Cerebral Cortex, Vol. 10, No. 4, 371-399,
April 2000
© 2000 Oxford University Press
Cortical Connections of the Insular and Adjacent Parieto-temporal Fields in the Cat
Department of Morphology, School of Medicine, Autónoma University, Madrid, 28029 Spain
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionConcluding RemarksNotesReferences |
|---|
We present a comprehensive analysis of the cortical connections of the insular and adjacent cortical areas in the domestic cat by using microinjections of wheat-germ agglutinin conjugated to horseradish peroxidase. We examined the identity and extent of the cortical fields connected to each area, the relative anatomical weights of the various connections, their laminar origin, and their paths across the cerebral commissures. Our main finding is that despite their relatively small size and close apposition, the connections of the insular and adjacent areas are far more widespread and more specific to each area than previously realized, suggesting that each area is involved in disparate aspects of cortical integration. The granular insular area is linked to a constellation of somatosensory, motor, premotor and prefrontal districts. The dysgranular insular area is chiefly associated with lateral prefrontal and premotor, lateral somatosensory and perirhinal cortices. The dorsal agranular insular area is connected with limbic neocortical fields, while the ventral agranular insular area is associated with an array of olfactory allocortical fields. The anterior sylvian area is associated with visual, auditory and multimodal areas, with the dorsolateral prefrontal cortex, and with perirhinal area 36. The parainsular area is linked to non-tonotopic auditory and ventromedial frontal areas. Trajectories followed by the callosal axons of each of the investigated areas are extremely divergent. As a whole, the picture of the insular region that emerges from this and a parallel study (Clascá et al., J Comp Neurol 384:456–482, 1997) is that of an extreme heterogeneity, both in terms of histological architecture and neural connections. Comparison with earlier published reports on primates suggests that most, but not all, of the areas we investigated in cats may have an direct counterpart within the insula of Old World monkeys.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConcluding RemarksNotesReferences |
|---|
Extensive arrays of long cortico-cortical connections are a key anatomical feature of higher association cortices in primates and carnivores. These long pathways, which include intrahemispheric as well as comissural connections, are regarded as a pivotal substrate for the complex integration capabilities of the association cortices (Reinoso-Suárez, 1984
). Precise anatomical mapping of cortical pathways, therefore, is crucial for understanding the functional properties of the these cortices. However, data on the cortical connections of many association cortices remain scant and fragmentary even in model species as well-studied as the domestic cat.
A loosely defined ‘orbito-insular’ region situated in the anterior sylvian and orbital gyri and adjacent sulci of cats has long been regarded a major association cortex in this species (Imbert et al., 1966; Avanzini et al., 1969; Loe and Benevento, 1969; Benevento and Loe, 1975; Fallon and Benevento, 1977, 1978; Reinoso-Suárez, 1984; Guldin and Markowitsch, 1984; Guldin et al., 1986; Hicks et al., 1989b). Although some studies have examined the cortical connections in limited parts of the region (Avanzini et al., 1969; Cranford et al., 1976; Guldin and Markowitsch, 1984; Guldin et al., 1986; Yasui et al., 1987, Norita et al., 1991), large portions of the orbito-insular region remain unexplored with modern tracing methods. Moreover, these previous studies have not, in general, mapped their findings with cytoarchitectonic or stereotaxic references, making it difficult to compare their data. In addition, since their lesions or tracer injections often extend to the claustrum, and this nucleus itself has widespread cortical connections (Clascá et al., 1992), the significance of the reported findings is unclear. Although additional data is available in different studies focused on other cortical zones (Reale and Imig, 1980; Craig et al., 1982; Burton and Kopf, 1984; Reinoso-Suárez, 1984; Cavada and Reinoso-Suárez, 1985; Reinoso-Suárez and Roda, 1985; Room et al., 1985; Witter and Groenewegen, 1986; Room and Groenewegen, 1986; Bowman and Olson, 1988; Avendaño et al., 1988; Clarey and Irvine, 1990; Bowman and Olson 1988; Ghosh, 1997a,c), it is not possible to infer from the published data data either the precise extent of the cortical territories connected to the insular region, or the relative anatomical weight of the various connections. Overall, data on the laminar origin of cortical input to the insular region, or on the comissural connections of this region, are almost nonexistent.
