Anatomical relationship between specialized astrocytes and leptin-sensitive neurones

Abstract

We have previously reported that a specialized subpopulation of astrocytes in the arcuate nucleus of the hypothalamus show an unusually intense immunoreactivity for brain fatty acid binding protein (bFABP). Since bFABP has been shown to regulate the activity of an enzyme, fatty acid synthase, that has a potent influence upon the regulation of feeding by the hypothalamus, it was of interest to determine if bFABP + astrocytes are positioned to potentially influence the activity of feeding-regulating neurones. In this study, we examined the anatomical relationship between specialized arcuate astrocytes immunoreactive for bFABP and feeding-regulating neurones that are responsive to leptin and which are immunoreactive for the transcription factor STAT3. The results show that both cell types are abundant in the arcuate nucleus of the hypothalamus and are frequently closely adjacent to each other. This study provides an anatomical basis for the possibility that specialized arcuate astrocytes regulate the function of leptin-sensitive, feeding-regulating neurones of the arcuate nucleus.

Introduction

Recent data have shown that molecules released from astrocytes have a potent influence upon neuronal function. For example, the metabolic breakdown of astrocyte glycogen into lactate and the transport of lactate from astrocytes to neurones allows astrocytes to protect neurones from glucose deprivation (Wender et al. 2000. Also, transport of cholesterol from astrocytes to neurones has a stimulatory effect upon synapse formationMauch et al. 2001

This astrocyte–neurone relationship may be of particular importance in the hypothalamus. A number of hypothalamic functions, such as the control of feeding behaviour and the secretion of pituitary and pancreatic hormones, are influenced by circulating levels of glucose. These effects of glucose may be mediated or modulated by unusual astrocytes possessing high-capacity glucose transporter proteins (GLUT2 transporters) that are present in the arcuate nucleus of the hypothalamus (Leloup et al. 1994; Ngarmukos et al. 2001). Uptake and metabolism of glucose by these astrocytes and subsequent transport of metabolites from astrocytes to neurones appears to influence hypothalamic function. Pharmacological manipulation of astrocyte function or of GLUT2 expression alters functional responses of the hypothalamus to glucose Young, 1988, 1989; Leloup et al. 1998; Wan et al. 1998; Young et al. 2000

Responsiveness of hypothalamic neurones to other nutrient molecules, such as lipids, has been less well studied, but may also be affected by astrocytes. Intraventricular infusion of oleic acid depresses food intake (Obici et al. 2002). Also, if hypothalamic concentrations of malonyl-CoA are increased by inhibiting hypothalamic fatty acid synthase, feeding behaviour is reduced by 90% (Loftus et al. 2000; Kumar et al. 2002). Factors that regulate hypothalamic levels of lipid molecules and fatty acid synthase activity may thus be of considerable importance in the control of feeding behaviour by the hypothalamus. Glucose itself is one such factor (Ferre, 1999). Also, a protein found in GFAP + astrocytes, brain fatty acid binding protein (bFABP), strongly regulates fatty acid synthase activity and is particularly abundant in a subset of hypothalamic astrocytes (Kurtz et al. 1994; Young et al. 1996; Mukhopadhyay & Mukherjea, 1998). These astrocytes are also immunoreactive for acyl-CoA binding protein, a lipid-transporting protein that is commonly co-expressed with FABP in lipidtransporting cells (Yanase et al. 2001; Young, 1994). The mRNA for acyl-CoA binding protein is highest in the arcuate nucleus Tong et al. 1991). Thus arcuate astrocytes possess many features that may modulate the hypothalamic response to glucose and lipid molecules.

The goal of this study was to see if astrocytes immunoreactive for bFABP showed any anatomical relationship with feeding-regulating neurones. Such neurones can be identified by their possession of receptors for a feeding-regulating hormone, leptin, which is secreted by adipocytes. Leptin-sensitive arcuate neurones are immunoreactive for leptin receptors and also for a leptin-stimulated transcription factor termed STAT3 (Signal Transducer and Activator of Transcription3) (Hakansson & Meister, 1998). Changes in hypothalamic lipids affect the function of numerous subsets of leptin-responsive neurones containing feedingregulating neurotransmitters such as neuropeptide Y, proopiomelanocortin and agouti-related peptide (Kumar et al. 2002). Astrocytes show functional responses to lipids like oleic acid, but do not appear directly responsive to leptin itself (Velasco et al. 2000; Suzuki et al. 2001).

Materials and methods

In this study, four adult female Sprague-Dawley derived albino rats were overdosed with chloral hydrate and perfused intracardially with 10% formalin in 0.2 M phosphate buffer. Serial 100-μm-thick coronal sections of brains were cut into phosphate buffer using a vibratome. Subsequently, free-floating sections were incubated in phosphate-buffered saline (PBS) containing 0.1% Triton-X 100 and 2% normal goat serum for 30 min, and then were placed into either non-immune rabbit serum, diluted 1 : 2000 in PBS—0.1% Triton—2% normal goat serum, rabbit antibody to bFABP, diluted 1 : 2000 in the same solution, or a rabbit polyclonal antibody to STAT3, diluted 1 : 15 000 (Santa Cruz Biotechnology) (Hakansson & Meister, 1998; Young et al. 1996).

