Parvalbumin, calbindin, or calretinin in cortically projecting and GABAergic, cholinergic, or glutamatergic basal forebrain neurons of the rat

Animals and surgery

Fourteen adult male Wistar rats (Charles River, St. Constant, Canada), weighing approximately 250 g were used in this study. The rats were anesthetized with sodium pentobarbital (Somnotol, 60–67 mg/kg, intraperitoneally) for all surgical procedures and for perfusion. The McGill University Animal Care Committee and the Canadian Council on Animal Care approved all procedures.

For the study of the CBP-containing cells that project to the cerebral cortex with the use of immunoperoxidase techniques, eight rats were first operated on under general anesthesia for injection of the retrograde tracer, CT b subunit (choleragenoid; LIST Biological Laboratories, Campbell, CA). The tracer was placed into the prefrontal cortex in four rats (CT22, CT25, CT26, and CT27) and into the parietal cortex in four rats (CT24, CT29, CT30, and CT31). Injections were performed under stereotaxic guidance according to coordinates for the right prefrontal cortex: anterior–posterior (AP), +11.5; vertical (V), +5.5; lateral (L), −0.75 (with reference to ear bar zero); and for the right parietal cortex: AP, −1.0; V, −3.0; L, −4.6 (with reference to bregma; see Paxinos and Watson, 1986). CT was dissolved as a 1% solution in Tris buffer (pH 7.5) and injected in a volume of 100 nl over 5–10 minutes through a 1-μl Hamilton microsyringe that was fitted with an oil-filled micropipette with a tip diameter of 50–100 μm. Two rats (CT22 and CT24) also received colchicine treatment approximately 24 hours after the CT injections and approximately 24 hours before death. All rats were killed under anesthesia by perfusion with a paraformaldehyde fixative approximately 48 hours after the CT injections.

For the study of the colocalization of CBPs and the transmitter enzymes using immunofluorescent staining, six rats were used (CT80, CT81, CT82, CT83, PP1, and PP2), four of which (CT80, CT81, CT82, CT83) received colchicine treatment approximately 24 hours before death and were killed under anesthesia by perfusion with a Zamboni fixative.

As previously described (Gritti et al., 1997), colchicine (Sigma, St. Louis, MO) was injected in a dose of 50 μg in 25 μl of saline solution into the left ventricle of anesthetized rats according to stereotaxic procedures (coordinates AP −0.8, V −4.4, L +1.5, with reference to bregma). Colchicine was found to enhance GAD immunostaining within nerve cell bodies but to not alter that of CBPs for which it was deemed unnecessary. Although it was also found to be unnecessary for revealing GAD when using the anti-GAD67 antibody in paraformaldehyde-fixed tissue (Gritti et al., 1997), colchicine was important for revealing GAD with this antibody in the Zamboni-fixed tissue used for the immunofluorescent staining of PAG.

Perfusion and fixation

Brains used for immunoperoxidase staining of CBP and CT (from eight rats) were fixed with a paraformaldehyde solution. After a brief rinse with phosphate buffered saline (0.1 M, containing 20 mM MgCl₂, pH 7.4), the animals were perfused through the ascending aorta with 1,000 ml of 3% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4 over ∼30 minutes), followed by 10% sucrose in phosphate buffer (pH 7.4, over ∼15 minutes), as previously described (Gritti et al., 1993, 1997). The brains were removed and immersed overnight in 30% sucrose in phosphate buffer for cryoprotection. Brains were subsequently frozen at −50°C with isopentane and stored at −80°C.

Brains used for immunofluorescent staining of CBP and GAD, PAG, or ChAT (from six rats) were fixed with Zamboni's solution, according to a slight modification of the procedure developed by Kaneko and colleagues for immunostaining of the PAG enzyme (Kaneko and Mizuno, 1988; Kaneko et al., 1989; Manns et al., 2001). After a brief rinse with phosphate buffered saline, the animals were perfused through the ascending aorta with 500 ml of a modified Zamboni's solution containing 0.3% paraformaldehyde and 75% saturated picric acid in 0.1 M sodium phosphate buffer (pH 7.0, over ∼30 minutes). The brains were postfixed overnight at 4°C in a solution of 3% paraformaldehyde with 75% saturated picric acid. Brains were subsequently immersed in a solution of 30% sucrose at 4°C for approximately 48 hours for cryoprotection. Brains were then frozen at −50°C and stored at −80°C.