We set out to analyze systematically the cortical connections of the orbito-insular region with the following specific goals: (i) to elucidate the areas connected to the various areas of this region; (ii) to determine the relative anatomical weight of the various connections, their laminar origin and their paths across cerebral comissures; (iii) to gain insight into the functional realm of each field through a comparison of its cortical connections to data from previously published physiological studies; and (iv) to explore the possibility that some of these pathways resemble cortical connections described in the insular areas of Old World primates. For this study, we took advantage of data from a parallel study of thalamic connections in the cat's orbito-insular region (Clascá et al., 1997). Results show that the cortical connections of the various areas in the orbito-insular region are far more widespread and more specific to each area than previously realized, and suggest that each area may be involved in disparate aspects of cortical integration. Preliminary results have been reported previously in abstract form (Clascá et al., 1996)
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConcluding RemarksNotesReferences |
|---|
The brains of 31 adult domestic cats of either sex were used in the present study. All procedures involving live animals were carried out in accordance with European Communities Council Directive 86/609/EEC guidelines.
Surgery
Animals were anesthetized with sodium pentobarbital (30 mg/kg), and additional doses (10 mg/kg) were administered as required throughout the surgical procedure to keep the animal arreflectic while preserving spontaneous ventilation. We exposed the target zone through a small craniotomy. To address the possibility that some cortical connections are not homogeneously distributed within the areas investigated, we performed either small injections involving limited parts of an area, or larger ones, covering most or all of the area. Under direct visual guidance, we made unilateral microinjections in the cortex of a mixture of 30% horseradish peroxidase (HRP) + 2% wheat germ agglutinin conjugated to HRP (WGA–HRP). In two animals (nos 201 and 363), we injected a 50% solution of HRP in distilled water. In most cases, we made a single injection of 40–60 nl. In four experiments, we made two or three contiguous deposits (60 nl each) to impregnate a more extensive zone. We adjusted the depth and angle of injection to impregnate all cortical layers as evenly as possible. In most cases, we used a 1 µl Hamilton syringe with a beveled and gauged tip; however, in experiments aimed at the deep sulcal cortex, we air-pressure injected the tracer through a glass micropipette (15–25 µm external diameter at the tip) using a Picospritzer II (General Valve, Fairfield, NJ). After injection was completed, we covered the exposed cortex with a film of hemostatic gelfoam, sealed the bone with dental cement, and sutured muscle and skin. Amoxycillin (3 mg/kg/day) was administered preoperatively and throughout the postoperative period.
Histology
Between 46 and 54 h after the injection, the animals were overdosed with sodium pentobarbital (80 mg/kg), and transcardially perfused with saline (5 min), 1% paraformaldehyde + 1.25% glutaraldehyde in phosphate buffer (pH 7.4, 4°C, 45 min), and 10% sucrose in the same buffer for 20 min. We then split the brains along a coronal plane, and subsequently cryoprotected the tissue by soaking in phosphate-buffered 30% sucrose for 48 h at 4°C. Using a freezing microtome, we cut the whole brain into 50 µm thick serial coronal sections, collecting six parallel series of sections. Two series of sections were used for histochemically revealing HRP using tetramethylbenzidine (TMB) (Mesulam, 1978). These sections were then mounted, air dried, lightly counterstained with thionin and coverslipped. Other series of sections underwent either staining with cresyl violet, acetylcholinesterase histochemistry (Geneser-Gensen and Blackstadt, 1971) or cytochrome oxidase histochemistry (Wong-Riley, 1979). The remaining series were discarded.
Microscope Examination of the Sections
For each brain, we analyzed and drew an entire series of TMB-reacted sections throughout the rostrocaudal extent of both cerebral hemispheres (one section every 250 µm). Using either a camera lucida mounted on a stereomicroscope, or an inverted projector, we traced section contours, heavier labeling and tissue landmarks (vessels, the inner borders of cortical layers I and VI, and the outer limit of layer V) as revealed by the thionin counterstain at 6x. Subcortical fibers, anterogradely labeled axon terminals and faintly labeled cell somata were subsequently recorded, re-examining the sections under brightfield and/or darkfield optics and polarized light at 50–300x in a Zeiss microscope. This was done by hand on the camera lucida drawings, using the previously drawn labeling and tissue landmarks as references for accurately positioning the labeling.
Delineation of Cortical Areas
Correct identification of the areas labeled by the axonal transport required comprehensive delineation of the cat's cerebral areas adjusted to stereotaxic coronal planes. To this end, we revised and updated the Reinoso-Suárez map (Reinoso-Suárez, 1984) with data collated from a large number of studies from this and other laboratories. For clarity, the studies relevant for the delineation of each area are referenced with the abbreviations given in Table 1. Figure 1 summarizes the resulting cortical map over ‘flat’ medial and lateral views of the cat's cerebral hemisphere. In this type of diagram, the anteroposterior extent of the areas is accuratedly matched to stereotaxic coronal planes, but the relative extent of sulcal cortex becomes substantially underrepresented. The map is not based on the analysis of any single brain but, rather, it represents an idealized ‘average’ shape and extent of cortical areas. As a ‘working diagram’, this map may need revision as new experimental data become available.
