THE JOURNAL OF COMPARATIVE NEUROLOGY 417:73 87 (2000)

Transcription

1 THE JOURNAL OF COMPARATIVE NEUROLOGY 417:73 87 (2000) Does the Visual System of the Flying Fox Resemble That of Primates? The Distribution of Calcium-Binding Proteins in the Primary Visual Pathway of Pteropus poliocephalus J.M. ICHIDA, 1 M.G.P. ROSA, 2 AND V.A. CASAGRANDE 1,3,4 * 1 Department of Psychology, Vanderbilt University, Nashville, Tennessee Vision, Touch and Hearing Research Centre, Department of Physiology and Pharmacology, University of Queensland, St. Lucia QLD 4072, Australia 3 Department of Cell Biology, Vanderbilt University, Nashville, Tennessee Department of Ophthalmology, Vanderbilt University, Nashville, Tennessee ABSTRACT It has been proposed that flying foxes and echolocating bats evolved independently from early mammalian ancestors in such a way that flying foxes form one of the suborders most closely related to primates. A major piece of evidence offered in support of a flying fox-primate link is the highly developed visual system of flying foxes, which is theorized to be primate-like in several different ways. Because the calcium-binding proteins parvalbumin (PV) and calbindin (CB) show distinct and consistent distributions in the primate visual system, the distribution of these same proteins was examined in the flying fox (Pteropus poliocephalus) visual system. Standard immunocytochemical techniques reveal that PV labeling within the lateral geniculate nucleus (LGN) of the flying fox is sparse, with clearly labeled cells located only within layer 1, adjacent to the optic tract. CB labeling in the LGN is profuse, with cells labeled in all layers throughout the nucleus. Double labeling reveals that all PV cells also contain CB, and that these cells are among the largest in the LGN. In primary visual cortex (V1) PV and CB label different classes of nonpyramidal neurons. PV cells are found in all cortical layers, although labeled cells are found only rarely in layer I. CB cells are found primarily in layers II and III. The density of PV neuropil correlates with the density of cytochrome oxidase staining; however, no CO or PV or CB patches or blobs are found in V1. These results show that the distribution of calcium-binding proteins in the flying fox LGN is unlike that found in primates, in which antibodies for PV and CB label specific separate populations of relay cells that exist in different layers. Indeed, the pattern of calcium-binding protein distribution in the flying fox LGN is different from that reported in any other terrestrial mammal. Within V1 no PV patches, CO blobs, or patchy distribution of CB neuropil that might reveal interblobs characteristic of primate V1 are found; however, PV and CB are found in separate populations of non-pyramidal neurons. The types of V1 cells labeled with antibodies to PV and CB in all mammals examined including the flying fox suggest that the similarities in the cellular distribution of these proteins in cortex reflect the fact that this feature is common to all mammals. J. Comp. Neurol. 417:73 87, Wiley-Liss, Inc. Indexing terms: chiroptera; parvalbumin; calbindin; visual cortex; lateral geniculate nucleus Grant sponsor: NIH; Grant numbers: EY01778, EY08126, and HD *Correspondence to: V.A. Casagrande, Department of Cell Biology, Vanderbilt Medical School, Medical Center North RM C2310, Nashville, TN vivien.casagrande@mcmail.vanderbilt.edu Received 19 February 1999; Revised 16 August 1999; Accepted 13 October WILEY-LISS, INC.