Immunohistochemistry

Analysis

Distribution of CBP-containing neurons

Through the cholinergic cell area of the BF, where cholinergic and GABAergic cells are codistributed, there were neurons containing the CBPs Parv, Calb, or Calret. The CBP⁺ cells were of different sizes and shapes, from small to large and oval, fusiform, or polygonal (Fig. 1A,C,E). The Parv⁺ BF cells often were multipolar in shape and appeared to be commonly and relatively large in size (Fig. 1A; with a maximum large diameter of 29.6 μm and mean ± standard deviation of 16.9 ± 3.7 μm; Table 1). Calb⁺ BF cells were highly variable in shape and size but, in addition to many small cells, included relatively large multipolar cells (Fig. 1C; with a maximum large diameter of 27.7 μm and 14.8 ± 3.6 μm; Table 1). The Calret⁺ BF cells also were variable, often oval or fusiform and bipolar (Fig. 1E) and, in addition to many small or medium cells, included some large cells (with a maximum large diameter of 25.8 μm and 15.5 ± 3.7 μm; Table 1).

Through the BF, the CBP⁺ cell groups were distributed throughout the cholinergic cell nuclei (Figs. 2, 3, 4), where they appeared to be more numerous than GAD⁺ cells (Gritti et al., 1993). According to estimates of total numbers of cells in one brain (Table 1), Parv⁺ (∼23,000), Calb⁺ (∼72,000), and Calret⁺ (∼27,000) cells greatly outnumbered GAD⁺ and ChAT⁺ cells in the BF nuclei (estimated at ∼39,000 and ∼18,000 according to similar calculations by Gritti et al., 1993). This also was the case with the exclusion of cells in the olfactory tubercle, where many small Calb⁺ cells are located, for BF totals of 18,243 Parv⁺, 34,690 Calb⁺, 22,527 Calret⁺, 30,791 GAD⁺; and 15,056 ChAT⁺.

Distribution of cortically projecting CBP-containing neurons

To determine whether the different CBP-containing cell groups comprised cortically projecting neurons, retrograde labeling with CT was examined in CBP⁺ neurons by dual immunostaining using immunoperoxidase in brains with CT injections into the prefrontal or parietal cortex.

CT injection sites.

The position and extent of the CT injections in the prefrontal and parietal cortices were assessed by the CT deposit in the cortex (Fig. 5) and then by the distribution of retrogradely labeled projection neurons in the contralateral cortex and in the ipsilateral thalamus within each brain, as previously documented (Gritti et al., 1997).

Injections were centered in the medial prefrontal cortex in four rats. The CT deposit was commonly located in the medial/ventral orbital cortex (MO/VO) and extended into the prelimbic cingulate cortex, area 3 (Cg3), more dorsally (Fig. 5A), similarly to previously published prefrontal cortex injection sites (Gritti et al., 1997). Across rats the injection sites also overlapped to varying degrees into the ventrolateral orbital (VLO) cortex laterally (CT25 and CT26), the infralimbic (IL) cortex caudally (CT25 and CT26) and the prelimbic cingulate cortex, area 1 (Cg1), cortex dorsally (CT27). Retrogradely labeled corticocortical projection neurons were located in the contralateral homotypic cortex, including MO/VO (CT22, CT25, CT26, and CT27), VLO (CT25 and CT26), Cg3 (CT22, CT25, CT26, and CT27), IL (CT25 and CT26), and Cg1 (CT27). Retrogradely labeled thalamocortical projection neurons were commonly located in the mediodorsal nucleus (MD), the centrolateral nucleus (CL), and the midline, reuniens (Re), and rhomboid (Rh) nuclei. A few retrogradely labeled neurons were also observed in different animals in the paratenial (PT), paraventricular (PV), centromedial (CM), paracentral (PC), anteromedial (AM), interanteromedial (IAM), intermediodorsal (IMD), and ventromedial (VM) nuclei.

Injections were centered in the parietal cortex in four rats. The CT deposit was located over the primary somatosensory cortex, parietal cortex 1 (Par1, Fig. 5B), similarly to previously published parietal cortical injection sites (Gritti et al., 1997). All injections (CT24, CT29, CT30, and CT31) were centered in the dorsomedial region of Par1. The retrogradely labeled corticocortical projection neurons were located predominantly in the contralateral homotypic Par1 cortex. In all rats, the retrogradely labeled thalamocortical neurons were located predominantly in the ventroposterior lateral and medial nuclei (VPL and VPM) and in the posterior (PO), centromedial (CM), centrolateral (CL), and PC (paracentral) nuclei.