|
|
Reconstruction of the Injection Sites
We first analyzed and reconstructed the injection sites. We considered valid for subsequent analysis only the injection experiments which had impregnated all cortical layers in a roughly proportionate manner, and had no significant spread of tracer to the subcortical white matter or the claustrum. A total of 13 injection experiments that did not meet these criteria were discarded. The results reported here, therefore, are based on the analysis of a total of 18 valid cases (Figure 2).
|
In TMB-stained material, the core of the tracer deposits contains a dense black precipitate, which glows purple under dark-field optics. A translucent halo of precipitate, which has a golden glow in dark-field, extends for few hundred microns around the core. As a conservative estimate, we considered ‘injection site’ to be both core and halo. For comparison of the position of the injections in the various brains, we reconstructed the tracer deposits, section by section, on a single ‘unfolded’ cortical map of the orbitosylvian region [Figure 2
; for details on the procedure to generate this map see Clascá et al. (Clascá et al., 1997)]. The coronal level of the sections containing the injection site was determined with the aid of Reinoso-Suárez's brain atlas (Reinoso-Suárez 1961), while the dorsoventral spread of the deposit was estimated according to cytoarchitectonic and gyral landmarks. Analysis of the Cortical Labeling
In our material, the cytoarchitecture revealed by the thionin counterstain of the TMB-reacted sections made it possible to delineate most of the cortical fields. When thionin staining was inconclusive, adjacent cresyl violet or acetylcholinesterase stained sections were used to elucidate the border of a cortical field. However, a number of fields, such as the extrastriate visual areas in the suprasylvian sulcus (Tusa et al., 1978; Tusa and Palmer, 1980; Updyke, 1986; Grant and Shipp, 1991), or the auditory fields in the anterior ectosylvian and posterior ectosylvian gyri (Reale and Imig 1980; Clarey and Irvine, 1990; Winer, 1992), are largely defined by physiological mappings, and in these regions we relied on gyral patterns and stereotaxic references reported in the original studies.
The spread of labeling across the cerebral hemispheres and the large number of sections tended to obscure the relative weight of the various projections. We therefore decided to complement the examination of single sections with a numerical analysis of labeled cells, area by area, across an entire series of sections. It must be emphasized that these counts were never intended as a quantitative estimate of the actual total population of labeled cells, but rather as an aid in perceiving the overall amount of the various sets of labeled neurons. Cell counts were made by hand on the section drawings. The resulting numbers fluctuated widely between the various experimental cases (range 577–5969 cells; mean ± SD 2344 ± 1341). To normalize for comparison between experiments, cell numbers were converted to percentages against the sum of all the labeled cortical cells counted in the same experiment.
To visualize the spatial distribution of the labeled connections across the cortical mantle, we generated reconstructions of the labeling onto lateral and medial views of the individual cerebral hemispheres. In these reconstructions, the individual labeled cells in each serial section were represented as dots along parallel lines matched to the anteroposterior level of the section.
In addition to labeling in the cortical gray matter, examination of TMB-stained material under dark-field and polarized light revealed the entire course of the axons through the white matter, including the interhemispheric comissures. We recorded these fibers on the drawings of the serial coronal sections. To facilitate comparison between cases, we reconstructed, section by section, the position of the labeled interhemispheric axons on a standard midsagittal section of the cat's cerebral commissures.
To determine the cortical layers of origin for the labeled projections more precisely than by an inspection of single sections, we decided to count, for each area in both cerebral hemispheres, the neuronal somata labeled either in the superficial (IV–II) or deep (V–VI) cortical layers. We then calculated the ratio between both groups of layers, and subsequently compared the ratios of the various areas.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConcluding RemarksNotesReferences |
|---|
Microinjections of WGA–HRP in the cortex of the orbito-insular region produced widespread anterograde and retrograde labeling in both cortical hemispheres. A detailed account of the percentage distribution of retrograde labeling in every case is provided in graphic format in Figure 3
, which also gives a global overview of our findings. When this information is compared with the locations of injections (Fig. 2, and black squares in Fig. 3), it can be seen that the patterns of labeling in experiments injected in the same area are remarkably consistent. Additionally, this analysis reveals a fine-grained variability in the relative weight of the various connections that, to some extent, can be correlated with small differences in the location of the injections.
|
The following account reports the observations for some ‘representative’ cases, usually one for each area investigated. Only features departing from the patterns seen in the representative case are described for the remaining cases. The spatial distribution of the labeling is documented by the drawings and photomicrographs in Figures 4–15
. Note that although axonal tracing with WGA–HRP produced both anterograde and retrograde labeling, for clarity only retrogradely-labeled somata are represented in the line drawings.
|
|
and 6aß. (E) Labeling in the ventromedial prefrontal cortex produced by an injection in the agranular insular areas (no. 788). The most ventral aspect of both left and right frontal cortices is shown. An arrow indicates the midline. Note the absence of labeling in the right frontal cortex.
|
|
|
|
|
|
|
|
|
|
Injections in the Granular Insular Area (GI)
Area GI encompasses the posterodorsal orbital gyrus and ventral lip of the orbital sulcus between anteroposterior planes (AP) +20 and +17 (Clascá et al., 1997). Four experimental cases (nos 983, 935, 818 and 847) have injections restricted to, or primarily located in, area GI (Fig. 2).
