Feeding suppression by fibroblast growth factor-1 is accompanied by selective induction of heat shock protein 27 in hypothalamic astrocytes

Animals

All animals received humane care in accordance with AIST guidelines; all animal protocols were approved by the AIST animal experimentation committee (Tsukuba, Japan). Male Wistar rats (Clea Co., Japan) weighing 250–300 g were maintained in individual cages and experienced a 12 h : 12 h light-dark cycle (lights on at 07.00 h) at a room temperature of 24–25 °C.

Sendai virus vector construction and surgical procedures

Recombinant SeV harbouring the mouse FGF-1 gene (FGF-1/SeV) or the green fluorescent protein (GFP) gene (GFP/SeV) were constructed and propagated, as described previously (Kato et al., 1996). Viral titres were determined using a standard chicken red blood cell haemagglutination assay. To administer the vector, the animals were anaesthetized by intraperitoneal (i.p.) injection of pentobarbital sodium (50 mg/kg), positioned in a stereotaxic frame, and unilaterally injected with 5 × 10⁶ pfu of FGF-1/SeV or GFP/SeV into the right LV using a 33-gauge 10-µL Hamilton syringe. The injection coordinates were −1.0 mm posterior to bregma, 1.5 mm lateral to the midsagittal sinus and 3.5 mm ventral to the skull surface (Paxinos & Watson, 1986).

Profiling mRNA expression

Three days after SeV infusion, total hypothalamic RNA were prepared using Isogen solution (Nippon Gene, Tokyo, Japan) according to the manufacturer's protocol. Gene expression in the GFP/SeV- and FGF-1/SeV-infused hypothalami was then analysed using the Atlas Rat cDNA expression array (Clontech, Palo Alto, CA, USA) with materials provided in the kit according to the manufacturer's instructions. Briefly, 1 µg samples of mRNA purified using a PolyA tract mRNA isolation system IV (Promega, Madison, WI, USA) was converted into ³²P-labelled first-strand cDNA using M-MLV reverse transcriptase. Unincorporated ³²P-labelled nucleotides were removed by Chroma spin-200 (Clontech) column chromatography. The cDNA fractions with the highest activity were then pooled and hybridized to an Atlas membrane. After prehybridization for 30 min at 68 °C in Express Hyb (Clontech) supplemented with 200 µg/mL herring sperm DNA (Sigma, Deisenhofen, Germany), the heat-denatured probe was added. Hybridization was carried out overnight at 68 °C, after which the membranes were washed for 20 min in 2 × SSC containing 1% SDS at 68 °C, followed by two washes in 0.1 × SSC containing 0.5% SDS. The membranes were then exposed to X-ray film for 72 h at −70 °C.

Quantification of heat shock protein (HSP) 27 expression by Northern-blot analysis cRNA probe

Fragments of rat HSP27 were amplified by polymerase chain reaction (PCR) using the following primer set: sense, 5′-GTTAAGACCAAG GAAGGCGTGG-3′; antisense, 5′- CTACTTGGCTCCAGACTGT TCC-3′. The amplified fragments were inserted into PCR-Trap vector (GenHunter, Nashiville, TN, USA) and amplified by PCR using a pair of riboprobe primers (SP6 and T7) fused to the DNA sequences flanking the cloning site of the vector. The resultant product contained the SP6 and T7 promoters, which were then transcribed in vitro using appropriate RNA polymerase (Life Technologies, Rockville, MD, USA) in the presence of digoxygenin-UTP (Roche Diagnostics, Mannheim, Germany) and stored in 50% ethanol. Mouse FGF-1 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) riboprobes were prepared from the constructs based on pBluescript II SK+, which was previously constructed in our laboratory (Ozawa et al., 1996).

Northern hybridization analysis

Samples of total RNA (1 or 5 µg) were electrophoresed in a formaldehyde-agarose gel and blotted onto a nylon membrane (Hybond-N+, Amersham Pharmacia Biotech, Amersham, UK) with 20 × SSC, and then hybridized with digoxygenin-labelled cRNA probes using Easy hyb buffer (Roche Diagnostics) according to the manufacturer's instructions. The membranes were then washed twice with 2 × SSC containing 0.1% SDS at room temperature and twice with 0.1 × SSC containing 0.1% SDS at 68 °C. Bound riboprobe was detected with an antidigoxygenin-AP Fab fragment and CDP-Star (Roche Diagnostics). Thereafter, semiquantitative analysis of HSP27 was performed and corrected using GAPDH (a housekeeping gene normally present in constant amounts within samples) as an internal standard. HSP27/GAPDH mRNA ratios were calculated using image processing software running on a Macintosh computer (NIH Image, Dr Wayne Rasband, NIH, Bethesda, MD, USA).