2 74 J.M. ICHIDA ET AL. The order Chiroptera includes all bats and is divided into two main groups, the flying foxes and the echolocating bats. The latter are relatively small, use echolocation for navigation, and encompass species with a wide variety of diets, including many that are insectivores. Flying foxes are larger in size and are commonly non-echolocating and phytophagous. The origin of these mammals has been the focus of debate over the past 10 years since Pettigrew et al. (1989) proposed that the suborders within Chiroptera evolved independently from different ancestors. This hypothesized diphyletic origin argues that flight evolved twice in mammals, once in early insectivore-like mammals, leading to echolocating bats, and again in an early primate ancestor, producing the flying foxes. The proposed flying fox-primate link comes from behavioral, physiological, and anatomical comparisons among small echolocating bats, flying foxes, and various species of primates (Pettigrew, 1986; Pettigrew et al., 1989). Among these comparisons, the flying fox visual system was regarded as strong evidence to support a flying fox-primate link (Pettigrew et al., 1989). Studies of the flying fox visual system have revealed anatomical similarities to the primate visual system at the level of the retina, lateral geniculate nucleus (LGN), and superior colliculus (SC) (Pettigrew, 1986; Pettigrew et al., 1989; Dann and Buhl, 1990). Examination of the retinal ganglion cells in flying foxes reveals three classes of cells with similarities to those found in cats and primates (Dann and Buhl, 1990); the functional properties of these cells have yet to be determined. Both flying foxes and primates have laminated LGNs. The flying fox LGN has three clearly defined cell layers containing layer-specific large and small cells, with the large cells located closest to the optic tract (Pentney and Cotter, 1981; Pettigrew et al., 1989). Pettigrew et al. (1989) describe these layers as magnocellular, parvocellular, and koniocellular layers (M, P, and K layers), similar to those found in primates (see Casagrande and Norton, 1991, for review). The degree to which the flying fox LGN organization actually resembles that of primates, however, remains a matter of debate. Based on 3 H amino acid labeling, Pettigrew et al. (1989) argue that the flying fox LGN has six layers defined by retinal input, three contralaterally innervated and three ipsilaterally innervated, just as described in primates (see Kaas et al., 1978; Casagrande and Norton, 1991, for review). Other studies of the retinogeniculate projections in flying foxes using similar methods have produced some contradictory results. Pettigrew et al. (1989) found sharp boundaries between contralateral and ipsilateral input into each of the three cellular layers. Cotter and Pentney (1979), however, found that although all three cellular layers of the LGN receive both contralateral and ipsilateral input, there are significant areas of binocular overlap within the layers, a non-primate characteristic found in some marsupials (Sanderson et al., 1984). Pentney and Cotter (1981) also report the existence of a layer of cells directly adjacent to the optic tract that receive projections from the ipsilateral eye; this pattern is unlike that found in most primates. They include this superficial (S) layer as a sublamina of their layer 1. Pettigrew (1986) also emphasizes the similarity between the organization of flying fox and primate superior colliculus (SC). Primates show a distinct pattern of projections to the SC that is different from that of almost all other mammals (Kadoya et al., 1971; Cynader and Berman, 1972; Kaas et al., 1974; Hubel et al., 1975; Cowey and Perry, 1980; see Rosa and Schmid, 1994, for review). In most mammals the topographic map in the superficial layers of the SC includes a complete representation of the contralateral retina, including all of its temporal retina. This extensive representation is usually the result of the entire contralateral retina projecting to the superficial layers of the SC. In contrast, the SC in primates represents only the contralateral visual hemifield, a characteristic that is correlated with a lack of projections from the temporal hemiretina to the contralateral SC. Using electrophysiological recordings and neuroanatomical tracers, Pettigrew (1986) found that flying foxes (genus Pteropus) exhibit the primate pattern of projections to the SC, and Rosa and Schmid (1994) confirmed, based on more extensive recordings, the lack of a representation of the ipsilateral hemifield in the rostral part of the SC. Although Thiele et al. (1991) found that flying foxes of a different genus (Rousettus) had the more general mammalian pattern of projections to the SC, their electrophysiological results have been regarded as inconclusive due to the lack of control of eye position (Rosa and Schmid, 1994). Electrophysiological mapping of the flying fox visual cortex has revealed a large and precise visuotopically organized cortex, which emphasizes the importance of vision in these bats (Rosa et al., 1993, 1994; Rosa, 1999). The goal of the present study was to examine the organization of the flying fox LGN and primary visual cortex (V1) using the calcium-binding proteins calbindin (CB) and parvalbumin (PV). The distribution of these proteins has been examined in several species of primates including Old World monkeys (Blumcke et al., 1990; Hendry and Carder, 1993; Hendry and Yoshioka, 1994; Yan et al., 1996), New World monkeys (Hendry and Carder, 1993; Goodchild and Martin, 1998), prosimians (Diamond et al., 1993; Johnson and Casagrande, 1995), and humans (Blumcke et al., 1990; Hendry and Carder, 1993; Leuba and Saini, 1996), and they have been shown to have very distinct and consistent distributions in the primate visual system (Hendry et al., 1989; Jones and Hendry, 1989; Blumcke et al., 1990; Diamond et al., 1993; Hendry and Yoshioka, 1994; Johnson and Casagrande,1995). In primates, these proteins have complementary, layerspecific distributions within the LGN, with CB located primarily in the K relay cells (the small cells lying within the interlaminar zones) whereas PV is located in the relay cells of both the M and P layers. Cytochrome oxidase (CO) staining matches PV localization with dark staining in the M and P layers and light staining in the K layers (Johnson and Casagrande, 1995). In the primate LGN there are more PV cells than CB cells, and when CB cells are found in the M and P layers they are usually smaller than the PV cells that dominate those layers, indicating that they represent different populations (see Johnson and Casagrande, 1995, for review). The distribution of PV and CB has also been examined in the LGN of nonprimate species with well-developed visual systems. The tree shrew shows a pattern somewhat similar to that of primates. CB labels the two smallest (presumptive K) cell layers and PV the other four cell layers (Diamond et al., 1993); however, PV cells in the tree shrew LGN are often found within the predominantly CB layers, a pattern that is not observed in primates. The cat LGN contains more CB cells than PV cells, and neither protein appears to be layer specific or confined to relay cells; about

