midbrain containing alternating cellular and fibrous layers (Kanaseki and Sprague, 1974; Tokunaga and Otani, 1976). Based on histological and functional considerations, the structure is classically divided into superficial and deep parts (Casagrande et al., 1972; Edwards, 1980). The superficial SC comprises three main layers: the stratum zonale, the stratum griseum superficiale (SGS), and the stratum opticum (SO). These layers are primarily involved in visual functions (see review by Stein and Meredith, 1991).

Briefly, they received their main inputs from the retina, visual cortex, lateral geniculate, and parabigeminal nuclei (Graybiel, 1972, 1975; Swanson et al., 1974; Hubel and Wiesel, 1977; Chalupa and Rhoades, 1979; Edwards et al., 1979; Graham, 1982; Mize, 1983). Their efferent connections are in turn directed to vision-related stuctures in the diencephalon (the lateral pulvinar/posterior complex, the geniculate nucleus), the pretectal area, and the deep SC (Caldwell and Mize, 1981; Holstedge and Collewijn, 1982; Huerta and Harting, 1983; Sugita et al., 1983; Grantyn et al., 1984; Rhoades et al., 1989). In the superficial SC, neuronal reactivity is predominantly related to onset-offset of stationary visual stimuli and also to moving cues (Schiller and Koerner, 1971; Goldberg and Wurtz, 1972; Tiao and Blakemore, 1976). The deep SC is composed of the stratum griseum intermediale (SGI), the stratum album intermediale (SAI), the stratum griseum profundum (SGP), and the stratum album profundum (SAP). As extensively reviewed elsewhere (Huerta and Harting, 1984; Grantyn, 1988; Sparks and Hartwich-Young, 1989; Stein and Meredith, 1991), these layers are mainly involved in multimodal integration and premotor functions. They process a variety of sensory, motor, and associative information arising from about 40 cortical and subcortical centers.

Concerning the sensory reactivity in the deep SC it is, by contrast with that of the superficial layers, visual, somatic (including tactile, proprioceptive, and nociceptive submodalities), and auditory. These inputs converge to these layers in a topographic fashion, and the various sensory maps they form are shown to be functionally registered (Gordon, 1973; Dräger and Hubel, 1975; Stein et al., 1976b; Tiao and Blakemore, 1976; Chalupa and Rhoades, 1977; Finlay et al., 1978; Donaldson and Long, 1980; Harris et al., 1980; McHaffie et al., 1989; Redgrave et al., 1996). In rodents for instance, the tactile and nociceptive reactivities of the face and the nasal visual field overlap in the rostral SC; the limb and trunk representations coincide with the auditory and temporal visual fields in the caudal SC. Likewise, the upper visual field and the tactile representation of the upper head, neck, and back are superimposed in medial SC, and lower visual and somatosensory fields are mapped together in the lateral portion of the structure (Dräger and Hubel, 1975; Tiao and Blakemore, 1976; Chalupa and Rhoades, 1977; Finlay et al., 1978). Thus, visual and/or somatic and/or auditory information coming from a given location in the environment appears to be processed in the same area of the collicular sensory matrix and even to converge at the level of single neurons (Meredith and Stein, 1986; Stein, 1998). A possible explanation for such a sensory alignment would be to ensure a coherent superposition of the physical descriptors of multimodal objects in the environment.

The deep SC also contains premotor neurons that, through extensive projections to motor-related structures in the brainstem, the spinal cord, and the diencephalon, promote orienting responses (Harting, 1977; Harting et al., 1980; Rhoades and Della-Croce, 1980; Holcombe and Hall, 1981; Grantyn and Grantyn, 1982; Murray and Coulter, 1982; Chevalier and Deniau, 1984; Redgrave et al., 1986; Yamasaki et al., 1986; Cowie et al., 1994; Muto et al., 1996). In monkey, they are predominantly gaze orientation involving eye or combined eye-head movements (Robinson, 1972; Stryker and Schiller, 1975; Cowie and Robinson, 1994), whereas in cat and rodents, they also comprise movements of pinnae, whiskers (Syka and Radil-Weiss, 1971; Harris, 1980; McHaffie and Stein, 1982) and even of the entire body (Northmore et al., 1988; Dean et al., 1989). The direction and amplitude of orienting movements are spatially encoded in the deep SC. Focusing on gaze control, which has been the most studied feature of the collicular physiology, early stimulation studies led to the concept of a motor map in deep SC in cat (Syka and Radil-Weiss, 1971; Stein et al., 1976a; Roucoux et al., 1980; Harris, 1980; Paré et al., 1994), monkey (Robinson, 1972; Stryker and Schiller, 1975), and rodents (McHaffie and Stein, 1982). To summarize, localized electrical stimulation of the intermediate layers induces foveating visual “grasping” by conjugate contraversive eye movements toward a particular region of the contralateral visual space.