The representative GI case is no. 983, illustrated in Figures 4 and 5A–D. Labeled cells are spread over a broad zone of the injected hemisphere. Heavy labeling is located in (i) the dorsal bank of the anterior ectosylvian sulcus (fourth somatosensory area, S-IV), as well as nearby zones of the anterior ectosylvian gyrus (second somatosensory area, S-II); (ii) sectors of the lateral bank and bottom of the presylvian sulcus that include area 6a
(Avendaño et al., 1992), and the dorsal border of the dorsolateral prefrontal sector (DlP) (Cavada and Reinoso-Suárez, 1985); (iii) areas 3a, 3b and 4 in the dorsal lip of the coronal sulcus and adjacent portions of the sygmoid gyri; (iv) a ventral zone of area 2 in the dorsal bank of the orbital sulcus; and (v) the fifth somatosensory area (S-V) (Mori et al., 1991) in the dorsal bank of the suprasylvian sulcus. There is additional labeling in both lips of the cruciate sulcus (areas 6a
and 6aß, and medial sectors of areas 3a and 4), in the dysgranular insular area, as well as in caudal and ventral portions of the anterior ectosylvian sulcal cortex1 (fields PAE and VAE). In the medial aspect of the hemisphere, numerous cells are labeled in area 7m (Avendaño and Verdú, 1992) and adjacent zones of the posterior cingulate area (CgP). A further collection of cells is labeled in the medial bank of the posterior rhinal sulcus (area 35 and the dorsolateral entorhinal area, DlE). In the contralateral hemisphere, areas GI, DI, S-IV, S-II and 3a contain the largest numbers of labeled neurons; however, in contrast to the injected hemisphere, no cells are labeled in 7m and S-V (Figs 4 and 5B).
In most areas, the distribution of anterogradely labeled fibers closely matches that of the labeled somata (Fig. 5A,B), although there appear to be some differences in the overall amount of anterograde labeling in the various areas. The heaviest anterograde labeling is present in areas S-IV, 3a and 6a, where it is arranged in a columnar fashion that largely matches the distribution of the retrogradely labeled somata. On the other hand, areas 3b, S-II and the motor fields of the cruciate sulcus show faint anterograde labeling (Fig. 5C,D). In most areas, layers I, III and VI contain the densest aggregates of anterogradely labeled fibers and terminals.
The remaining three GI injection experiments (Fig. 6) largely concur with the findings in case no. 983. On the other hand, each case shows some particular features that, at least in part, may reflect the specific location of the tracer deposits within GI. For example, the injection in no. 935 partially overlaps that in no. 983, but it also spreads to a more rostral and dorsal portion of GI (Fig. 2). Compared to case no. 983, labeling in no. 935 is almost absent in areas 5, 7m and Cg; even scarcer in 3b; but fairly heavier in S-II and PAE (Figs 3 and 6). Likewise, in case no. 818, which involved a caudodorsal portion of GI as well as a small border zone of the ventral anterior ectosylvian field (VAE), labeling of the somatosensory cortex is less extensive, and mainly restricted to S-IV, 3a, S-II and S-V. Moreover, areas weakly labeled in case no. 983, such as area 36 and the anterolateral lateral suprasylvian area (AlLS), contain significant labeling in no. 818. Case no. 847 received a large tracer deposit that encompasses most of GI and two small bordering zones of areas DI and AId. Despite the fact that the zone impregnated nearly doubles in extent that of no. 983 (Fig. 2), this injection basically yielded the same labeling pattern, with some additional cells and fibers labeled in area 36 of the perirhinal cortex, as well as in the infralimbic (IL), prelimbic (PL) and anterior cingulate (CgA) areas.
Injections in the Dysgranular Insular Area (DI)
Area DI extends over the anteroventral aspect of the orbital gyrus and the lateral lip of the presylvian sulcus between AP +22 and +18 (Clascá et al., 1997). Two valid experimental cases (nos 851 and 907) have injections centered in DI (Fig. 2).