3 CALCIUM-BINDING PROTEINS IN FLYING FOX LGN AND V1 75 half of cat LGN cells contain both proteins (Demeulemeester et al., 1989, 1991a). In V1 of primates, CB and PV label separate populations of GABAergic interneurons. CB cells are found primarily in the upper layers (II and IIIA), with a few cells scattered throughout the remaining layers; PV cells are found in all the layers, with a band of dense fiber labeling in layer IV, which is coincident with heavy CO reactivity in that layer (Johnson and Casagrande, 1995). Although this pattern of labeling is generally true of primates and is similar to the pattern observed in other mammals including cats (Demeulemeester et al., 1989, 1991b) and rats (Celio, 1990), there are reported differences between primate species in the distribution of labeled cells and neuropil (Hendry and Carder, 1993). CB and PV do, however, consistently label the same types of cells in cortex of all species examined (Hendry et al., 1989; van Brederode et al., 1990; Glezer et al., 1993; Yan et al., 1996). Additionally, CB and PV distinguish the functional compartments in V1 of most primates. PV neuropil is dense within the CO blobs in layer IIIB and lighter in the interblob zones; CB shows the complementary pattern with denser neuropil in the interblob regions (Hendry et al., 1989; Blumcke et al., 1990; Hendry and Carder, 1993; Johnson and Casagrande, 1995). With these patterns in mind, we can hypothesize that if flying foxes and primates do share a relatively recent common ancestor (i.e., one that lived after the separation of the major mammalian radiations), then the distribution of CB and PV in the LGN should be complementary and perhaps involve different cell layers. In cortex, the only unique primate feature we would expect to find in terms of calcium-binding proteins is one that matches CO blob and interblob compartments. MATERIALS AND METHODS Subjects and histology Four cortical hemispheres and six thalami from three flying foxes (Pteropus poliocephalus) were used in this study. The care and experimental use of animals conformed with the guidelines of the Institutional Animal Care and Use Committee at Vanderbilt University. The animals were perfused with 3% paraformaldehyde in 0.1 M phosphate buffer, ph 7.4. Following perfusion, the brains were removed and placed in a 30% sucrose/buffer solution, ph 7.4. The brains remained in the sucrose/ buffer solution at room temperature for 5 days during shipping. Upon arrival, the brains were hemisected, blocked, and stored frozen at 70 C until use. Frozen sections were cut at 40 or 10 m (double-labeled sections) on a sliding microtome. Three cortical hemispheres were cut in the parasagittal plane, and one was cut in the coronal plane. One thalamus was cut in the parasagittal plane, two in the coronal plane, and three in the horizontal plane. All sections were collected in 0.1 M Tris-buffered saline (TBS; ph 7.4). Immunocytochemistry and CO histochemistry Single and double labeling for CB and PV was performed using primary antibodies raised in rabbit for CB (CB-38 polyclonal, Swant, Bellinzona, Switzerland) and in mouse for PV (P-3171 monoclonal, Sigma, St. Louis, MO). Prior to incubation in the primary antibodies, the sections were placed in 0.3% H 2 O 2 for 30 minutes (to block endogenous peroxidase activity) and then rinsed in 0.1 M TBS. For single-labeling experiments, the primary antibodies were used at dilutions of 1:2,000 for CB and 1:5,000 for PV. All antisera were diluted in 0.1 M TBS with 4% normal horse serum, 0.2% Triton X-100, 0.5% sodium azide, and 0.2% bovine serum albumin. Sections were incubated in the primary antibody for hours at 4 C, rinsed with 0.1 M TBS, and then incubated in speciesspecific biotinylated secondary antibodies (donkey antirabbit [Chemicon, Temecula, CA] for CB and horse antimouse [Vector, Burlingame, CA] for PV; diluted at 1:200) for 1 2 hours. Following rinses with 0.1 M TBS, the sections were placed in tertiary antibody diluted with 0.1 M TBS (Elite ABC kit, PK 6100; Vector, Burlingame, CA) for 1 2 hours. Following rinses in 0.1 M TBS, the sections were visualized by a minute preincubation with 0.05% 3,3 -diaminobenzidine (DAB) 0.02% nickel ammonium sulfate in 0.1 M TBS before adding 0.003% H 2 O 2. Negative controls, in which the primary antibody was omitted, were performed as described above for each experiment, and no staining was observed in these sections. Adjacent sections were assayed in series for PV, CO, CB, and Nissl bodies. CO staining was performed using a modification (Boyd and Matsubara, 1996) of the method originally described by Wong-Riley (1979). In some cases, DAB-treated sections were enhanced with Giemsa stain using the method described by Singleton and Casagrande (1996). Photomicrographs were taken on a Zeiss Axiophot with black and white print film (Kodak TMAX). Double-label immunofluorescence Colocalization of PV and CB was examined using either fluorescently labeled secondary antibodies or biotinylated secondary antibodies followed by fluorescently labeled streptavidin. Solutions and incubation times were as described above with the following modifications. Background fluorescence was minimized by incubating the sections in 1.0% bovine serum albumin and 10% normal horse serum in 0.1 M lysine. Primary antibodies were used at dilutions of 1:5,000 for CB and 1:10,000 for PV and were applied individually. After rinses in 0.1 M TBS, the sections were incubated in a mixture of secondary antibodies at 1:200 (fluorescein isothiocyanate [FITC]- conjugated donkey anti-rabbit [Chemicon] for CB; biotinylated horse anti-mouse [Vector] for PV) for 1 2 hours. Finally, PV was labeled by incubating in either streptavidin-texas Red or streptavidin-cy3 (Amersham, Arlington Heights, IL) at 1:200 for 2 3 hours. Sections were then mounted on gelatinized slides and partially dried before being coverslipped with Vectashield (Vector). Negative controls were performed as described above and resulted in either no staining or easily distinguished artifact. Photomicrographs of fluorescent double labeling were taken on the same microscope as single-labeled sections, only under fluorescent optics. Cell measurements in the LGN Measurements of cell area were made in one singlelabeled experimental case. Cell areas were calculated using the Bioquant Image Analysis System (R and M Biometrics, Nashville, TN). Measurements were made in adjacent sections reacted for PV or CB and Giemsa counterstained. This procedure allowed us to make measurements of labeled cell populations and total cell populations