Colocalization of CBPs with GAD, ChAT, or PAG

To determine whether the CBPs are colocalized with the synthetic enzymes for the neurotransmitters GABA, acetylcholine, or possibly glutamate in BF neurons, dual immunostaining was performed with immunohistofluorescence for the CBPs and EPs. Dual immunostaining was assessed in individual CBP⁺ cells across the entire population through the BF nuclei, except cells in the olfactory tubercle (Table 2).

In sections dually immunostained for CBPs and GAD, a very large number (>90%) of Parv⁺ neurons was GAD⁺ (Fig. 10A,B; Table 2). In contrast, a very small number (<5%) of Calb⁺ neurons was GAD⁺ (Fig. 10E,F; Table 2). Similarly, a very small number (<10%) of Calret⁺ neurons was GAD⁺ (Fig. 10I,J; Table 2).

In sections dually immunostained for Calb and ChAT, no Calb⁺ neurons were ChAT⁺ (not shown).

In sections dually immunostained for CBPs and the synthetic enzyme for the transmitter glutamate, PAG, a proportion (∼50%) of Parv⁺ cells was PAG⁺ (Fig. 10C,D; Table 2). Slightly less than half (∼45%) of Calb⁺ neurons was PAG⁺ (Fig. 10G,H; Table 2). Most Calret⁺ (>85%) neurons were PAG⁺ (Fig. 10K,L; Table 2).

Methodologic considerations

The initial aim of the present study was to determine whether subgroups of GABAergic neurons contain different CBPs that would allow their subsequent differentiation and identification. The findings in immunoperoxidase-stained material indicated that estimated numbers of CBP-containing neurons greatly exceeded previously estimated numbers of GAD⁺ cells. To permit this comparison, we plotted cell profiles with the aid of the same two-dimensional image analysis system and estimated cell numbers with the same method (Abercrombie, 1946) used in our original work for estimating GAD⁺ and ChAT⁺ cells in delineated nuclei of the BF (Gritti et al., 1993, 1994, 1997). Although such methods are increasingly being replaced by three-dimensional stereologic counting techniques (Coggeshall and Lekan, 1996), they nonetheless allow approximate estimates of cell numbers and are deemed appropriate when they are employed as in the present study for comparison with data obtained in the same manner (Guillery and Herrup, 1997). This comparison indicated that the combined numbers of the CBP⁺ cell groups greatly exceeded the number of GAD⁺ cells and that the combined proportions of cortically projecting CBP⁺ cell groups greatly exceeded that of the cortically projecting GAD⁺ cells. Given these results, we examined in dual-immunofluorescent stained material whether CBP⁺ cells were also immunopositive for GAD, ChAT, or PAG. In this examination by fluorescence microscopy, which did not allow processing by our image analysis system, we made no attempt to estimate cell numbers but simply calculated the proportions of double-labeled cell profiles, thereby permitting an estimation of the proportion of CBP⁺ cells with the capacity to synthesize GABA, acetylcholine, or glutamate, as discussed below.

Distribution of CBP-containing neurons

Parv⁺, Calb⁺, and Calret⁺ neurons varied in shape and size in the BF, as found by others in the medial septum–diagonal band nuclei (Kiss et al., 1990b, 1997), magnocellular basal nuclei (Celio, 1990; Chang and Kuo, 1991), and other forebrain structures (Freund and Buzsaki, 1996; DeFelipe, 1997; McDonald and Betette, 2001). Of the three groups of CPB-containing cells in the BF, the Parv⁺ neurons appeared to be the largest and most homogeneous in cell size, thus most closely resembling the prominent, magnocellular GAD⁺ cells in the BF (Gritti et al., 1993).

The CBP⁺ cell groups, like GAD⁺ cells, are distributed coextensively with cholinergic cells throughout the BF, thus substantiating reports by others of this distribution in the medial septum–diagonal band (Kiss et al., 1990b, 1997) and magnocellular basal nuclei of the rat (Celio, 1990; Chang and Kuo, 1991; Smith et al., 1994; Zaborszky et al., 1999). Whereas the distribution of the Parv⁺ cells resembled that of the large GAD⁺ cells, that of Calb⁺ and Calret⁺ cells could not be simply likened to that of particular GAD⁺ cells.