The direction and amplitude of saccade basically depends on the location of the stimulated site. For example, stimulation in the anterolateral region of SC, produces forward saccades with a downward component, whereas caudomedial stimulation induces backward saccades with a clear upward component. Concerning more specifically the amplitude of saccades, it has generally been observed that they are small when evoked from anterior sites and large and combined with head movements when induced caudally (Syka and Radil-Weiss, 1971; Roucoux et al., 1980; Northmore et al., 1988; Cowie and Robinson, 1994; Paré et al., 1994).The notion of a motor map in SC has also been substantiated by chronic unit recordings in awake monkeys performing orienting tasks. From these studies emerged the concept of a motor field; briefly, neurons in intermediate layers discharge before saccades, and they do so maximally for saccades of a particular direction and amplitude (Wurtz and Goldberg, 1972; Sparks et al., 1976). Thus, cells are topographically organized in intermediate layers, according to their movement field.

Finally, a major advance in the physiology of the deep SC has been provided by the observation that sensory and motor maps align, even if the way in which they do so is still unknown (Sparks, 1988). As initially shown in monkey by Schiller and Stryker (1972), local stimulation of the deep SC produces a saccade that moves the fovea to the region of the visual field represented in this area. Furthermore, convergent observations were drawn from recordings in behaving monkey, showing that cells with both sensory and motor responses, display overlapping (but not necessarily co-extensive) movement and receptive fields (Wurtz and Goldberg, 1972). In conclusion, the deep SC appears suitably equipped with a modular organization of the sensorimotor linkages to perform grasping movements of the cephalic captors and even body displacements toward attractive cues in the environment.

The concept of modularity in the deep SC, and in particular in the well-developed intermediate layers, has also been substantiated by anatomical data, which have provided decisive evidence for a discontinuous cyto- and chemo-architecture of these layers. Viktorov (1968) was the first to draw attention to the clustering of large multipolar neurons lying in the cat SGI, suggesting that such clusters may constitute functional subunits for orientation. This observation was later confirmed by retrograde labeling of different classes of efferent collicular neurons (Castiglioni et al., 1978; Huerta et al., 1981; Murray and Coulter, 1982; Yamasaki et al., 1986). In parallel, there is a patchy innervation of most, if not all, of the collicular afferents (from cortex, Goldman and Nauta, 1976; Künzle and Akert, 1977; Leichnetz et al., 1981; Segal et al., 1983; Clemo and Stein, 1983; from subcortical sensory relays, Kudo and Niimi, 1980; Tashiro et al., 1980; Huerta et al., 1981; Kudo, 1981; Huerta and Harting, 1982; Harting and Van Lieshout, 1991; from basal ganglia, Jayaraman et al., 1977; Graybiel, 1978a; Beckstead et al., 1979; Harting and Van Lieshout, 1991; Redgrave et al., 1992; from cerebellum, Kawamura et al., 1982; from hypothalamus, Rieck et al., 1986). Patches in the intermediate layers are regularly described in the frontal plane as a series of groups of either terminals or neurons with generally increased size from medial to lateral SC. In cat, for instance, the trigemino-collicular terminals segregate in a series of mediolateral patches with diameters ranging from 50 to 500 μm and spaced about 150 μm apart (Tashiro et al., 1980; Huerta et al,. 1981). In the same species, the nigrocollicular innervation distributes in 300–500-μm-wide clumps (Graybiel, 1978a). The frontal eyefield-related patches in monkey range from 300 to 500 μm, with equivalent space intervals (Künzle and Akert, 1977). Finally, in rat, neurons projecting to the thalamic intralaminar nuclei form clusters ranging from 80 to 220 μm, and the intervening gaps vary from 80 to 100 μm (Yamasaki et al., 1986).