4 76 J.M. ICHIDA ET AL. within the same section, thus decreasing the potential effects of differential shrinkage due to different histological procedures. Only cells that were clearly labeled, as determined by the dark reaction product and the clarity of the cell boundary, were included in the PV and CB measurements. Cells that were non-immunoreactive for PV or CB were included in the measurement when the nucleus and nucleolus were visible. This was done to ensure that our measures were made when we focused on the center of the cell. To ensure that the same region of the LGN was measured in adjacent sections, landmarks (e.g., blood vessels) were used to align the sections and as guides for the area to be measured. Statistical comparisons were made using a two-tailed t-test for independent samples, and differences were considered statistically significant at the P 0.01 level. RESULTS Lateral geniculate nucleus: Cytoarchitecture Nissl-stained sections (Fig. 1A) reveal a laminated LGN with three distinct cell layers and two cell-sparse interlaminar zones (ILZs). This is the same laminar pattern as described by Pentney and Cotter (1981). Figure 1A shows a representative section through the LGN, where this lamination is evident. The largest layer (cytoarchitectural layer 1, encompassing the magnocellular layers of Pettigrew et al., 1989) lies adjacent to the optic tract and includes within it the S sublayer also described by Pentney and Cotter (1981). Layers 2 and 3 lie medial to layer 1 and contain cells that are smaller than those found in layer 1. The ILZ between layers 1 and 2 is larger and more clearly defined than the ILZ between layers 2 and 3. CO-reacted sections (Fig. 1B) reveal a dark layer 1 with lighter layers 2 and 3. Layers 2 and 3 are often difficult to differentiate from each other, because the smaller ILZ between these layers is not clearly demarcated by CO staining. The large ILZ between layers 1 and 2 appears pale in CO reactions except for intermittent darkly stained patches, which give it a dotted line appearance (Fig. 1B, arrows). Lateral geniculate nucleus: CB and PV distribution CB- and PV-positive cells have distinct distributions in the LGN. The few PV cells are found only in layer 1, with a concentration of darkly labeled cells directly adjacent to the optic tract in the S layer (Fig. 1C, arrowheads). PV neuropil labeling in the LGN is very dense throughout the nucleus and tends to obscure the PV cells when examined at low power (Fig. 1C); however, higher power examination reveals clearly labeled cells (Fig. 2A). CB cells are numerous and located throughout the LGN (Fig. 1D). CB cells do not show layer specificity, but in the ILZ between layers 1 and 2, patches of fibers and cells stain heavily for CB, and these patches colocalize with the light patches seen in the CO stains (Fig. 1B, arrows). Highpower inspection of the immunopositive cells in Giemsacounterstained sections reveals that both CB and PV are found in a population of the larger cells, which are often oriented perpendicular to the layers, especially in the case of CB cells (Fig. 2A,B). Examination of CB labeling at high power also shows the darkly labeled fibers in the ILZ (Fig. 2B, arrow). Figure 3 shows an example of a section double labeled for PV and CB. Figure 3A shows fluorescently labeled PV cells near the optic tract viewed under RITC optics. The same section viewed under FITC optics shows CB labeling (Fig. 3B). Examination of doublelabeled sections reveals that all PV cells are also positive for CB but not all CB cells contain PV (Fig. 3, arrows). In an additional analysis, sections from a single animal were used to examine the sizes of neurons. (Sections from other cases were not prepared in a way that allowed for accurate measurements, and it was not possible to obtain additional animals because of import restrictions due to rabies infection in this population.) Cell measurements in this case reveal that PV cells have an average size of m 2 (n 79), CB cells have an average size of m 2 (n 110), and the average cell size for the total population is m 2 (n 379). Figure 4 shows a histogram of the cell size distributions. Statistical analysis within this animal shows that there is no significant size difference (P 0.05) between PV and CB cells and that both of these populations are significantly larger (P 0.01) than the average cell size for the total population. These data, of course, cannot be generalized to the larger population, but they provide a clear hypothesis for future testing. Primary visual cortex (V1) Nissl and CO stains reveal six cortical layers in flying fox V1 (Fig. 5A,B). Layer I appears cell sparse in Nissl stains and shows light CO staining except for a dark band near the pial surface. In Nissl-stained sections, layer II is made up of a narrow, dense band of cells; layer III is large and consists of less densely packed cells than layer II. Layers II and III are not differentiated from each other in CO sections, but the overall staining is darker than layer I; there is no evidence of discontinuous staining equivalent to CO blobs or subdivisions of layer III. Layer IV is characterized by a dense band of darkly stained cells in Nissl sections and very dark CO staining. Layer V is formed by less densely packed cell bodies and contains large pyramidal cells that are visible in both Nissl and CO sections. These cells show heavy CO staining of the soma and proximal dendrites and are located in the middle of layer V (Fig. 5B). Layer VI contains two subdivisions visible in both Nissl and CO sections. The upper subdivision of VI is cell dense and stains moderately for CO; the lower subdivision contains fewer cells and has paler staining than the upper subdivision. Figure 5C and D shows the overall distribution of PV and CB fibers and cells in adjacent sections of V1. Neuropil labeling for both proteins shows some indication of laminar distribution, with PV showing clearer layer boundaries than CB. PV reveals only light staining in layer I, made up primarily of beaded dendrites from cells in lower layers (see Fig. 7B). Layers II and III both contain dense PV neuropil, but there is no differentiation between them. PV neuropil in layer IV is comparable to that in II/III but is noticeably darker than in layer V. The distribution of PV does reveal the two subdivisions of layer VI, with the upper subdivision containing slightly more neuropil than the lower subdivision. However, the boundary between the two subdivisions is often fuzzy, being clearest along the medial bank (peripheral visual field representation) of the cortex. CB staining shows light labeling in layers I and II, with no clear boundary between the two. At higher power, fibers running parallel to the

5 CALCIUM-BINDING PROTEINS IN FLYING FOX LGN AND V1 77 Fig. 1. Adjacent horizontal sections through the lateral geniculate nucleus (LGN) showing staining patterns for Nissl (A), cytochrome oxidase (CO; B), parvalbumin (PV; C), and calbindin (CB; D). Lateral is to the right and anterior to the top. A: Nissl stain showing the three cell layers in the LGN. The largest cells are located next to the optic tract in layer 1; layers 2 and 3 contain smaller cells. A large, cellsparse interlaminar zone (ILZ) separates layers 1 and 2 and a smaller ILZ separates layers 2 and 3. B: CO staining is darkest in layer 1 and lightest in the ILZ between layers 1 and 2 (white arrows). C: PV cells are sparse and located next to the optic tract (arrowheads). D: CB cells are located throughout the LGN. The ILZ between layers 1 and 2 shows patchy cell and fiber immunoreactivity for CB. Scale bar 500 m.