The total number of CBP⁺ neurons (estimated at ∼117,000) across the BF greatly exceeded that of the GAD⁺ neurons (∼39,000; Gritti et al., 1993). A larger number of CBP⁺ cells than of GAD⁺ cells in the BF was noted by Zaborszky et al. (1999) and interpreted by them to mean that the number of GAD⁺ cells had been previously underestimated by us. It appeared more probable to us that neurons in addition to GAD⁺ neurons might be immunostained for CBPs, in particular for Calb and Calret. It was thus considered possible that Calb, contained in the largest number of cells, could also be contained in cholinergic neurons in the rat brain, as found in the monkey brain, despite claims to the contrary (Chang and Kuo, 1991; Ichitani et al., 1993). Given that the total number of CBP⁺ neurons actually exceeded the total number of GAD⁺ and ChAT⁺ neurons across the BF, it also appeared possible that other non-GABAergic and non-cholinergic neurons could contain CBPs.

Distribution and number of cortically projecting CBP-containing neurons

Parv⁺, Calb⁺, and Calret⁺ neurons were retrogradely labeled from the cerebral cortex, and Parv⁺ and Calb⁺ neurons were labeled in substantial numbers and thus potentially representing the GABAergic cortically projecting cell group, in contrast to Calret⁺ neurons, which were retrogradely labeled in numbers considered too small to pursue systematically. It was also apparent in different dual-immunostained series that different numbers of retrogradely labeled cells were apparent, suggesting a differential sensitivity in the dual immunostaining for reasons of antibody competition and/or displacement. Accordingly, total numbers of retrogradely labeled neurons differed between the Parv-CT and Calb-CT series of sections, which were thus considered independently for the calculation of proportions of retrogradely labeled cells. However, we cannot be sure that these proportions were not biased by the different sensitivities and thus should be considered as estimates. We also noted, as described previously, that, of the total population of retrogradely labeled cells, many were lightly retrogradely labeled with CT (Gritti et al., 1997), which is a very sensitive retrograde tracer. To be sure that the proportions of retrogradely cells calculated from the total numbers were not determined by a different density of retrograde labeling and thus projection, we also checked the proportions of the darkly retrogradely labeled cells. As established previously for the GAD⁺ and ChAT⁺ cells, the relative proportions did not change when comparing groups of darkly retrogradely labeled CBP⁺ cells. We interpreted these results to indicate that, like the GAD⁺ and ChAT⁺ cells, the Parv⁺ and Calb⁺ cell groups comprise substantial proportions of cortically projecting neurons that provide a dense terminal innervation to a focused cortical target area in addition to a potentially more dispersed innervation to surrounding areas.

Although they varied to some degree in size and shape, Parv⁺, Calb⁺, and even the Calret⁺ cortically projecting cells, like the GABAergic and cholinergic cortically projecting cells of the BF, were generally medium to large in size. Despite differences attributed to variations in injection site and sensitivity of dual-immunostaining, retrogradely labeled Parv⁺ cells and Calb⁺ cells were approximately similarly distributed through the nuclei of the BF from the medial septum to the globus pallidus and overlapped in their distribution with that of GABAergic and cholinergic neurons of the BF projecting to the prefrontal and parietal cortexes (Gritti et al., 1997).

Through the medial septum to the globus pallidus, Parv⁺ cortically projecting cells were present in substantial numbers, representing an average of ∼40–45% of the cortically projecting cells across the BF. This proportion was similar to, albeit greater than, that of the GAD⁺ cortically projecting neurons (∼30–37%), as calculated in our previous study (Gritti et al., 1997). The results indicated that the Parv⁺ cells could account for the GABAergic cortically projecting neurons of the BF. A similar conclusion was reached for the GABAergic neurons in the medial septum–diagonal band nuclei projecting to the hippocampus (Freund, 1989). Calb⁺ cortically projecting cells also were present throughout the BF and accounted for ∼35% of cortically projecting neurons. These results were at odds with those of Smith et al. (1994) who did not find retrograde labeling of Calb⁺ cells in the BF. In view of our findings, such a negative result is surprising. It may be attributed to a lower sensitivity of the retrograde tracer (fluorogold) used or of the dual immunostaining for that tracer with Calb, which we also found to be less sensitive than that for CT-Parv. Given that the proportions of Calb⁺ and Parv⁺ retrogradely labeled cells were ∼30–45%, it appeared that both could not correspond to the cortically projecting GABAergic cells or that Calb and Parv could be colocalized in the same magnocellular GABAergic neurons. This possibility cannot be ruled out because such colocalization has been documented in the amygdala (McDonald and Betette, 2001; McDonald and Mascagni, 2001). The Calret seen in a small number of retrogradely labeled cells also could be colocalized in the same cells, given evidence for colocalization of the three CBPs in neurons within certain forebrain regions (Wouterlood et al., 2000). However, such colocalization appears less likely in the BF neurons because it is relatively rare in other forebrain structures (Gulyas et al., 1991; Kubota et al., 1993; DeFelipe, 1997) and not at all within the medial septum–diagonal band nuclei (Kiss et al., 1997). Thus, Parv⁺, Calb⁺, and Calret⁺ cortically projecting neurons likely corresponded predominantly to distinct cell subgroups and that those subgroups might not be all GABAergic.