The idea of collicular compartmentalization has been reinforced by the neurochemical finding of a regular distribution of a number of neuroactive substances in the intermediate layers such as somatostatin (Spangler and Morlay, 1987), enkephalin (Graybiel et al., 1984; Miguel-Hidalgo et al., 1989; Mize, 1989; Graybiel and Illing, 1994), cytochrome oxidase (Sandell, 1984; Wiener, 1986), choline acetyltransferase (Ross and Godfrey, 1985; Beninato and Spencer, 1986; Hall et al., 1989; Hashikawa, 1989; Illing, 1990; Henderson and Sheriff, 1991; Jeon and Mize, 1993), parvalbumin (Illing et al., 1990), substance P (Miguel-Hidalgo et al., 1989; Behan et al., 1995), hexokinase (Sandell, 1984), and others (Illing, 1996).

Among the impressive number of markers currently available to delineate patterning in the SC, acetylcholinesterase (AChE) represents the most frequently used. In the deep SC, the presence of AChE was shown to be associated, at least for a part, with the cholinergic projections arising in the pedunculopontine and dorsolateral tegmental nuclei (Woolf and Butcher, 1986; Hallanger and Wainer, 1988; Jones and Webster, 1988; Wallace and Fredens, 1988; Hall et al., 1989). The intracollicular distribution of this enzyme has been well documented in cat (Ramon-Moliner, 1972; Graybiel, 1978b; Illing and Graybiel, 1985), rat (Sandell, 1984; Ross and Godfrey, 1985; Beninato and Spencer, 1986; Toga and Santori, 1995), mouse (Wiener, 1986; Wallace, 1986), monkey (Graybiel, 1978b), and human (Graybiel, 1979). In spite of the powerful three-dimensional (3D) digital imaging techniques that have recently been introduced to visualize the stereometric arrangement of the AChE in rat (Toga and Santori, 1995), the descriptions were promising but remained incomplete. In fact, it is in cat that the most detailed description is available, starting with the pioneering studies by Graybiel (1978b) and Illing and Graybiel (1985). According to these authors, AChE labeling appears, when examined in the frontal plane, as a series of regularly spaced dense patches about 200–600 μm wide, at 300–600-μm intervals, extending from the lateral to medial aspects of the intermediate gray layer. This pattern is most clearly visible in the caudal third of the SC. Further forward it tends to become less regular. Rostral patches are irregular in shape and frequently fuse so that the regular spacing may eventually become fully disrupted. A vertical arrangement of the enzyme has also been identified. The AChE patches are frequently split into dorsal and ventral tiers, sometimes with the two parts linked by a wisp-like AChE extension. When reconstructed in the dorsal view from serial frontal sections, the cores of patches tend to line up so as to form longitudinal stripes that seem to radiate from a major anastomotic node, near the caudal pole of the SC, where AChE forms intermingled latticeworks.

Surprisingly, in the course of routine studies in rat, in which AChE activity was systematically reconstructed from tangential sections, we discovered some unexpected characteristics. AChE staining identified nearly perfect circular distributions in the rostral portion of the intermediate layers. This consistent observation was puzzling because it partially conflicted with the well-accepted description of AChE distribution in cat and monkey (Graybiel, 1978b). Following this observation, we carefully reexamined the stereometric pattern of the AChE activity in rat and then extended our comparison to other mammalian species (hamster, mouse, cat, and monkey). In the rat, we attempted to visualize the 3D architecture of the cholinesterase domains by reconstructing the enzyme labeling from serial sections cut in frontal, sagittal, and tangential planes. For the other species, the study was limited to a reconstruction of labeling from tangential sections.

Taken together, the present findings support a unified model of the AChE arrangement within the intermediate layers of the mammalian SC. Furthermore, this model comprehensively incorporates the classical description of the patchy and stripy features of the enzyme distribution.

MATERIALS AND METHODS

We studied the distribution of AChE activity in the superior colliculus of 30 adult male albino rats (Sprague-Dawley), 3 male adult mice (C57BL), 2 adult male golden hamsters (Syrian hamster), 2 adult male cats (Felis domestica), and 1 adult male monkey (Macaca fasciculata).