6 78 J.M. ICHIDA ET AL. Fig. 2. Higher power photomicrographs of PV (A) and CB (B) reacted sections. Lateral is to the right and anterior to the top. A: PV cells in layer 1. Cells are located near the optic tract B: CB cells in layer 1. Cells are oriented perpendicular to the optic tract and are distributed evenly across the layers. The patches of densely stained CB fibers in the ILZ are also visible (arrow). Scale bar 30 m. layer boundary are visible in layer I (see Figs. 6 and 8A). Layer III contains the densest CB fiber labeling, and layers IV, V, and VI show light CB fiber staining, with layer V exhibiting slightly darker staining than layers IV or VI. Again, these divisions are clearest along the medial bank of V1. CB does not reveal any subdivisions of layer VI, although there is a gradual decrease in the density of staining toward the base of the layer. Both PV and CB show labeled fibers in the white matter underlying V1. The distribution of labeled cells reveals distinct patterns for both PV and CB. PV cells are sparsely distributed throughout all six cortical layers (although rare in layer I), and CB cells are found sparingly in all layers, although there is an obvious concentration of labeled cells throughout layers II and III (Fig. 5D). This concentration of CB cells consists of darkly and lightly labeled cell populations (Fig. 6). In contrast, all PV labeled cells are heavily stained. Figures 7 and 8 show high-power photomicrographs of PV- and CB-labeled cells in different cortical layers. High-power examination of CB and PV cells reveals a mostly nonpyramidal, nonspiny morphology (Figs. 7, 8). In many cases, primary and sometimes secondary dendrites are clearly labeled, and there are some examples of labeled axons in the cortex. One example of a CB-labeled axon can be seen in Figure 8A. Although both proteins label dendrites, CB-labeled dendrites appear more completely labeled in individual sections. This could be due to better staining with CB, or it could be caused by the type of labeled cells and the cellular localization of the protein. Examination of cell morphology indicates that PV and CB label different types of cells. Darkly labeled PV cells are typically multipolar (Fig. 7A), and those in the upper layers (II and III) often extend beaded dendrites vertically into layer I (Fig. 7B). Figure 7C shows PV cells in layer V as well as PV labeled terminals surrounding large pyramidal cell bodies (which are not PV ) in layer V. This type of connectivity suggests that these PV cells are -aminobutyric acid (GABA)ergic basket cells (van Brederode et al., 1990). PV terminals on pyramidal cell somata are also seen in the supragranular layers, but they are most distinct in layer V. These terminals, although visible in the DAB-reacted sections, are greatly enhanced by the Giemsa stain intensification. CB cells often exhibit a bitufted morphology with processes extending vertically through the cortex, an example of which is shown in Figure 8B. There are, however, CB cells that exhibit more typical stellate morphology as well (Fig. 8D). In

7 CALCIUM-BINDING PROTEINS IN FLYING FOX LGN AND V1 79 Fig. 3. Section through the LGN fluorescently double labeled for PV (A) and CB (B). Lateral is to the right and anterior to the top. A and B show the same section viewed under RITC optics for PV and FITC optics for CB. Note that all PV cells are also CB. Arrows indicate the location of cells that are labeled with CB (see B) but not PV. Scale bar 40 m. general, CB cells have quite delicate processes that extend for some distance through an individual section. Double labeling indicates that PV and CB exist in separate populations of cells. Figure 9 shows the same cortical section viewed under FITC optics showing PV labeling (Fig. 9A) and under RITC optics showing CB labeling (Fig. 9B). Although bleed-through of label was present while using FITC optics, it was possible to distinguish between true labeling and bleed-through artifact on the basis of color. Fluorescence revealed the same pattern of PV and CB distribution as seen using DAB as a chromagen. DISCUSSION Distribution of PV and CB in the LGN The results show that PV and CB are differentially distributed within the flying fox LGN. PV cells are found in layer 1, the outermost layer of the LGN, and are few in number; CB cells are found throughout the LGN and are greater in number than PV cells. Fiber and neuropil labeling also differ for PV and CB, with PV neuropil found throughout the LGN and CB neuropil located in a patchy distribution in the intralaminar zone between layers 1 and 2 of the LGN. Results of double labeling with fluorescent tags reveal that all PV cells are also CB. Preliminary results also indicate that these populations consist of similarly sized cells whose size range suggests that they are relay cells. The distribution of calcium-binding proteins seen in the flying fox LGN is unlike that found in primates. The primate pattern of PV and CB distribution is distinct and consistent across species (Jones and Hendry, 1989; Diamond et al., 1993; Hendry and Yoshioka, 1994; Johnson and Casagrande, 1995). Labeling in primates is complementary and layer specific, PV cells are more common than CB cells, and, when they are found in the same layers, they almost always represent different populations of cells (Jones and Hendry, 1989; Johnson and Casagrande, 1995). In the flying fox LGN there are few PV cells, whereas CB cells are profuse; all PV cells are also CB. Although this pattern is not primate-like, there are some similarities with PV and CB distributions in the cat LGN. In the cat, CB cells are more numerous than PV cells, and the majority colocalize with PV cells. However, unlike in the flying fox, all PV cells do not contain CB (Demeulemeester et al., 1989, 1991a). Recent studies have uncovered other similarities between the flying fox and carnivore LGN, including a widespread pattern of projections from the main LGN layers to both visual areas V1 and V2 (Funk and Rosa, 1998; Manger and Rosa, unpublished observations). Besides the similarities with carnivores, the distribution of calcium-binding proteins in the flying fox LGN resembles that reported for the bottlenose dolphin. Glezer et al. (1998) found that the bottlenose dolphin LGN consists of two main divisions (magnocellular and parvocellular) and that CB cells are found in both divisions whereas PV cells are restricted to the magnocellular division. It is not known, however, whether these proteins are actually colocalized within single LGN cells in this species. The significance of the heavily labeled CB fibers in the interlaminar zone of the flying fox LGN is unclear. This aspect of the flying fox pattern resembles the pattern seen in primates, where CB fibers are localized primarily in the K layers, which lie between the main layers (Diamond et al., 1993; Johnson and Casagrande, 1995). The important distinction is that, in primates, CB labels specific cell bodies within all the K layers, as opposed to the primarily neuropil labeling within one ILZ as in the flying fox. In addition, the source of these CB fibers in the flying fox ILZ remains unclear, although they could originate in the retina since CB fibers are observed in the optic tract (Fig. 1D). The relationship between PV and CB localization and CO staining in the flying fox LGN is also quite different from that seen in primates. In primates, CO staining mirrors PV localization, with dark M and P layers and light K layers (Hendry and Yoshioka, 1994; Johnson and Casagrande, 1995; Goodchild and Martin, 1998). In the flying fox LGN, CO staining does not fully correlate with the distribution of either protein. All the main layers of the LGN stain relatively darkly for CO, whereas the ILZs remain CO pale. The only coincident CO and calciumbinding protein staining occurs within the ILZ that lies between layers 1 and 2, where the CO pale patches correlate with CB fiber patches.