Case no. 851 is described as the representative, and illustrated in Figures 7 and 8A,C. The tracer deposit in this case covers a large extent of DI, along with a small border zone of AId (Fig. 2). Despite the relatively large size of the injection, however, labeling spreads over a smaller zone than after similar, or even smaller, injections in adjacent area GI. Labeling is heavy in 6a, DlP and 2 in the dorsal lip of the orbital sulcus, as well as in area AS. However, unlike in any of the GI-injected cases, the remaining somatosensory and motor fields are not labeled. Moreover, also unlike injections limited to GI (case nos 983, 935 and 818), there are labeled neurons in IL, and in a rostral zone of the posterior rhinal sulcus that is transitional between 35 and DlE. In addition, labeling in the gustatory area (G) is heavier than following injections in GI. In the opposite hemisphere, the densest labeling involves DI and GI. The tangential distribution of anterograde labeling basically matches that of the labeled somata. The densest anterograde labeling is present in the presylvian sulcus and area 35, where it mainly involves layers VI and I (Fig. 8A,C).
Being limited to a rostral and dorsal portion of DI and a border zone of GI (Fig. 2), the tracer deposit in no. 907 is substantially smaller than that in no. 851. As would be expected from the smaller size of the deposit, fewer neurons are labeled in the cortex; nevertheless, their distribution is a virtual replica of case no. 851 except for the absence of labeling in the rostral perirhinal cortex and S-IV, and a few labeled cells in 3a and S-V (Figs 3 and 6I).
The minute tracer deposit in case no. 363 is placed in a region that, according to our map, corresponds to a ‘junction’ zone between areas DI, GI, AId and AS (Fig. 2). Accordingly, labeling in this case (Fig. 3) involves some of the areas labeled by injections in DI or GI (areas 4, 6, 5 and S-IV); others labeled by injections in AId (PL and AL); as well as some further areas typically labeled by injections in AS (posterior suprasylvian area, PS; temporal auditory field, Te — see below).
Injections in the Agranular Insular Areas (Areas AId and AIv)
The agranular insular cortex comprises the dorsal bank and fundus of the anterior rhinal sulcus between AP +19 and +13 (Fig. 2). An isocortical agranular dorsal field (AId) extends along the dorsal bank. The ventral agranular subfield (AIv) is cytoarchitectonically transitional with the olfactory allocortex, and makes up the bottom and deep part of the ventral bank of the sulcus. Since AId and AIv are ‘folded’ within the anterior rhinal sulcus, it is not easy to reach them as selectively as would be desirable, and none of our three valid injections in the anterior rhinal sulcus involved a single area independently. Moreover, the claustrum is wrapped around AId and AIv, and narrowly separated from them by a thin extreme capsule. Thus, it was technically difficult to avoid some tracer spill over the claustrum, and it was decided to include two injections with some tracer spread to the claustrum (nos 788 and 711) among the valid cases. Overall, even if none of the three valid injections in AId and AIv is per se an ideal experiment, one can draw a consistent picture of the cortical connections from the comparison of the labeling patterns in the three experiments.
The injection in no. 788 involves a sector of AId and an adjacent zone of AIv at about AP +16.5. The tracer spilled over a small ventral portion of the dorsal claustrum (Fig. 2). Cortical labeling (Figs 9 and 5E) is confined to ventral isocortical areas and allocortical olfactory fields. The largest set of HRP-positive cells is situated in the ventral and medial frontal region [ventral prefrontal sector (VPf), IL and PL]. Area DI and portions of AS adjacent to the injection also contain numerous labeled cells. The perirhinal cortex, particularly area 35, is labeled across an extensive anteroposterior range. In the allocortex, the densest labeling involves the prepyriform cortex (PpC), while some few neurons are labeled in taenia tecta (TT). In the contralateral hemisphere, AId, VPf and 35 show the heaviest retrograde labeling. Interestingly, there are virtually no HRP-positive cells in contralateral AIv, and allocortical fields are not labeled.
Anterograde labeling is heavy in all the sectors of the prefrontal cortex, but particularly in VPf (Fig. 5E). Additional anterograde labeling is present in the medial part of 35, as well as in DlE, where it is situated deep to the lamina dissecans.
The tracer deposit in case no. 811 involves AId, AIv and AS between AP +14.5 and +16 (Fig. 2), and completely spares the claustrum. Retrograde labeling in this case (Fig. 3) shows features similar to no. 788, plus others typical of AS injections, such as labeling of auditory fields A-2 and Te, visual fields of the suprasylvian sulcus and CgP (see below). Anterogradely labeled fibers in DI and VPf show a density and laminar distribution similar to case no. 788; however, unlike no. 788, anterograde labeling in the prefrontal cortex is mainly restricted to VPf.
The tracer deposit in case no. 711 involves AIv and the adjacent PpC. Labeling is restricted to allocortical and ventral frontal isocortical regions (Fig. 9). While labeling in the isocortical and transitional areas is circumscribed to VPf, IL, and the agranular orbital area2 (AO), labeling in allocortical fields such as TT and Pp is heavier and more widespread than the previous two cases.