After an overdose (160 mg/kg, i.p.) of sodium pentobarbital (Nembutal, Sanofi), animals were perfused transcardially with a 0.9% saline wash followed by a fixative solution of 3% paraformaldehyde and 0.5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4) and finally cryoprotected with a solution of 10% sucrose in 0.1 M phosphate buffer (pH 7.4) at 4°C. The brains were kept overnight in 30% sucrose in phosphate buffer (pH 7.4) at 4°C.

The rat brains were sectioned at 40–60 μm on a freezing microtome, in either frontal, sagittal, or tangential planes, whereas those of the mice, hamster, cat, and monkey were cut only in the tangential plane, at 40–60 μm. For the cat and monkey, the left and right colliculi were separated on the inter-collicular midline and sectioned separately in a slightly rotated plane.

Floating sections were then processed for AChE histochemistry according to a slightly modified protocol of Geneser-Jensen and Blackstadt (1971) using as the developing medium a 10% potassium ferricyanide solution. The sections were mounted onto chrome-alum-coated slides, dehydrated through alcohols and xylene, coverslipped, and examined using brightfield microscopy.

The resulting pattern of staining was observed at ×100 magnification factor and drawn with a camera lucida. Serial-section reconstructions were assembled in each plane as follows. For the sake of clarity, we remind the reader that the mammalian SC is shaped approximately like a slightly flattened portion of a sphere in which the main vascular processes course in a radial direction toward the SC surface. We chose, after Illing and Graybiel (1985), to use them as fiducial landmarks for the stereometric reconstruction of the enzyme domain in each of the species studied. In the sagittal and frontal planes, we lined up the corresponding segments of the blood vessels, section after section. In the tangential plane this time, the cross sections of the blood vessels were vertically aligned through the depth of the deep SC. Consequently, in the tangential plane the procedure straightened the radial vascular trajectory and resulted in a flattened but very useful view of the SC.

An alternative procedure for the tangential reconstruction was attempted in two rats. It consisted of the use of artificial marks to assemble the serial collicular sections. After being fixed with the aldehyde solution, the rats were repositioned in the stereotaxic frame. Two holes were then stereotaxically drilled by the vertical motion of a metallic electrode through the pretectum and the inferior colliculus (atlas of Paxinos and Watson, 1982). The two holes were kept well visible on each section of tissue and thus could be reliably used as reference marks to align the sections. This procedure was not, however, retained for further reconstructions because owing to the collicular curvature, it resulted in complex AChE pictures difficult to analyze without the help of 3D imaging computer software.

Shrinkage due to dehydration was evaluated only from rat tangential sections, as follows. In the same animal we compared pictures produced by AChE staining in both wet and dry tissue. Immediately after the staining procedure, the wet sections were rapidly photographed (using a standard Leica camera) at ×100 magnification. A photomontage of each individual section was prepared from a set of silver prints that we used for drawings. After dehydration and mounting, the same sections were drawn with a camera lucida and the distribution and measurements of the AChE deposits compared with the first set of wet images.

For processing, the figures were prepared with Adobe PHOTOSHOP from scanned (1200 dpi) drawings and silver-printed photomicrographs. The contrast of certain photomicrographs was sometimes altered to obtain a uniform tone for the photomontages presented.

All experimental protocols were approved by the French Office of Agriculture, Fisheries, and Food (authorization no. 6805) and complied with the veterinary regulations on the use of laboratory animals.

RESULTS

AChE patterning in intermediate layers of the rat SC

Frontal plane.

The enzyme labeling appears in this plane in two different staining densities that we have defined as the “dense” and “diffuse” AChE domains respectively (Fig. 1).

The dense AChE domain.