8 80 J.M. ICHIDA ET AL. Fig. 4. Histogram showing the sizes of parvalbumin (PV) cells, calbindin (CB) cells, and the total cell population in the LGN measured in one animal. There is no significant difference (P 0.05) between the size of PV (mean m 2 ) and CB (mean m 2 ) cells; however, both PV and CB cells are significantly larger than the average of the total cell population (mean m 2 ; P 0.01). There is a large body of evidence showing that, in adult primates, most LGN cells containing PV and CB are relay cells, rather than local GABAergic interneurons (Jones and Hendry, 1989; Tigges and Tigges, 1991; Mize et al., 1992; Johnson and Casagrande, 1995). This evidence comes from studies of retrograde tracers from the cortex and GABA labeling in the LGN, which show that PV and CB are present in cells retrogradely labeled from the visual cortex but not in GABAergic cells within the LGN (Jones and Hendry, 1989; Johnson and Casagrande, 1995). Cell size comparisons also show that PV and CB cells are usually larger than GABAergic interneurons (Johnson and Casagrande, 1995). Unlike in primates, GABA labeling in the cat LGN shows that over half the CB cells and almost all the PV cells are also GABAergic interneurons (Banfro and Mize, 1991; Demeulemeester et al., 1991a). Our preliminary cell measurements in the flying fox (one case) reveal that PV and CB cells are among the largest in the LGN. This finding suggests that these are relay cells, although this hypothesis remains to be tested. Distribution of PV and CB in V1 Our results show a distinct pattern of PV and CB labeling in flying fox V1. Although both proteins are found in cells in all cortical layers, they exhibit differential distributions. PV and CB cells are roughly equal in number, and PV neuropil colocalizes with CO staining, being densest in layer IV. CB cells are mainly concentrated within layers II and III. Neuropil and fibers labeled with CB are most numerous in layer I and sparse throughout the other layers. There is no evidence of patchy cell or neuropil labeling in layers II and III with PV, CB, or CO staining. The laminar pattern of PV and CB labeling observed in flying fox V1 is similar to distributions seen in many other mammals including rats (Celio, 1990), cats (Demeulmeester et al., 1989, 1991b), nonhuman primates (Blumcke et al., 1990; van Brederode et al., 1990; Hendry and Carder, 1993; Goodchild and Martin, 1998), and humans (Blumcke et al., 1990; Leuba and Saini, 1996). Differing patterns of distribution have been reported in whales and insectivores, including echolocating bats (Glezer et al., Fig. 5. Parasagittal sections through V1 stained for Nissl (A), CO (B), PV (C), and CB (D). Dorsal is to the top and anterior to the left. A: Nissl stain showing the laminar arrangement of V1 in the flying fox. B: Adjacent CO section showing heavy staining in layer IV and large, dark cells in layer V. C: PV cells are distributed across layers II through VI. Neuropil staining is present in all layers with a concentration in layers III and IV. D: CB cells are located mainly in layers II and III. Scale bar 200 m.

9 Figure 5

10 82 J.M. ICHIDA ET AL. Fig. 6. CB cells in layers II and III of V1 showing the darkly and lightly stained populations of CB cells within these layers. This photo also shows the CB fibers coursing through layer I. Dorsal is to the top and anterior to the left. Scale bar 50 m. 1993; see, however, Garey et al., 1989). The patterns observed in the latter species are similar to each other, with CB cells found in layers I, II, IIIc, and V, whereas PV cells are found mainly in layers IIIc and V. In whales and insectivores, CB cells are also more abundant than PV cells, whereas in primates the ratio of PV cells to CB cells is closer to 1:1 (Glezer et al., 1993, 1998). Although cell counts were not made in the present study, qualitative evaluation of the tissue suggests that PV and CB cell populations in flying fox V1 are approximately equal. Although the laminar distribution and relative numbers of calcium-binding proteins may differ among species, the types of cortical cells labeled by PV and CB are consistent across species with PV and CB labeling subsets of GABAergic interneurons (Hendry et al., 1989; van Brederode et al., 1990; Glezer et al., 1993; Yan et al., 1996). Based on morphological analysis and comparisons with other mammals, it seems likely that PV and CB also label subpopulations of GABAergic interneurons in flying fox cortex. PV cells in cortex have been described as nonspiny, stellate cells and have been classified as basket cells and chandelier cells based on morphological characteristics and connectional patterns (DeFelipe et al., 1989a; Hendry et al., 1989; Lewis and Lund, 1990; DeFelipe, 1997). Basket cell axons often terminate on the somata of large cortical pyramidal cells (Freund et al., 1983; Hendry et al., 1989). Our observations of the morphology of PV cells with terminations on pyramidal cell somata in the flying fox suggest that these PV cells are GABAergic interneurons. Some CB cells in the cortex of other mammals have a bitufted morphology (DeFelipe et al., 1989b, 1990). In the flying fox, CB cells often have a bitufted nonpyramidal morphology, indicating that they, too, are a subpopulation of GABAergic interneurons. In other mammals, some CB cells exhibit basket cell-type morphologies, suggesting that PV and CB may label two subpopulations of basket cells in the cortex (van Brederode et al., 1990; DeFelipe, 1997). Although no CB cells with basket-like terminations on pyramidal cell somata were observed in the present study, some CB cells in flying fox V1 do have a stellate morphology. Double labeling in the current study reveals that PV and CB cell populations are indeed non-overlapping, as reported for cats (Stichel et al., 1987; Demeulemester et al., 1989), as well as primates (Johnson and Casagrande, 1995; DeFelipe, 1997). Although the distribution of these calcium-binding proteins within the layers of V1 in the flying fox is similar to that found in many other species, the flying fox lacks some specific patterns of distribution that are found in most primates. In primates, PV neuropil in layer III colocalizes with CO blobs, whereas CB neuropil colocalizes with interblobs (Celio et al., 1986; Hendry et al., 1989; Blumcke et al., 1990, 1994; Blumcke and Celio, 1992; Hendry and Carder, 1993; Johnson and Casagrande, 1995; see, however, Goodchild and Martin, 1998). In flying fox V1 there is no evidence of blobs in CO-reacted sections or in PV-