Injections in the Parainsular Area (Pi)
Area Pi covers the ventral bank and bottom of the pseudosylvian sulcus, except for its caudal end. Our series includes two valid Pi injection experiments (nos 709 and 778), which are partially overlapping (Fig. 2).
Case no. 709 (Fig. 10) received an injection in the rostral tip of the ventral bank of the pseudosylvian sulcus, a zone that corresponds to the rostral third of Pi. In the injected hemisphere, the largest collections of HRP-positive cells are situated in VPf and Te. Other labeled areas include IL or PL, Te, 35 and 36. In the opposite hemisphere, aside from homotopic labeling in Pi, the heaviest labeling is situated in neighboring area Te, with some additional cells labeled in VPf, IL and PL. Anterogradely labeled fibers overlap the regions containing retrogradely labeled somata. Most abundant in VPf, Te and 36, they are mainly distributed in layers I, III and VI.
The deposit in case no. 778 is situated roughly at the center of area Pi (Fig. 2). As in the previous case, numerous neurons are labeled in VPf, Te, 35 and 36 (Fig. 10). Additional cells are labeled in AO, in caudal portions of AS and in PL. On the other hand, unlike no. 709, large collections of cells and terminals are labeled in EP and A-2. In the opposite hemisphere, the heaviest anterograde and retrograde labeling involves areas Pi, Te, AS and 36.
Injections in the Anterior Sylvian Area (AS)
Area AS covers the rostral two-thirds of the anterior sylvian gyrus and dorsal lip of the pseudosylvian sulcus (Clascá et al., 1997), and there were six valid injection cases in this area (Fig. 2). The representative case is no. 677 and consisted of two contiguous injections that impregnated a relatively large zone in the crown of the anterior sylvian gyrus (Fig. 2). Several major arrays of labeled cells and terminals are present in the injected hemisphere (Figs 8B,D, 11 and 12). One array is spread along the lateral bank and lip of the suprasylvian sulcus and adjacent portions of the posterior ectosylvian and fusiform gyri. Most of the labeling is situated in retinotopic areas PlLS, DLS, PS and an adjacent zone3 referred to as EPp (Fig. 12B,C), while other labeling probably belongs to areas VLS, 21b and AlLS. A second array of labeled cells and terminals involves a zone of DlP (Fig. 11, 8B), with some additional cells scattered in DmP. A third array is spread on the anterior and posterior sylvian gyri (A-2, Te; Fig. 12A), and the ventral bank of the anterior ectosylvian sulcus (field VAE). There is a smaller labeling focus in the ventral lip of the splenial sulcus, a zone that would correspond to a border between Cg and the cingulate visual area4 (CVA; Fig. 12D). Further collections of labeling are situated in area 36 and lateral parts of area 35 (Fig. 8D). In the contralateral cortex, the heaviest retrograde labeling is located in AS, A-2 and VAE, and there is additional labeling scattered in PlLS, DLS, EPp and PS. However, in stark contrast to the injected hemisphere, the prefrontal and perirhinal cortices of the contralateral hemisphere are not labeled.
In general, anterogradely labeled terminals largely overlap the locations of labeled somata. Anterograde labeling is heaviest in DlP (Fig. 8B) and the suprasylvian visual areas (Fig. 12B,C), but area 36 contains few labeled terminals (Fig. 8D). In most areas, the densest anterograde labeling is seen in layers I, III and VI.
The remaining five valid experiments with an injection in AS involved some sectors of this field not affected by the injection in no. 677. Despite small differences, the resulting labeling patterns (Figs 3 and 13) are basically like the one just described for no. 677. The deposit in case no. 201 is limited to a rostral zone of AS that was not involved by the injection in no. 677 (Fig. 2). In comparison to no. 677, labeling in no. 201 (Fig. 13A) is scarcer in caudal portions of the suprasylvian sulcus, as well as in areas Te and 36. On the other hand, no. 201 has HRP-positive cells in areas not noticeably labeled in no. 677, specifically S-IV, PAE and 6aß. In a further experiment (no. 705), the injection spread over an anteroventral sector of AS within the lip of the anterior rhinal sulcus (Fig. 2). The scarcity of labeling in the posterior sylvian gyrus and suprasylvian sulcus, and the relatively large numbers of neurons labeled in the orbital gyrus (DI, GI, AId), are salient features of this case (Fig. 13C). The injection in no. 570 involved the crown of the anterior sylvian gyrus between AP +15 and +13 (Fig. 2). Scant labeling of PS and of the perirhinal cortex are the only significant departures from the pattern seen in no. 677 (Fig. 13B). A further case (no. 399) injected in a caudal and ventral portion of AS, largely spared by the injection in no. 677 (Fig. 2), yielded fairly heavier labeling in Te, EP and VAE than in no. 677, and scant labeling of AlLS, PlLS, VLS and Cg (Fig. 13D). The injection in case no. 760 impregnated a cortical territory that, as far as can be said from our reconstruction of the injection sites, is almost totally contained within the zone injected in no. 677 (Fig. 2), and the labeling resembles that of case no. 677. However, the labeled cells in auditory area A-2 and the ventral auditory field (V) in no. 760 are fairly more numerous and more dorsally located than those in no. 677 (compare Figs 11H and 13E).