This appears predominantly as a heavily stained fibrous system that constitutes a well-defined regular network throughout the intermediate layers. In the caudal pole of the rat SC, where the intermediate layers are limited to the SGI, the AChE labeling forms two semi-parallel AChE tiers of about 40 μm thickness each, which delineate the dorsal and ventral limits of the layer, respectively. They are linked by regular vertical AChE fibrous fascicles or strands (30–50 μm thick), giving the labeling a ladder-like appearance and dividing the SGI into a mediolateral row of three to four square compartments (with 150–250 μm sides). On a given section, their contours are generally seen as a fragmented AChE line continuous with the adjoining sections. In some of the frontal sections, the ventral AChE tier appears broken up into regularly spaced small stained dots, taking on the appearance of cut fiber tracts from which the vertical AChE strands often arise. In the rostral two-thirds of the SC, the AChE staining contrast decreases, but the geometrical arrangement is maintained for the SGI and is even extended into the underlying SAI (Fig. 1). Three AChE tiers are visible here and roughly delineate the SGI and SAI boundaries. The ventral tier spreads at the interface between the SGP and SAI layers whereas the intermediate and the dorsal ones mark the SAI/SGI and the SGI/SO boundaries, respectively.

Compared with the ventral tier, which appears as predominantly fibrous, the dorsal and intermediate tiers show a rather fleecy texture suggestive of the presence of terminal fields (Fig. 1E). As for the vertical strands in which fibers and more moderate terminal fields intermingle, they course through the intermediate layers in a radial direction with respect to the curved SC surface; they follow the trajectory of blood vessels, with which they are often parallel. On some sections, they seem to be at the origin of the intermediate and dorsal tiers through lateral arborizations. Thus, the horizontal and vertical dense AChE strands together form in the rostral two-thirds of the SC a geometrical frame dividing both the intermediate strata into two sets of aligned compartments. However, this double-decker organization clearly blurs in the rostralmost pole, where the intermediate AChE tier becomes difficult to distinguish. The same observation can be made for the AChE organization near the midline, where the SGI and SAI greatly reduce in thickness and fuse. Finally, the enzymatic grid pattern is absent in the anterolateral portion of the intermediate layers, designated as the “flank” by Wiener (1986), i.e., the area that protrudes beyond the overlying superficial layers. This region is only outlined by a thin enzymatic sheet that seems to be a lateral continuation of the dorsal tier (Fig. 1A).

The diffuse AChE domain.

This is well developed only in the rostral SC, where the intermediate strata are in two layers. We systematically distinguish a faint and dispersed AChE labeling filling the spaces delineated by the dense AChE frame, as well as the flank (Fig. 1A). As a rule this labeling shifts from one section to another, making its precise arrangement difficult to identify in the frontal plane. It is sometimes seen right in the middle of the densely stained compartments, clinging near the edges or even absent. Under higher magnification, we noted that the diffuse-AChE domain coincides with rich fibrous areas. The ventral SGI and the SAI are, in rat, regions where richly fibrous areas run along the longitudinal axis of the SC. Thus, the diffuse-AChE domain spreads into the interstices between non-AChE fibers, which reduces its visibility still further in the frontal plane (Fig. 1D). Finally, in three rats, some sparse and weakly AChE-stained medium to large-sized perikarya were encountered in the lateral SAI. For the sake of clarity, we will maintain in the following sections the nomenclature for the AChE domains set out here from observations in the frontal plane. So we will refer to the compartment frame as the dense AChE domain and to its scattered contents as the diffuse-AChE domain.

Sagittal plane.

In this plane, the two distinct AChE domains are still obvious but now the visibility of the diffuse domain has improved (Fig. 2).

The dense-AChE domain.

A pattern similar to that described for the frontal plane is observed for the caudal third of the SC. AChE labeling is seen in the SGI as a heavily stained fibrous network that composes a one-layered series of four to five square compartments with sides of 150–250 μm. Rostralward, the dense frame, while present, becomes harder to detect, probably due to its out-of-plane orientation and weakening of its staining. In this plane, neither of the dorsal and intermediate AChE tiers are easily distinguishable because they are often confused with the longitudinal strips of the diffuse domain (see below). Only the ventral tier remains discernible as a fairly regular series of AChE-positive fiber tracts. They form a longitudinal line, accurately delineating the boundary between SAI and the underlying SGP from which, in some cases, the vertical strands were seen to arise (Fig. 2A,B). Although a partial reconstruction of the fiber network has been undertaken from sagittal sections, the best view is from the tangential plane. In fact, in the sagittal plane, it is the well-structured diffuse-AChE domain that is the most striking.

Diffuse-AChE domain.