11 CALCIUM-BINDING PROTEINS IN FLYING FOX LGN AND V1 83 Fig. 7. High-power photomicrographs of PV cells in different cortical layers. Dorsal is to the top and anterior to the left. A: PV cells in layer IV of V1 displaying typical basket cell morphology. B: PV cell in layer III of V1 with fine, beaded processes extending up into layer I. C: PV cell in layer V of V1 with well-labeled terminals surrounding a neighboring pyramidal cell. These types of terminations were most visible within layer V. Scale bar 15 m in A and C; 30 m inb.

12 84 J.M. ICHIDA ET AL. Fig. 8. High-power photomicrographs of CB cells in different cortical layers. Dorsal is to the top and anterior to the left. A: CB cell in layer III of V1 with clear labeling of its axon. Also visible in this photo are the CB fibers found throughout layer I. B, C: CB cells in layers III (B) and IV (C) displaying typical bitufted morphology. D: CB cell in layer IV displaying basket type morphology. Although not common, CB basket cells were observed occasionally. Scale bar 30 m ina;15 m inb,c,andd.

13 CALCIUM-BINDING PROTEINS IN FLYING FOX LGN AND V1 85 Fig. 9. Section through V1 fluorescently double labeled for PV (A) and CB (B). A and B show the same section viewed under FITC optics for PV and RITC optics for CB. Arrows indicate cells that are labeled for CB but not PV. Cells in V1 were rarely double labeled for PV and CB. Scale bar 100 m. stained sections, nor is there evidence of a patchy distribution of CB neuropil that might reveal interblobs. Ideally, flat-mounted sections cut parallel to the cortical layers need to be studied before completely ruling out the existence of a modular organization of distribution of calcium-binding proteins, as faint periodic patterns may not be evident in cross sections (cf. Preuss and Kaas, 1996). Evolutionary implications The debate about the phylogenetic relationships among flying foxes, echolocating bats, and primates has been based on two broad lines of evidence: the molecular biological analysis of DNA and protein sequences, and comparative anatomical studies (including analyses of characteristics of the nervous system). Most recent studies based on molecular comparisons suggest that echolocating bats and flying foxes are closely related and that neither group of bats is a likely sister group of primates (Adkins et al., 1991; Mindell, et al., 1991). However, Pettigrew (1994) showed that bat DNA has a highly skewed nucleotide composition wherein the AT content makes up more than 70% of the bases. He argued that this skewed composition could create an artifactual impression of similarity between the two bat groups. Thus, earlier genetic analyses could have been confounded by a genetic bias that is present in the DNA of animals with high metabolisms and body temperatures (Pettigrew, 1994). It seems, therefore, that molecular analyses alone cannot provide an unambiguous solution to this evolutionary controversy. According to the cladistic analysis carried out by Pettigrew et al. (1989), which was mainly based on anatomical characteristics, flying foxes form a close sister group to primates, branching off after tree shrews and before prosimians. Although other anatomical characteristics were included, Pettigrew s data matrix relied heavily on visual system comparisons. For example, Pettigrew et al. (1989) highlight the fact that the LGN layers with the largest cells are located nearest to the optic tract, a character shared with primates. According to the flying primate theory, the distribution of calcium-binding proteins in the flying fox LGN should resemble that found in primates (or, at least, represent some earlier stage from which the primate pattern could have developed). The pattern should also be distinct from that found in echolocating bats. A close affinity between the flying fox and primate