Colocalization of CBP with ChAT, GAD, or PAG in the BF

Given the high degree of colocalization of Parv, Calb, and Calret with GAD or GABA in the neocortex, hippocampus, and the neostriatum (Kubota et al., 1993; Freund and Buzsaki, 1996; DeFelipe, 1997), we expected that such colocalization would also be frequent in the BF. Although this was the case for Parv, it was not for Calb or Calret. Of the Parv⁺ cells throughout the BF, more than 90% were dual immunostained for GAD in the present study, indicating that the vast majority of Parv⁺ cells has the capacity to synthesize GABA. In previous studies using dual immunostaining of anterogradely labeled septohippocampal fibers and terminals, it was believed that the Parv⁺ fibers emerging from the septum–diagonal band neurons contained GABA (Freund, 1989) and that the Parv⁺ neurons thus corresponded to the GABAergic septohippocampal projection neurons. Our present transmitter enzyme immunohistochemical and retrograde transport results also indicated that the Parv⁺ BF neurons have the capacity to synthesize GABA and appear to correspond in large part to the basal cortical GABAergic projection neurons. In contrast, few Calb⁺ cells (<5%) contained GAD. Such minimal colocalization of Calb with GAD had been noted in the magnocellular basal nucleus (Smith et al., 1994). We also confirmed that Calb⁺ neurons do not colocalize ChAT in the rat brain, although they had been found to do so in the primate brain (Chang and Kuo, 1991; Smith et al., 1994). We thus concluded from our retrograde and transmitter enzyme immunohistochemical results that Calb is predominantly contained within a distinct group of non-GABAergic, non-cholinergic basal cortical projection neurons. Like Calb⁺ cells, few Calret⁺ cells (<10%) contained GAD, leading us to conclude that Calret is contained predominantly in a distinct subgroup of non-cholinergic, non-GABAergic, caudally or locally projecting BF neurons.

Having recently established that the enzyme used in the synthesis of the transmitter glutamate, PAG, is localized in BF cortically projecting neurons (Manns et al., 2001), we sought to establish whether this enzyme was contained in the different CBP⁺ cell groups. In studies within the neocortex, where all CBP⁺ neurons also contained GAD, none contained PAG (Kaneko et al., 1992). The situation was different in the present study. Approximately half of the Parv⁺ cells was also immunopositive for PAG, indicating that a proportion of these cells may have the potential to synthesize glutamate as a neurotransmitter. For GAD⁺ neurons, a similar proportion was found to contain PAG, indicating that, as in other areas of the brain, some BF neurons may have the capacity to synthesize GABA and glutamate (as discussed by Manns et al., 2001). Whether such cells use glutamate as a neurotransmitter can be determined only by immunostaining for the glutamate vesicular transporter proteins, which are evident in nerve terminals but not in cell bodies (Fremeau et al., 2001; B.E. Jones, unpublished observations). Among all Calb⁺ cells, slightly less than half was immunopositive for PAG, indicating that a significant proportion of these cells have the capacity to synthesize glutamate as a neurotransmitter. Because most cortically projecting neurons were found to be PAG⁺ (Manns et al., 2001), we concluded that the Calb⁺ cortically projecting neurons most probably contain this enzyme and are thus potentially glutamatergic. Such cells may parallel in their projections and actions the Calb⁺ glutamatergic neurons of the thalamus (Frassoni et al., 1997), which provide a widespread projection to the superficial layers of the cerebral cortex (Rausell and Jones, 1991; Jones, 1998). The vast majority of Calret⁺ neurons was immunopositive for PAG and not retrogradely labeled from the cortex, thus likely corresponding to glutamatergic neurons with caudally or locally directed projections. Such a possibility has been considered in the piriform and entorhinal cortex, where, in addition to containing glutamate, Calret⁺ cells were found to provide asymmetric, and thus excitatory, synapses to local cortical neurons (Frassoni et al., 1998; Wouterlood et al., 2000).