After incubation in these solutions overnight at 4 °C, sections were washed in PBS−0.1% Triton, incubated for 30 min in biotinylated goat-anti rabbit IgG (Vectastain ABC kit, Vector Laboratories, Burlingame, CA, USA) and further processed as described elsewhere (Young et al. 1996). Initially, sections were stained to demonstrate immunoreactivity for only bFABP or for STAT3. After confirmation under the microscope that immunoreactivity for bFABP was present only in astrocytes, and that immunoreactivity for STAT3 was present only in neurones, a portion of sections from each brain were then retrieved and stained a second time to demonstrate the antigen that had not been stained in that section previously.

After staining, vibratome sections were either mounted onto slides or dehydrated, infiltrated with JB-4 methacrylate resin, and flat-embedded between two plastic coverslips (Young et al. 1996). After hardening, embedded sections were attached to plastic stubs with cyanoacrylate glue and were cut at a thickness of 1–2 μm using dry glass knives and a JB-4 microtome. Sections were then dried down onto glass slides and counterstained with 0.1% toluidine blue.

Results

As reported previously, astrocytes intensely immunoreactive for bFABP were abundant in the arcuate nucleus of the hypothalamus; astrocytes elsewhere showed a much weaker immunoreactivity (Young et al. 1996) (Fig. 1). Immunoreactivity was present in both cytoplasmic processes and in astrocyte nuclei (Fig. 2).

Fig. 1.

Vibratome sections of the arcuate nucleus of the hypothalamus, stained using immunocytochemistry to illustrate: (A) astrocytes immunoreactive for brain fatty acid binding protein (bFABP) (arrow), (B) leptin-sensitive neurones immunoreactive for Signal Transduction and Activator of Transcription-3 (STAT3) (arrow), (C) control section stained with non-immune rabbit serum. Scale bar = 100 μm.

Fig. 2.

Immunostained vibratome sections, flat embedded in methacrylate and cut at a thickness of 1 μm to provide higher magnification views of arcuate astrocytes and neurones. Panel A shows that bFABP-immunoreactive astrocytes (arrowheads) occupy the same general area as STAT3-immunoreactive neurones (asterisks). Scale bar = 20 μm. Panels B—D show examples of bFABP-immunoreactive astrocytes in close proximity to STAT3-immunoreactive neurones. Scale bar = 20 μm.

Immunoreactivity for STAT3 was detected in neuronal nuclei and perikarya, but not in dendritic processes. Sections stained for both antigens showed that both cell types were common in the same areas of the arcuate nucleus. The staining intensity of bFABP- immunoreactive astrocytes was noticeably darker than that of STAT3-immunoreactive neurones, so that astrocytes were readily distinguished from stained neurones; this was probably due to the greater recommended dilution of the anti-STAT3 antibody compared to that for the antibFABP antibody. bFABP-immunoreactive astrocytes could often be seen closely adjacent to STAT3-immunoreactive neurones (Fig. 2).

Discussion

These results demonstrate that arcuate astrocytes not only contain bFABP, a known modulator of fatty acid synthase, but also often are in close proximity to feeding-regulating neurones immunoreactive for a leptin-activated transcription factor. The degree of contact between these two cell types could not be quantitatively assessed and was likely greatly underestimated in this study, for several reasons. First, antibodies to either STAT3 or to leptin receptors can identifity leptin-sensitive neurones, but do not demonstrate the dendritic processes of these neurones (Hakansson & Meister, 1998). Thus, contact between a leptin-sensitive neurone and a bFABP-immunoreactive astrocyte could be assessed only in the neighbourhood of the neuronal nucleus. Secondly, antibodies only penetrated a few micrometres into the thickness of each vibratome section, limiting the number of neurones and astrocytes that could be examined for each section.

This study provides an anatomical basis for the possibility that specialized arcuate astrocytes influence the function of feeding-regulating neurones. Astrocyte bFABP, by regulating either the extracellular or intracellular concentration of free fatty acids, could affect the activity of fatty acid synthase either in adjacent neurones or in the astrocyte cytoplasm itself, and thereby modulate the hypothalamic control of feeding behaviour (Mukhopadhyay & Mukherjea, 1998; Kusakabe et al. 2000; Loftus et al. 2000). Also, since a high-fat diet induces hypothalamic resistance to satiating effects of leptin and thereby promotes diet-induced obesity, the ability of astrocytes adjacent to leptinresponsive neurones to bind fatty acids is likely to have important functional consequences (El-Haschimi et al. 2000). It has been proposed that the anorexic action of intraventricular infusions of oleic acid may involve an altered function of neurones that possess ATP-sensitive K+ channels (Obici et al. 2002). However, since astrocytes also possess ATP-sensitive K+ channels and show functional responses to oleic acid, an involvement of astrocytes in feeding effects of lipid molecules seems likelyVelasco et al. 2000).

These data do not rule out the possibility of a functional interaction between bFABP + astrocytes and other subsets of arcuate neurones. Such astrocytes could, for example, modulate the ability of free fatty acids to control arcuate neurones containing somatostatin (Briard et al. 1998). Another function postulated for bFABP is to deliver lipids to neurones in brain regions undergoing neurogenesis or remodelling of axonal processes (Kurtz et al. 1994). However, such events appear relatively rare in the adult arcuate nucleusLeal et al. 1998).

A physiological interaction between neurones of two different phenotypes can be assessed anatomically by confirming the presence of synapses between the two cell types. However, a physiological interaction between an astrocyte and a neurone cannot be assessed with confidence by using purely anatomical methods, even at the ultrastructural level, since an anatomical signature of cell–cell interaction, i.e. a synapse, does not occur between glia and neurones. One means of assessing the physiological significance of the anatomical findings reported here would be to examine effects of hypothalamic infusion of bFABP upon feeding behaviour. This is a goal of future studies.