Patchy Tangential Distribution of Labeled Neurons and Terminals
On the individual coronal sections, HRP-positive neurons and fibers are most often found gathered together, forming small clusters or column-like arrays of variable size, separated by zones of non-labeled or poorly labeled tissue. When serial reconstructions were made (Figs 4, 6, 7, 9–11 and 13), it became apparent that, in many areas, these aggregates of labeling corresponded to domains of variable size and shape that involve only limited portions of the labeled areas. In some areas, these domains are fairly small (~300–800 µm in diameter; Fig. 5C–E). In other areas, the labeling spreads over wider zones; however, even here, there was a tendency to waxing and waning of the labeled cells and terminals (Figs 5A, 8A,B, 12C,E) that suggests a preferential labeling of smaller cortical domains. Although the distance between the sections sampled precludes a more fine-grained assessment, present observations show that the cortical connections of the insular fields do not involve the whole extent of the labeled areas, but rather a patchwork of restricted cortical domains within these areas. These domains are irregularly shaped, are ~300–800 µm in diameter, and are separated by zones of either non-labeled or poorly labeled cortex. Together with similar findings on the connections of the somatic, auditory and visual cortices (Avendaño et al., 1988; Rouiller et al., 1991; Schwark et al., 1992, Morley et al., 1997), our observations in a variety of sensory, association and limbic areas strongly suggest that this pattern may reflect a general underlying organization of the cortico-cortical connections.
Interhemispheric Pathways Across the Cerebral Comissures
In addition to labeling in the cortical gray matter, our experiments revealed the entire course of the axons through the white matter, including the interhemispheric comissures. Figure 14 summarizes these observations. Injections in each of the areas investigated labeled two sets of interhemispheric axons. One set followed a ventral route through the external capsule, before crossing the midline in the posterior limb of the anterior commissure, while the other set follows a dorsal route, crossing the midline in the corpus callosum.5 It is noteworthy, however, that the injection limited to AIv and PpC (case no. 711) labeled only ventrally directed commissural axons.
While the paths of ventrally directed axons labeled after injections in all areas seem to be largely overlapping, paths taken by dorsally directed axons diverge markedly with each area injected (Fig. 14). Injections in AId and AIv labeled axons in the rostroventral edge of the corpus callosum. After the DI injections, labeled fibers turn dorsally and then extend across the ventral portions of the genu of the callosum. In the brains that received injections in GI, the majority of dorsally routed commissural axons surround the tapetum, and then cross the corpus callosum between AP +18.5 and +16, with some few additional axons scattered up to AP +14.5. Following injections in AS, either in the crown of the anterior sylvian gyrus or in the dorsal bank of the pseudosylvian sulcus, the main bundle of labeled axons extends first dorsocaudally and then crosses the body of the corpus callosum between AP +12 and +10, although additional labeled axons are scattered over a broader zone (AP +13 to +7.5). Finally, following injections in Area Pi, labeled fibers form a conspicuous bundle that extends first caudally in the lateroventral wall of the temporal horn of the lateral ventricle up to about AP –1, and then turns medially, around the occipital edge of the ventricle, to join the inferior branch of the forceps minor. This bundle crosses the midline through the dorsal hippocampal commissure and ventral splenium (AP +6.5 and +3.5).
Laminar Origin of Afferent Cortical Projections
The layers with the most abundant retrogradely labeled somata were, in decreasing order, layers III, II, V and VI. This pattern can be observed in Figures 4 and 5 and 7–12. To substantiate our impressions based on the observation of single sections, we counted, in each experiment, the somata labeled in the infragranular (VI–V) or supragranular (IV–II) layers of the cortex in each cortical area in both cerebral hemispheres on all the drawn sections. The ratio of supragranular to infragranular cells was calculated for each area, and then averaged among the cases injected in the same area.
The charts in Figure 15 summarize the results of this analysis. Note that, in most cortical areas, 75–95% of the projections to the insular and adjacent areas originate in layers III–II. In fact, some projections like those of field G to DI and to GI are 97–99% supragranular (Fig. 8A). There is an interesting exception, however: ~80% of the projections from the perirhinal cortex (areas 35 and 36) to areas GI, DI, AI and AS arise from neurons in layers V–VI (Fig, 15C,D).