As in the frontal plane, the diffuse domain is seen primarily developed in the rostral two-thirds of the intermediate layers. It displays more often a series of regular blocks that vertically segment both the intermediate strata. Every block is traversed longitudinally by non-AChE fibers, giving it a striated appearance (Fig. 2C,D). In the more medial SC, the AChE staining pattern is a one-layered series of rectangular blocks (100–150 μm thick and 50–300 μm long) separated by free enzyme zones. When followed from section to section, each AChE block regularly splits in two lateral fragments that move away and further fuse to again form a single block. Thus, the interblock interval shifts center to center from 0 to 200–250 μm. At the mid-level of the SC, the AChE blocks organize in two aligned sets lying in the SGI and the SAI layers. In the lateral SC, the upper set of blocks remains visible only in the SGI, and the lower set in the SAI tends to lose its segmental arrangement and becomes an evenly stained area. This even texture will further rule the enzyme staining in the flank area where the SAI layer predominates.

Tangential plane.

This plane was crucial for the integration of the frontal and sagittally derived observations into a logical framework. Indeed, the tangential plane gave the best overall picture of the complementarity of both dense and diffuse AChE domains (Figs. 3, 4A). In rat, taking into account the curvature of the SC and the thickness of the intermediate layers, 10–12 serial sections (50-μm-thick) are generally required from dehydrated tissue, to reconstruct the picture of the horizontal enzymatic arrangement. Both enzymatic domains combine to form a conspicuous honeycomb-like network of about 100 ovoid and rounded compartments.

It is in the caudal SC that the lattice shows its heaviest staining contrast. Here the enzyme latticework is composed of dense cross-linked bands (40-μm-thick). They delineate 15 or so ovoid compartments (150–200 μm axis) that roughly compose four transverse rows. In the rostral two-thirds of the SC, in agreement with the observations made in the frontal and the sagittal planes, the AChE staining fades, but the compartmentation is maintained. Both the dense and diffuse domains join to form a fairly regular matrix of about 80 rounded compartments with thick contours (50–100 μm) surrounding enzyme-free hollow centers (150–300 μm diameter, the largest being near the medial edge of the SC).The compartments are distributed with a spacing of 200–350 μm from center to center. In this plane, the dense and diffuse domains align perfectly in the depth of the intermediate layers so that the transition from one to the other is not obvious. The vertical strands of the dense domain are nevertheless well identified since they compose an array of heavily stained AChE profiles (40–60 μm in size) regularly spaced by 150–300 μm in any direction (Fig. 4A,B). Their number could not be fully determined, because they often surround blood vessels like a sheath, making their identification sometimes ambiguous. Similar to the other planes, these vertical strands clearly originate from the ventral AChE tier.

In the tangential plane, this tier appears as a fibrous plexus extending from caudal to rostral SC composed of a series of 12–13 regularly spaced parallel fascicles (200–250-μm interval). They enter laterally into the intermediate layers at the SGP/SAI interface and course in a transverse direction toward the collicular midline (Fig. 4C). During its course each fascicle is seen to be periodically (150 to 250–300 μm) at the origin of a lateromedial series of several vertical fibrous strands. We sometimes clearly observed that the 12–13 AChE fascicles also give off a net of divergent secondary branches, spreading this time in a horizontal direction at the SGP surface. They radiate from the bifurcating nodes where the vertical strands arise, and they cross-link the series of nodes of a given fascicle with that of the next (see Fig. 6A). So the ventral AChE tier appears to form latticework whose arrangement reflects the overlying lattice.

Finally, the honeycomb-like net spreads over the full extent of the intermediate layers with the exception of the flank area, where the AChE staining is evenly distributed.

An estimate of the amount of shrinkage due to dehydration has been carried out from tangential sections. It is approximatively 15% following measurements of the internal and external diameters of the honeycomb compartments in the wet and dry tissue, respectively.

DISCUSSION

The present study was aimed at reinvestigating the stereometric pattern of AChE-related labeling in the intermediate layers of the SC in rat and several other mammals. The findings have led us to conclude that this enzymatic marker demonstrates the same pattern in all species examined. It consists of a regular honeycomb-like lattice of AChE-positive elements throughout virtually the whole extent of the intermediate layers. Such an arrangement was first suspected in cat and monkey by Graybiel (1978b) and then reinvestigated in the cat by Illing and Graybiel (1985); unfortunately, the entire pattern was not systematically revealed. In cat, for instance, these authors stressed as a basic arrangement for AChE a dense anastomotic grid-like node in the caudal SC that rapidly disorganizes rostralward to become a vanishing longitudinal band-like network. We propose that the reason for such a discrepancy between our observations is that these authors lacked pieces of the AChE jigsaw puzzle. The complexity of the 3D enzymatic network is such that it requires a comparison of views in frontal, sagittal, and tangential planes and a complete reconstruction in each of them. Indeed, the three planes are complementary since separately they show only a part of the lattice and cannot reveal the others. Any reconstruction of profile from a single plane can be misleading. In these classic papers, the tangential view of AChE labeling was inferred from frontal sections, and the reconstruction probably failed to take into account what we call in the present report, the diffuse AChE domain. This domain is found predominantly in the anterior two-thirds of the SC, and its visibility is weak in the frontal plane. Although in both studies the same AChE protocol has been followed, it cannot be excluded that minor “tweakings” of protocol that unavoidably take place in different ways, in different labs, might account for somewhat different results.

What does an AChE compartment mean?

It has long been known that the deep SC is a critical component of the neural circuitry that processes orienting movements of eyes, head, pinnae, vibrissae, and even the whole body toward novel and behaviorally significant cues (Huerta and Harting, 1984; Grantyn, 1988; Sparks and Hartwich-Young, 1989; Stein and Meredith, 1991). Besides orientation, defensive movements such as escape or avoidance have also been described in rodents (Northmore et al., 1988; Dean et al., 1989). The premotor cells involved in the generation of these orienting movements are organized topographically in the intermediate layers and form a map of motor space. Converging onto this neuronal matrix, somatic, visual, and auditory signals are also aligned topographically and thereby form maps of sensory space. Finally, it is generally accepted that the motor and sensory maps are in functional register so that in SC, sensory space would be transformed in motor space and hence in “grasp orienting responses” (Sparks, 1988). Since physiologically the intermediate layers appear to be a high-resolution matrix for the generation of directed orienting movements, there is a possibility that the AChE lattice reflects this organization. In keeping with that hypothesis, one may put forward the proposal that “AChE compartments” function as orienting modules. Each compartment would contain the pool of neurons reactive to various sensory inputs from a location in sensory space and that through efferent connections with brainstem motor circuitry would perform the spatiotemporal transformation required to trigger orienting movements of the appropriate direction and amplitude. With reference to the orienting vector that could be logically expected in each compartment or group of compartments, careful studies combining single unit recordings of sensory and motor fields and AChE staining are clearly required.

However, to guide further experiments, a proposal is presented Figure 7. By drawing one's inspiration from the motor map for eye movement set up by Roucoux and Crommelinck (1976), we drew in the movement vectors we hypothesize to be present throughout the cat AChE lattice. So, compartments in the mid-rostral SC, where the central vision is represented, would be responsible for forward gaze shifts of small amplitude, whereas those in the caudal SC would trigger large eye movements combined with head deviation for the retina's eccentricity beyond 30°. The latter region would correspond to the most heavily stained rows of compartments. Finally, saccades governed by compartments lying parallel to the midline would have a clear upward component, whereas those evoked by the lateral compartments would have a downward component.

All these assertions rely on further evidence that AChE is a relevant marker for disclosing the very structure of the collicular matrix for orientation. In following up this idea, further research should investigate a series of modular markers, such as calretinin, parvalbumin, and cytochrome oxidase, for instance, which constitute with AChE a mosaic of complementary instead of overlapping domains (Wallace, 1986; Wiener, 1986; Illing, 1996). Likewise, some somatic afferents in the intermediate layers such as those from the bulbospinal junction, the somatosensory area SIV, and also visual afferents from the lateral suprasylvian cortex were shown in the cat to form complementary pictures with the AChE-rich domain (Illing and Graybiel, 1986). An important question is how these afferents and chemoarchitectural domains are distributed with respect to the AChE lattice. With regard to this, we report data in a companion paper in favor of a single modular structure that links AChE with the alignment of some of the major input and output pathways of the intermediate SC.