Functional significance of CBP content in BF cell groups

In the hippocampus and neocortex, neurons containing the CBPs, which are commonly GABAergic interneurons, have been distinguished in their electrophysiologic properties from those lacking CBPs, i.e., the pyramidal neurons, by their frequently continuous and fast spiking discharge (Celio, 1986; Kawaguchi et al., 1987; Kawaguchi and Hama, 1988; Kawaguchi and Kubota, 1993; Freund and Buzsaki, 1996). Indeed, the CBPs serve to buffer Ca²⁺ and accordingly prevent Ca²⁺-activated K⁺ currents from hyperpolarizing the membrane and inhibiting spike generation (Baimbridge et al., 1992). CBPs accordingly can also shape discharge patterns and in their absence or paucity permit phasic discharge by neurons (Li et al., 1995). Throughout the forebrain, Parv⁺ neurons are commonly fast spiking cells, as evident from in vitro and in vivo recording studies of identified neurons (Celio, 1986; Kawaguchi et al., 1987; Kawaguchi, 1993; Kawaguchi and Kubota, 1993; Sik et al., 1995). In the medial septum–diagonal band, Parv⁺ cells have been found to be fast spiking by in vitro recordings (Morris et al., 1999) and in the BF (in one cell) by in vivo recordings (Duque et al., 2000). In our in vivo recording studies, we identified a large number of cortically projecting GAD⁺ neurons that discharge tonically at relatively high rates (average > 30 Hz) in association with cortical activation (Manns et al., 2000a) and thus could correspond to the cortically projecting, Parv⁺ BF neurons identified here. ChAT⁺ cells that do not contain the CBPs in the rat brain (Kiss et al., 1990b, 1997; Smith et al., 1994) discharge in a slow tonic manner and in a phasic bursting manner in association with cortical activation (Manns et al., 2000b). Other cortically projecting BF neurons that were GAD⁻ and presumed noncholinergic cells discharged tonically at a moderate rate and may correspond to Calb⁺ neurons (Manns et al., 2000a), which in other forebrain areas have been shown to discharge tonically (Li et al., 1995). Conversely, in the neocortex and hippocampus, Calb⁺ neurons displayed phasic discharge properties in addition to fast spiking (Kawaguchi and Kubota, 1993; Sik et al., 1995; Freund and Buzsaki, 1996). Thus, recently discovered GAD⁻, ChAT⁻, PAG⁺ cortically projecting BF neurons that were found to discharge at a moderate rate (average ∼ 20 Hz) and to discharge in a phasic manner in association with cortical activation (Manns et al., 2000c) could also be Calb⁺ cells.

CBPs also have the capacity to prevent or attenuate damage to cells due to excessive entry of Ca²⁺ during intense activity (Scharfman and Schwartzkroin, 1989). Such protection has been thought to underlie the selective survival and, conversely, selective vulnerability of neurons containing or lacking the different CBPs (for review, see Morrison et al., 1998). Given that Parv⁺ and Calret⁺ neurons appear the most resistant and Calb⁺ neurons the least among CBP-containing neurons in the cortex of patients with Alzheimer's disease, it is also possible that similarly different degrees of resistance would apply to BF neurons. Thus, Parv⁺ GABAergic cortically projecting and Calret⁺ caudally or locally projecting neurons, documented here in the rat brain, might be expected to be more resistant than the Calb-containing cholinergic neurons in the human brain (Ichimiya et al., 1989) and other Calb⁺ potentially glutamatergic cortically projecting neurons therein.

Before gleaning more precise knowledge concerning the discharge properties and vulnerability to damage of the specific CBP-containing BF cell groups, the present results provide important information concerning the existence of noncholinergic, non-GABAergic basal cortical projection neurons and an immunohistochemical means of identifying them. From these results, it can be concluded that the basal cortical projection comprises, in addition to cholinergic neurons, likely GABAergic Parv⁺ and possibly glutamatergic Calb⁺ neurons. The existence of these noncholinergic basal cortical projection neurons may explain the finding that selective lesions of cholinergic neurons alone do not produce the serious attention and memory, as well as electroencephalographic, deficits of nonselective lesions of the entire BF cell population do in the rat (Dunnett et al., 1991; Lee et al., 1994; Voytko et al., 1994; Baxter et al., 1995). Accordingly, in addition to cholinergic and likely GABAergic Parv⁺ neurons, possibly glutamatergic Calb⁺ basal cortical projection neurons may play a critical role in modulating cortical activity and facilitating arousal, attention, and memory, functions that are greatly compromised with the degenerative changes occurring in the BF with Alzheimer's disease.