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConcluding RemarksNotesReferences |
|---|
The main finding of this study is that, despite their relatively small size and close apposition, the insular region areas connect to largely non-overlapping sets of cortical areas, suggesting that they may be involved in rather different aspects of cortical integration. The following discussion explores: (i) the patterns of cortical connections characteristic to each field; (ii) the functional affiliations that can be inferred from the presence of such connections; and (iii) the similarities to connections described in the insular fields of Old World primates.
Methodological Considerations
Injections placed in the same area produced largely similar patterns of cortical labeling in different animals. However, our quantitative and topographical analysis of the labeling consistently reveals fluctuations between cases in the relative amount and spatial distribution of the connections labeled in particular areas. Differences in the efficiency of the axonal transport labeling method are an unlikely explanation for these fluctuations, since they would affect the global amount of labeling, rather than the amount in any particular area. In our view, these fluctuations probably reflect the combination of two factors. On one hand, the distribution of cortico-cortical connections within the injected areas may not be homogeneous, but patchy. Previous data showing that injections in distant areas selectively label small, patch-like domains of the orbital or anterior sylvian gyrus (Vicario et al., 1983; Burton and Kopf, 1984; Yasui et al., 1987; Bowman and Olson, 1988; Musil and Olson, 1992) support this interpretation. Conceivably, our small injections could have randomly involved or spared small domains with specific cortical connections, and thus led to significant shifts in the labeling pattern. On the other hand, since each experiment was carried out in a different animal, the fluctuation may also, in part, reflect genuine interindividual differences in the relative size or topographical arrangement of the various sets of cortical connections. Individual differences have been shown to be relatively substantial in cortical connections (McNeil et al., 1997). Some of the ‘inconsistencies’ in the labeling produced by injections that, as far as can be seen from our reconstructions, were overlapping, strongly suggest such individual variation. This is the case, for example, for the extensive connections labeled in areas A-2 and V in case no. 760, which were not labeled by overlapping injections in nos 677 and 570 (Figs 2, 12 and 14); or the lack of labeling in area 36 after the injection in no. 570, as compared to nos 760, 705 or 677.
The demonstrated existence of direct connections to cortical areas whose functional significance is fairly well understood strongly suggests particular functional affiliations for each of the areas under study. In addition, the relative anatomical weight of the various projections reaching a given area might be interpreted as indicative of their relative functional impact. Thus, many connections would suggest a strong functional impact, whereas less abundant connections might be interpreted as being functionally weaker. Nevertheless, it should be remembered that there is evidence that the anatomical weight of cortico-cortical connections does not always correspond to their functional strength (Vanduffel et al., 1997), suggesting that the functional impact of the various connections can dynamically change with the different behavioral conditions under which cortico-cortical systems are activated.
Highly Convergent Inputs from Face, Neck and Upper Limb-related Sensorimotor Regions Characterize the Granular Insular Area
The cortical connections of the dorsolateral portion of the orbital gyrus now identified as area GI had not previously been investigated with modern methods. Results show that this area is strongly associated with a wide array of somatic and motor districts in both cerebral hemispheres, with additional connections to dorsolateral prefrontal and perirhinal cortices (Fig. 16).
|
A heavy reciprocal pathway links GI with a zone of area 3a in the rostral tip and dorsal lip of the coronal sulcus of both cerebral hemispheres. This zone responds to cutaneous and propioceptive stimulation of the face (Landgren and Olsson, 1980
; Iwata et al., 1990b), and elicits simple face and perioral movements upon intracortical microstimulation (Woolsey, 1958; Iwata et al., 1985, 1990a). More caudal zones of area 3a in the anterior coronal gyrus, which are also connected to GI, receive propioceptive information from the forelimb (Dykes et al., 1980), project to the cervical spinal cord, and elicit movements of shoulder muscles upon microstimulation (Woolsey, 1958; Ghosh, 1997a). Connections of GI with area 3b are weaker, and mostly concentrated in a zone that reportedly processes cutaneous input from forepaw digits, forelimb and trunk (Felleman et al., 1983). Our data indicate that GI lacks direct connections to more lateral portions of areas 3a and 3b, which process tongue, teeth and jaw inputs (Felleman et al., 1983; Iwata et al., 1986, 1987; Taira, 1987), although it does have dense interconnections with area 2 in the dorsal bank of the orbital sulcus and adjacent portions of the coronal gyrus, a zone associated with intraoral somatic input (Iwata et al., 1990b; Nishikawa et al., 1993).
Connections with area S-IV are heavy and involve the whole extent of this field in both hemispheres. Area S-II connections are heavier in the anteroventral zones of the area, which have been shown to respond to cutaneous stimulation of the face and digits (Burton et al., 1982). Following i