Histochemistry

Under sodium pentobarbital anaesthesia (50 mg/kg, i.p.), blocks of brain were removed, fixed for 12–16 h at 4 °C in 4% paraformaldehyde in phosphate-buffered saline (PBS), and then treated in a graded sucrose series for 24 h. After embedding the fixed specimens in OCT compound, 16-µm coronal sections were cut, coated onto slides, dried (50 °C for 1 h), and stored at −40 °C until use.

For the purposes of in situ hybridization, slides were first soaked in 0.2 m HCl for 20 min at room temperature to inactivate endogenous alkaline-phosphatases, then treated with proteinase K (5 µg/mL) for 5 min at 37 °C, and dehydrated in an ethanol series. An excess of digoxygenin-labelled FGF-1 riboprobe (in 60 µL of hybridization buffer consisting of 300 mm NaCl, 30 mm sodium citrate at pH 7.0, 50% v/v formamide, 10% w/v dextran sulphate, and 1 µg/µL Escherichia coli tRNA) was applied to each slide, and hybridization was allowed to proceed for 20 h at 50 °C. After then washing the slides (in 300 mm NaCl, 30 mm sodium citrate and 50% formamide at 55 °C for 1 h), bound riboprobe was detected with an antidigoxygenin antibody conjugated with alkaline-phosphatase (1 : 1000, Roche). A colourimetric reaction was carried out with nitro-blue tetrazolium and 5-bromo-4-chloro-3-indolyl-1-phosphate (Promega) for 3–5 h at room temperature.

For immunohistochemistry, sections were first incubated for 30 min in PBS containing 0.3% hydrogen peroxide to eliminate endogenous peroxidase activity. The sections were then rinsed in PBS, incubated for 30 min in PBS containing 2% normal goat serum, and then rinsed for an additional 30 min in PBS containing rabbit polyclonal antibody against HSP27 (1 : 1000) (StressGen, Victoria, BC, Canada) plus 2% normal goat serum. The primary antibody was localized with the aid of an avidin-biotin system (Vector Laboratories, Burlingame, CA, USA). Bound peroxide was visualized by incubating 0.05% 3,3′-diaminobenzidine tetrahydrochloride with 0.005% hydrogen peroxide, resulting in a brown reaction product. For double immunostaining experiments, sections were blocked with 5% BSA in PBS and then incubated for 1 h with polyclonal antibody against HSP27 and monoclonal antibody against glial fibrillary acidic protein (GFAP) or microtubule-associated protein 2 (MAP2) (Neo Markers), followed by coincubation in a mixed solution of 1 : 500 Alex488-conjugated goat antimouse IgG (Molecular Probes, Eugene, OR, USA) and 1 : 200 Texas red-conjugated goat antirabbit IgG (Amersham).

Intracerebroventricular infusion of recombinant polypeptides

Under sodium pentobarbital anaesthesia (50 mg/kg, i.p.), a 23-gauge stainless-steel guide cannula (15 mm long) was inserted into the left LV (−0.6 mm posterior to bregma, 1.4 mm lateral to the midsagittal sinus and 3.8 mm ventral to the skull surface; Paxinos & Watson, 1986) and fixed to the skull with dental cement; 3 days later, cannula placement was functionally confirmed by demonstration of increased thirst after administration of angiotensin II (30 ng/rat, ICV).

Measurement of food intake was begun following a 1-week recovery from the implantation using procedures similar to those described by Li et al. (1998b). The rats were fed ad libitum except between 18.00 and 19.00 h daily, when food was removed, measured and then replaced. Food intake over selected time-spans was measured at 21.00, 01.00, 09.00 and 18.00 h (beginning 4 days before the ICV infusion), which were, respectively, 0–2, 3–6, 7–14 and 15–23 h after infusion. During dark phase measurements, a red incandescent bulb was used to provide dim illumination.

Recombinant human FGF-1 (R&D Systems, Minneapolis, MN, USA) dissolved in 0.15 m NaCl containing 200 µg/mL heparin was infused into each rat (5 µL/rat, ICV) over a 5-min period using a microsyringe. Some animals received 0.15 m NaCl containing 200 µg/mL heparin as a vehicle control. Recombinant rat leptin (1 µg/rat; Genzyme, Cambridge, MA, USA) or rat ciliary neurotrophic factor (CNTF; 250 ng/rat; R&D) in saline was infused, as described earlier. Saline served as vehicle control for leptin and CNTF. Each rat received a single infusion.

Statistics

All results are presented as means ± SEM. The data were analysed using one-way analysis of variance (anova), followed by multiple comparisons between individual groups using a post hoc Fisher's protected least significant difference test (PLSD). For experiments with small animal numbers (n = 3–5 per group), significant differences were determined using Mann–Whitney U-test. Values of P < 0.05 were considered significant.

Gene expression profiling reveals selective induction of HSP27 in the hypothalamus by FGF-1/SeV

To analyse the mechanism by which FGF-1 suppresses feeding, FGF-1/SeV or GFP/SeV was infused into the LV of rats and the changes in hypothalamic gene expression were examined.

Initially determined were the regions transfected by these vectors. Using in situ hybridization with samples harvested on day 3 after the infusion of FGF-1/SeV, we found FGF-1 mRNA to be specifically expressed in the ependymal cell layers surrounding the 3V (Fig. 1A), contralateral LV (Fig. 1A), and ipsilateral LV (not shown). Control rats infused with GFP/SeV exhibited a similar distribution of GFP protein, as shown by its fluorescence (Fig. 1B). FGF-1 and GFP signals were not observed in other regions in the brain sections (coronal section from bregma −1.4 mm to −3.8 mm), including hypothalamus, hippocampus or cerebral cortex.

The strong expression of FGF-1 in ependymal cells of FGF-1/SeV-infused rats was accompanied by marked suppression of feeding and loss of bodyweight, which is consistent with our earlier observation (Hou et al., 2000), and suggests that exogenous FGF-1 expressed in transfected ependymal cells acts as a satiety factor, perhaps utilizing the same mechanism employed by endogenously expressed FGF-1.

To examine the hypothalamic gene expression profiles in more detail, on day 3 after infusion of either FGF-1/SeV or GFP/SeV into the LV, total hypothalamic mRNA was extracted, reverse transcribed, and hybridized to cDNA array membranes. As shown in Fig. 2, the most prominent change was an increase in the expression of HSP27 mRNA in the hypothalami of FGF-1/SeV-infused rats (Fig. 2A, brackets 1). The induced expression of HSP27 mRNA was confirmed by Northern blot analysis, which showed that FGF-1/SeV infusion drastically increased HSP27 expression, as compared with GFP/SeV, and that the level of HSP27 expression paralleled the enhanced expression of FGF-1 (Fig. 2B). Expression of other HSP family members, including HSP60 and HSP70, was not detected on the DNA arrays (Fig. 2A, brackets 2 and 3, respectively). In addition, experiments using RT-PCR showed that expression of HSP60, HSP70 and HSP90 was not affected by the FGF-1/SeV infusion (results not shown). The results indicate that the observed change in HSP27 expression was not a response to the stress of infusion. Expression of the housekeeping enzyme GAPDH was also unaffected by the infusion (Fig. 2A, brackets 4).

HSP27 immunoreactivity was distributed around the 3V in FGF-1/SeV-infused rats

We next analysed the localization of HSP27 protein in coronal brain sections of rats infused with FGF-1/SeV or GFP/SeV. When control rats were infused with GFP/SeV, faint HSP27 immunoreactivity was observed in a thin zone surrounding the 3V (Fig. 3A). Rats infused with FGF-1/SeV, by contrast, exhibited strong HSP27 immunoreactivity, widely distributed in the parenchyma surrounding the ependymal cell layer of the 3V (Fig. 3A). Indeed, HSP27 signals were observed in all tested coronal sections from FGF-1/SeV-infused brains, encompassing a region from bregma −1.4 mm to −3.8 mm. More dispersed and weak HSP27 signals were also observed around the contralateral (Fig. 3A) and ipsilateral (not shown) LVs, but in this case, there was no difference in their intensity among GFP/SeV- and FGF-1/SeV-infused brains. No HSP27 signals were detected from other brain regions.

In some experiments, brain sections were double immunostained for HSP27 and GFAP or MAP2, i.e. astrocyte and neuron markers, respectively. In large part, the HSP27 signals were localized to GFAP-positive cells that also showed hypertrophy and a stellate shape (Fig. 3B). Thus, the expression of HSP27 induced by the ependymal cell-derived FGF-1 appears to occur in astrocytes surrounding the 3V.

ICV infusion of recombinant FGF-1 polypeptide induces transient HSP27 expression

To confirm that FGF-1 in the CSF was responsible for the observed increases in HSP27 expression, we tested the capacity of recombinant FGF-1 polypeptide, infused into the LV, to induce HSP27 expression. Consistent with our previous study (Li et al., 1998b), significant feeding suppression was observed in the FGF-1-infused group (14.1 ± 1.4 g) compared with the saline-infused group (21.8 ± 0.8 g) on the infusion day (Fig. 4A), suggesting that the reduction of feeding was specific to ICV infusion of FGF-1 polypeptide, but not to infusion stress. Northern blot analysis revealed that infusion of FGF-1 elicited a clear, time-dependent increase in hypothalamic HSP27 expression: the expression was increased 6 and 14 h after FGF-1 infusion, and then gradually declined until, by 48 h, it was no longer apparent (Fig. 4B). The level of HSP27 expression 6 h after FGF-1 infusion was dose-dependent, with maximal responses being attained upon infusion of 200 ng of the protein (Fig. 5).

Temporally correlated with FGF-1-induced induction of HSP27 was a decline in feeding behaviour (Fig. 4B and C). When feeding during selected periods after infusion of FGF-1 (200 ng/rat) was compared with that during the same period on the pre-infusion day, feeding behaviour was found to be inhibited 0–2, 3–6, 7–14 and 15–23 h after ICV FGF-1 infusion by 28.5% (n = 14), 67.7% (n = 11), 25.9% (n = 8) and 16.8% (n = 4), respectively, with the most significant decline occurring during the 3–6-h period (P < 0.02, Fig. 4C). Thus, induction of HSP27 expression and suppression of feeding behaviour showed similar temporal patterns.

Immunohistochemical analysis of the hypothalamus showed that HSP27 immunoreactivity in FGF-1-infused brains was distributed around the 3V (Fig. 6A). HSP27-positive regions extended 221 ± 38 µm from the wall of the ventricle in FGF-1-infused rats, which was significantly greater than that seen in rats receiving either saline or heparin (Table 1). Note that HSP27 expression associated with heparin infusion was not significantly different from that seen with saline infusion (P = 0.1; Table 1).

No induction of HSP27 was seen in the area surrounding the contralateral LV in any experimental group (Fig. 6A). Double immunostaining following FGF-1 infusion confirmed the colocalization of HSP27 and GFAP signals (Fig. 6B) seen previously in FGF-1/SeV-infused rats (Fig. 3B). It is also noteworthy that HSP27 signals in both FGF-1 polypeptide- and FGF-1/SeV-infused rats were distributed not only within GFAP-positive cell bodies of astrocytes, but also in areas surrounding the MAP2-positive neurons 3, 6.

ICV infusion of recombinant CNTF, but not leptin, induces HSP27 expression

The results thus far suggest that suppression of feeding behaviour caused by FGF-1 is accompanied by the induction of HSP27 expression in periventricular astrocytes. To determine whether increased expression of HSP27 also accompanies responses to other known suppressers, we examined the effects of CNTF (Fantuzzi et al., 1995) and leptin (Zhang et al., 1994; Halaas et al., 1995). As reported previously, it was found that 24 h after ICV infusion of 200 ng of CNTF, the gain in bodyweight in CNTF-infused rats was significantly less than that in saline-infused, control rats (data not shown). Northern blot analysis of hypothalamus samples obtained 6 h after infusion showed that the level of HSP27 mRNA was augmented by infusion of CNTF (357 ± 104% compared with saline control; n = 3 in each case; P < 0.05; Fig. 7).

As with CNTF, rats infused with 1 µg of recombinant leptin gained significantly less bodyweight than rats infused with saline (data not shown). However, there was no significant increase in expression of HSP27 mRNA (230 ± 44% for leptin, 100 ± 15% for saline, P = 0.5; Fig. 7).

The HSP27-positive region extended 293 ± 13 µm from the wall of 3V in CNTF-infused rats, which was significantly greater than in either saline- or leptin-infused animals (Fig. 6A, Table 1). There was no significant difference between saline- and leptin-infused animals (P = 0.7).