14 86 J.M. ICHIDA ET AL. patterns of LGN organization has not been supported by the present study, although one might still argue that the distribution of these calcium-binding proteins in the flying fox may represent an early mammalian pattern from which the primate pattern arose. To evaluate this possibility, more studies, including the analysis of calciumbinding proteins in other mammalian groups (especially some marsupials and insectivores), as well as in echolocating bats, will be needed in order to establish the morphotypic mammalian pattern of LGN organization, with respect to calcium-binding proteins. In comparison with other mammals so far studied, the distribution of calciumbinding proteins in the flying fox LGN resembles that found in the cat, an intriguing result in view of the similarities between these species in the pattern of projections from the LGN to cortical areas V1 and V2. Indeed, as in the cat (Dreher, 1986), it can be argued that in the flying fox V1 and V2 together form the primary visual cortex, operating in parallel (Funk and Rosa, 1998). Additionally, the similarities between the flying fox and the bottlenose dolphin LGN suggest that large populations of CB cells may be representative of an early mammalian trait. In contrast to the flying fox, tree shrews have a more primate-like distribution of calcium-binding proteins in the LGN where CB labeling is confined to the CO-light, W-like cell layers that contain the smallest cells, and PV cell labeling is most dense in layers that contain dense CO stain (Diamond et al., 1993). Taken together with the present results, the latter findings support the view that, in the course of phylogeny, flying foxes branched off from the primate lineage before tree shrews did (see also Kaas and Preuss, 1993). However, as demonstrated by the debate based on the molecular biological evidence, any single piece of evidence alone cannot be taken as proof of evolutionary relationships. For instance, to use an example related to the nervous system, an analysis restricted to the visuotopy of the superficial layers of the superior colliculus would support the opposite point of view, as discussed by Rosa and Schmid (1994). Thus, analyses of comprehensive sets of anatomical and/ or molecular data are required before such evolutionary relationships can be postulated with confidence. In particular, one important step in this process needs to be the determination of the morphotypic characters in a large number of extant mammals that might suggest common ancestry. Our present observations certainly show that the proposed link between primates and flying foxes, based on comparative anatomy of the nervous system, is not straightforward. In any case, if flying foxes or tree shrews shared a more recent common ancestor with primates, this common ancestor was probably quite remote and certainly lacked most of the characteristics we usually associate with present-day primates, such as the extensive binocular overlap, welldeveloped eye-hand coordination, and large brains relative to body size (Rosa et al., 1997). ACKNOWLEDGMENTS The authors thank Dr. Jamie Boyd, Amy Wiencken, Julie Mavity-Hudson, and Rowan Tweedale for helpful comments on the paper. We also thank Jan Rosemergy for proofreading the manuscript, and we are grateful to Julie Mavity-Hudson for help with all phases of the project. V.A.C. was the recipient of NIH grant EY LITERATURE CITED Adkins RM, Honeycutt RL Molecular phylogeny of the superorder Archonta. Proc Natl Acad Sci USA 88: Banfro F, Mize RR Calbindin antibodies label specific cell classes in the cat lateral geniculate nucleus. Soc Neurosci Abstr 8:628. Blumcke, Celio MR Parvalbumin and calbindin D-28k immunoreactivities coexist within cytochrome oxidase-rich compartments of squirrel monkey area 18. Exp Brain Res 92: Blumcke I, Hof PR, Morrison JH, Celio MR Distribution of parvalbumin immunoreactivity in the visual cortex of Old World monkeys and humans. J Comp Neurol 301: Blumcke I, Weruaga E, Kasas S, Hendrickson AE, Celio MR Discrete reduction patterns of parvalbumin and calbindin D-28k immunoreactivity in the dorsal lateral geniculate nucleus and the striate cortex of adult macaque monkeys after monocular enucleation. Vis Neurosci 11:1 11. Boyd JD, Matsubara JA Laminar and columnar patterns of geniculocortical projections in the cat: relationship to cytochrome oxidase. J Comp Neurol 365: Casagrande VA, Norton TT Lateral geniculate nucleus: a review of its physiology and function. In: Leventhal AG, editor. The neural basis of visual function. London: MacMillan. p Celio MR Calbindin D-28k and parvalbumin in the rat nervous system. Neuroscience 35: Celio MR, Scharer L, Morrison JH, Norman AW, Bloom FE Calbindin immunoreactivity alternates with cytochrome c-oxidase-rich zones in some layers of the primate visual cortex. Nature 323: Cotter JR, Pentney RJ Retinofugal projections of nonecholocating (Pteropus giganteus) and echolocating (Myotis lucifugus) bats. J Comp Neurol 184: Cowey A, Perry VH The projection of the fovea to the superior colliculus in rhesus monkeys. Neuroscience 5: Cynader M, Berman N Receptive-field organization of monkey superior colliculus. J Neurophysiol 35: Dan JF, Buhl EH Morphology of retinal ganglion cells in the flying fox (Pteropus scapulatus): a Lucifer yellow investigation. J Comp Neurol 301: DeFelipe J Types of neurons, synaptic connections and chemical characteristics of cells immunoreactive for calbindin-d28k, parvalbumin and calretinin in the neocortex. J Chem Neuroanat 14:1 19. DeFelipe J, Hendry SH, Jones DG. 1989a. Visualization of chandelier cell axons by parvalbumin immunoreactivity in monkey cerebral cortex. Proc Natl Acad Sci USA 86: DeFelipe J, Hendry SH, Jones DG. 1989b. Synapses of double bouquet cells in monkey cerebral cortex visualized by calbindin immunoreactivity. Brain Res 503: DeFelipe J, Hendry SH, Hashikawa T, Molinari M, Jones EG A microcolumnar structure of monkey cerebral cortex revealed by immunocytochemical studies of double bouquet cell axons. Neuroscience 37: Demeulemeester H, Vandesande F, Orban GA, Heizmann CW, Pochet R Calbindin D-28K and parvalbumin immunoreactivity is confined to two separate neuronal subpopulations in the cat visual cortex, whereas partial coexistence is shown in the dorsal lateral geniculate nucleus. Neurosci Lett 99:6 11. Demeulemeester H, Arckens L, Vandesande F, Orban GA, Heizmann CW, Pochet R. 1991a. Calcium binding proteins as molecular markers for cat geniculate neurons. Exp Brain Res 83: Demeulemeester H, Arckens L, Vandesande F, Orban GA, Heizmann CW, Pochet R. 1991b. Calcium binding proteins and neuropeptides as molecular markers of GABAergic interneurons in the cat visual cortex. Exp Brain Res 84: Diamond IT, Fitzpatrick D, Schmechel D Calcium binding proteins distinguish large and small cells of the ventral posterior and lateral geniculate nuclei of the prosimian galago and the tree shrew (Tupaia belangeri). Proc Natl Acad Sci USA 90: Dreher B Thalamocortical and corticocortical interconnections in the cat visual system: relation to the mechanisms of information processing. In: Pettigrew JD, Sanderson KJ, Levick WR, editors. Visual neuroscience. Cambridge: Cambridge University Press. p Freund TF, Martin KA, Smith AD, Somogyi P Glutamate decarboxylase-immunoreactive terminals of Golgi-impregnated axoaxonic cells and of presumed basket cells in synaptic contact with pyramidal neurons of the cat s visual cortex. J Comp Neurol 221: