Altered Hippocampal Expression of Neuropeptides in Seizure-prone GALR1 Knockout Mice

Authors


Address correspondence and reprint requests to Dr. S.O. Fetissov at Department of Neuroscience B3:4, Karolinska Institutet, Retzius väg. 8, 17177, Stockholm, Sweden. E-mail: Serguei.Fetissov@neuro.ki.se

Abstract

Summary: Purpose: Mice carrying a deletion of the GALR1 galanin receptor have recently showed spontaneous seizure phenotype with 25% penetrance. To better understand the role of neuropeptides, which are known to undergo complex plasticity changes with development of epileptic seizures, we characterized their expression in the hippocampal formation in GALR1- knockout (-KO) mice with or without seizures and in wild-type (WT) mice.

Methods: Immunohistochemistry and in situ hybridization were used to study expression of galanin, neuropeptide Y (NPY), substance P, enkephalin, dynorphin, and cholecystokinin (CCK).

Results: In GALR1-KO mice that had been displaying seizures, a strong upregulation of galanin immunoreactivity (ir) and messenger RNA (mRNA) was found in the polymorph layer of the dentate gyrus; galanin-ir also appeared in a dense fiber network in the supragranular layer. A strong upregulation of enkephalin was found in the granule cells/mossy fibers, whereas dynorphin mRNA levels were modestly decreased. NPY was strongly expressed in the granule cells/mossy fibers, and an increase of NPY mRNA levels in the polymorph cells was paralleled by an increase of NPY-ir in the molecular layer. An upregulation of substance P-ir was confined to the fibers in the granule and molecular layers, whereas substance P mRNA was increased in the cells of the polymorph layer. Both CCK-ir and mRNA were strongly downregulated in the granule cell/mossy fiber system, but CCK-ir appeared increased in the supragranular and molecular layers. No changes in neuropeptide-ir were found in GALR1-KO mice not displaying seizures.

Conclusions: Complex changes in neuropeptide expression in some principal hippocampal neurons and interneurons appear as a characteristic feature of the spontaneous-seizure phenotype in GALR1-KO mice. However, to what extent causal relations exist between this “epilepsia peptidergic profile” and development of seizures requires further clarification.

Galanin, a 29-aa neuropeptide (1), is widely distributed throughout the rat (2) and mouse (3) brain. Three receptor subtypes for galanin (GALR1, GALR2, and GALR3) have been cloned (4,5). Evidence has been found for involvement of galanin and its receptors in multiple neuronal and neuroendocrine functions (6,7). Galanin has been reported to have inhibitory properties in several pathways in the central nervous system (8–13), including exerting an antiepileptic effect. Thus galanin injected in the hippocampal formation has a powerful anticonvulsant effect in rats (14,15), whereas mice ectopically overexpressing galanin by using different gene promoters display resistance for status epilepticus (SE) and kindling epileptogenesis (16,17). In contrast, mice lacking galanin expression show increased propensity to develop SE (16). Recently mice carrying a deletion of GALR1 have been generated and phenotypically characterized, and both male and female GALR1 knockout (-KO) mice exhibited spontaneous seizures with a penetrance of 25%, ranging from facial movements to SE (18).

It has been demonstrated in a number of studies that expression of neuropeptides is dramatically changed, mostly increased, in animal models for epilepsy and that this occurs in specific subregions, especially in the hippocampal formation (19–22). For example, enkephalin is strongly upregulated in the granule cell/mossy fiber system (23), and neuropeptide Y (NPY) is apparently aberrantly expressed in the same cells (24,25), whereas cholecystokinin (CCK) is downregulated in the mossy fibers (23). Galanin is upregulated in mossy cells and in the supragranular layer (15,17).

In this work, to better understand the role of peptidergic plasticity in the development of the seizure-prone phenotype, we used immunohistochemistry and in situ hybridization to characterize expression of hippocampal neuropeptides in GALR1-KO mice, to see if any of the neuropeptide changes described in various epileptic animal models (see earlier) also occur in the GALR1-KO mouse.

METHODS

Animals and tissue preparation

GALR1-KO mice were generated by insertional mutagenesis of the gene encoding GALR1, resulting in absence of normal full-length transcript (18). Spontaneous seizure phenotype occurred with a penetrance of 25% in mice homozygous for mutation on C57BL/6J background, starting at age 11.1 ± 0.2 weeks, with 27% mortality (18). Some GALR1-KO mice (75%) were found not to develop any seizures when exposed to the standard conditions that consistently evoked seizures in GALR1-KO seizure-prone mice, such as bright overhead lighting or handling. C57BL/6J wild-type (WT) mice were used as controls. Before killing, 6-month-old mice had been housed under controlled environmental conditions with constant light–dark cycle (light on between 6:00 and 18:00 hr), a temperature of 21–22°C, and a relative humidity of 40–50%; food and water were given ad libitum. The day before killing, between 16:00 and 17:00, seizure-prone phenotype in GALR1-KO mice was confirmed by bright overhead lighting, resulting mainly in clonic unilateral or bilateral seizures lasting for ∼1 min. The next day between 9:00 and 10:00, GALR1-KO mice (n = 6), which had exhibited seizures in their anamnesis but were not in SE and were not observed to have a seizure in the morning before killing (referred to later as GALR1-KO unless specified), GALR1-KO mice that did not display seizures (n = 3), and WT mice (n = 6) were anaesthetized with sodium pentobarbital (0.15 mg/100 g body weight, i.p.) and perfused via the ascending aorta with Tyrode's Ca2+-free solution at 37°C, followed by a mixture of 4% paraformaldehyde and 0.4% picric acid in 0.16 M phosphate buffer (pH 6.9), 37°C and then by the same, but ice-cold mixture. The brains were rapidly dissected out, immersed in the same fixative for 90 min, and rinsed with 10% sucrose in 0.1 M phosphate buffer (pH 7.4) overnight. Brains were snap-frozen with CO2. Coronal, 14-μm thick brain sections were cut on a cryostat (Microm, Heidelberg, Germany), and thaw-mounted on chrome alum-gelatin–coated glass slides.

Immunohistochemistry

The tyramide signal amplification (TSA) immunohistochemical method (26) and rabbit polyclonal antisera raised against the following peptides were used: galanin (27), NPY (Walsh J. and H. Wong, unpublished), substance P (28), enkephalin (29), dynorphin (30), and CCK (31). Guinea pig anti–γ-aminobutyric acid (GABA) antiserum (1:400; Chemicon, Temecula, CA, U.S.A.) and CCK (1:400) antiserum (31) were used for double staining after processing for the TSA method. Rabbit antiserum against zinc transporter-3 (ZnT-3), 1:400, was used as a marker for mossy fibers (32). Incubation with primary antisera (1:4,000) overnight at 4°C was followed by horseradish peroxidase–conjugated, swine anti-rabbit immunoglobulin G (IgG; 1:100; Dako A/S, Copenhagen, Denmark) and incubations according to the TSA-Plus Fluorescein System protocol (DuPont, New England Nuclear, Boston, MA, U.S.A.). The specificity of antibodies was tested by preabsorption tests with an excess (10−6 or 10−5M) of the corresponding peptide, purchased from Peninsula (Belmont, CA, U.S.A.) or Bachem (Bissendorf, Switzerland). Sections were mounted in a mixture of glycerol and 0.1 M phosphate-buffered saline (3:1), pH 7.4, containing 0.1% 1,4-phenylenediamine (Sigma Chemicals, St. Louis, MO, U.S.A.) as antifading agent.

After processing, the sections were examined in a Radiance Plus confocal laser scanning system (Bio-Rad, Hemel Hemstead, U.K.), installed on an Eclipse E600 fluorescence microscope (Nikon, Tokyo, Japan). Digital images resulting from the confocal scanning microscopy were optimized for image resolution or merged for color illustrations in PhotoShop 6.0 (Adobe Systems Incorporated, San Jose, CA, U.S.A.). A subjective estimation of intensity of immunostaining was made in the microscope by using a rating scale from not detectable (ND), very low (±), low (+), medium (++), high (+++), and very high (++++) levels (Table 1).

Table 1. Intensity of neuropeptide immunoreactivity in the hippocampal formation of seizure-prone GALR1-KO and WT mice
Hippocampal formationGalaninNPYSPEnkephalinDynorphinCCK
KOWTKOWTKOWTKOWTKOWTKOWT
Dentate gyrus 
 Granule cell layerNDND+ND++++++/−+/−+/−+/−+/−
 Mossy fibersNDND+++NDNDND+++++++++/−++
 Polymorph cells+++/−++++/−+/−NDNDNDND+++
 Supragranular layer+++++/−+ND+NDNDNDNDND+++++++
Hippocampus 
 Molecular layer+/−+/−+++/−+++/−NDNDNDND++/−
 Stratum lacunosum mol.++++++++/−ND++NDND+++/−
 Stratum radiatum+/−+/−+++/−+/−+/−+/−+/−+/−+/−++++++
 Stratum oriens+/−+/−++/−++/−+++/−+/−++++++
Pyramidal cell layers 
 CA1 (cells)NDND+/−+/−+/−+/−+/−+ND+/−+++
 CA1 (fibers)+/−++/−+/−+++/−+/−+ND+/−+++++++
 CA3 (cells)NDND+/−ND+/−NDNDNDNDND++/−
 CA3 (fibers)+/−++/−+/−+ND++/−++/−+++++

In situ hybridization

Between 9:00 and 10:00, GALR1-KO mice (n = 5) that had exhibited seizures in their anamnesis according to conditions described earlier, and WT mice (n = 5) were killed by decapitation. The brains were dissected, immersed in ice-cold phosphate-buffered saline, immediately thereafter frozen, cut at 14-μm thickness with a cryostat (Microm), and thaw-mounted onto “Probe On” slides (Fisher Scientific, Pittsburgh, PA, U.S.A.).

Antisense oligonucleotides probes complementary to nucleotides 152–199 of galanin mRNA (33), 322–360 of enkephalin mRNA (34), 1,671–1,714 of NPY mRNA (35), 298–341 of CCK mRNA (36), 145–192 of preprotachykin mRNA (37), and 871–918 of dynorphin mRNA (38) were synthesized by CyberGene AB (Huddinge, Sweden). The oligonucleotides were labeled at the 3′ end by using terminal deoxynucleotidyltransferase (Amersham, Buckinghamshire, U.K.) with [α-35S]dATP (NEN, Boston, MA, U.S.A.) to a specific activity of 1–4 × 106 cpm/ng oligonucleotide. The labeled probes were purified through QIA quick-spin columns (Qiagen/GmbH, Hilden, Germany). Dithiothreitol (DTT) was added to a final concentration of 10 mM. Sections were hybridized as described previously (39,40). In brief, air-dried sections were incubated in a hybridization buffer [50% formamide, 4× SSC, 1 × Denhardt's solution (1% sarcosyl, 0.02 M phosphate buffer, 10% dextran sulfate), 500 μg/ml heat-denatured salmon sperm DNA, 200 mM DTT, 1 × 107 cpm/ml of the labeled probes] in a humidified chamber for 16–18 h at 42°C. After hybridization, the sections were washed in 1× SSC 4 ×15 min at 55°C and 30 min at room temperature, and then air-dried and dipped into Ilford K5 nuclear emulsion (Ilford, Mobberly, U.K.) diluted 1:1 with water. After exposure at 4°C for 2–6 days, the slides were developed in D19 (Kodak, Rochester, NY, U.S.A.), fixed in Kodak 3000, and mounted in glycerol-phosphate buffer. For the control of specificity of in situ hybridization results, adjacent sections were incubated with an excess (×100) of unlabelled probe, none of which yielded a hybridization signal.

Every seventh section from a series through the rostrocaudal extension of the hippocampus, giving in total 10 sections for each mRNA, was examined by using a Nikon Eclipse E600 fluorescence microscope equipped with a dark-field condenser. Digital images acquired with Nikon DXM1200 digital still camera (using a ×20 objective) were analyzed for intensity of mRNA labeling in Scion Image 4.0 (Frederick, MD, U.S.A.). Each captured image was calibrated to 256-pixel grey values (0 = white and 256 = black), and every labeled neuronal profile in the polymorph cell layer was individually selected, and its mean pixel density was measured. A rectangular selection over the granule layer upper blade with 200-µm long and 100-µm short side was used for the quantification of the single in the granule cells. The background levels were measured in separate images from outside of the section area. We assumed that silver grain density overlying neurons correlates directly with their level of mRNA expression. To estimate silver grain density, the mean pixel density was converted into the relative optic density (ROD) by using formula–log [(256 grey value)/256]. The background ROD level was calculated in the same way and was ultimately subtracted from the neuronal ROD. Unpaired t test was used to compare means of ROD between GALR1-KO and control groups of mice. Data expressed as a percentage of mean ROD relative to the control group equaled 100%± SD.

RESULTS

The expression of galanin, enkephalin, NPY, CCK, substance P, and dynorphin peptides and mRNA was studied at different rostrocaudal levels of the hippocampal formation of seizure-prone GALR1-KO mice and compared with the patterns seen in WT C57BL/6J mice, by using immunohistochemistry and in situ hybridization. In terms of general anatomy and structure of the hippocampal formation, no obvious differences were found between the two groups of mice. However, it appeared that, in the dorsal dentate gyrus, the upper and lower blades of granule cell layers were more separated in the GALR1-KO than in WT mice, so that in the former, the angle between the two blades was significantly larger (27 ± 3.2 vs. 15.5 ± 1.8 degrees; p < 0.001). In the majority of cases, the changes in mRNA and peptide levels were parallel, but the changes in transcripts appeared less pronounced than the effects on peptide levels. The results of neuropeptide immunostaining in the dorsal hippocampal formation are summarized in Table 1 and shown in Figs. 1–6, and changes in neuropeptide mRNA levels are demonstrated in Fig. 7. Figure 8 shows results from incubation with antiserum to ZnT3, a marker for mossy fibers.

Figure 1.

Galanin (Gal) and cholecystokinin (CCK) immunoreactivity (-ir) in the dorsal hippocampal formation of wild-type (WT) (A) and seizure-prone GALR1-KO (B) mice after processing for double labeling. Note a strong yellow band reflecting overlapping galanin-ir and CCK-ir in the supragranular layer in the KO (B) but not WT (A) mouse. Coexistence (arrowheads) of galanin-ir and CCK-ir was found in mossy cells of the polymorph layer in GALR1-KO mice (C–E). Arrow, a galanin-positive, CCK-negative fiber. Scale bars: 100 μm (A, B); 20 μm (C–E). CA1, CA3, Ammon's horn, fields CA1, CA3; gr, granule cell layer; mf, mossy fibers; mo, molecular layer; po, polymorph layer; sg, supragranular layer; slm, stratum lacunosum moleculare; sr, stratum radiatum; so, stratum oriens.

Figure 2.

Galanin-ir in the dorsal hippocampal formation of wild type (WT) (A, C, E, G), seizure-prone GALR1-KO (B, D, F, H) mice and GALR1-KO mice without seizures (nsz) (I). Note a strong increase of galanin-ir in the supragranular layer (sg) (B, D) and in cells in the polymorph layer (po) in seizure-prone GALR1-KOs (D). Cells indicated by arrows in C and D correspond to the cells in G and H, respectively. Scale bar: 100 μm (A, B); 50 μm (C, D); 5.5 μm (E–H); 20 μm (I). CA1, CA3, Ammon's horn, fields CA1, CA3; gr, granule cell layer; mf, mossy fibers; mo, molecular layer; po, polymorph layer; sg, supragranular layer; slm, stratum lacunosum moleculare; sr, stratum radiatum; so, stratum oriens.

Figure 3.

Enkephalin-ir in the dorsal hippocampal formation of wild-type (WT) (A, C, E), seizure-prone GALR1-KO (B, D, F) mice and GALR1-KO mice without seizures (nsz) (G). Note a strong increase of enkephalin-ir in the granule cells (gr) and mossy fibers (mf) in seizure-prone GALR1-KOs (B, D, F). The enkephalin-positive cells are mainly localized in the external part of the granule cell layer (gr) (E, F). Scale bar: 100 μm (A, B); 50 μm (C, D); 5,5 μm (E–F); 20 μm (G). CA3, Ammon's horn, field CA3; gr, granule cell layer; mf, mossy fibers; po, polymorph layer.

Figure 4.

Neuropeptide Y immunoreactivity (NPY-ir) in the dorsal hippocampal formation of wild-type (WT) (A, C, E), seizure-prone GALR1-KO (B, D, F) mice and GALR1-KO mice without seizures (nsz) (G). Note that NPY-ir is present in the mossy fibers in seizure-prone GALR1-KO (D) mice. Increases occur but are less pronounced in other layers, including more fibers in the granule cell layer (gr) (E, F). Scale bar: 100 μm (A, B); 50 μm (C, D); 5.5 μm (E-F), 20 μm (G). CA3, Ammon's horn, field CA3; gr, granule cell layer; mf, mossy fibers; mo, molecular layer; po, polymorph layer.

Figure 5.

Cholecystokinin immunoreactivity (CCK-ir) in the dorsal hippocampal formation of wild-type (WT) (A, C, E, G), seizure-prone GALR1-KO (B, D, F, H) mice and GALR1-KO mice without seizures (nsz) (I). Note a decrease of CCK-ir in the mossy fibers (mf) (Dvs. C), but an increase in the stratum lacunosum moleculare (slm) (B vs. A). There is only a small increase in the supragranular layer (sg) (F vs. E), whereas CCK-positive cells in the polymorph layer (po) can be more distinctly seen in the seizure-prone GALR1-KO mouse (H vs. G). Scale bar: 100 μm (A, B); 50 μm (C, D); 5.5 μm (E–H); 20 μm (I). CA3, Ammon's horn, field CA3; gr, granule cell layer; mf, mossy fibers; mo, molecular layer; po, polymorph layer; sg, supragranular layer; slm, stratum lacunosum moleculare.

Figure 6.

Substance P (SP) (A–E) and dynorphin (Dyn)-ir (F–H) in the dorsal hippocampal formation of wild-type (WT) (A, C, F), seizure-prone GALR1-KO (B, D, G) mice and GALR1-KO mice without seizures (nsz) (E, H). Note an increase of substance P-ir in the molecular layer (mo) in seizure-prone GALR1-KO mice (B). No major difference can be seen with regard to dynorphin-ir (E, F). Scale bar: 100 μm (A, B, E, F); 5.5 μm (C, D); 20 μm (G). gr, granule cell layer; mf, mossy fibers; mo, molecular layer; po, polymorph layer.

Figure 7.

A: Levels of messenger RNA (mRNA) for galanin (Gal), enkephalin (Enk), neuropeptide Y (NPY), cholecystokinin (CCK), substance P (SP), and dynorphin (Dyn) in seizure-prone GALR1-KO versus wild-type (WT) mice (100%) in granule (gr) or polymorph cells (poly) of the dentate gyrus. Examples of in situ hybridization signals are shown for galanin (B, C), enkephalin (D, E), and substance P (F, G) in WT (B, D, F) and GALR1-KO (C, E, G) mice. Note differences in signal for galanin in the polymorph layer (po) (B, C), for Enk in the granule cell layer (gr), CA1 and entorhinal cortex (ent) (D, E), and for SP in the polymorph layer (F, G). Arrows, cells with positive hybridization signals; ec, external capsule; DG, dentate gyrus. Scale bars: 100 μm (B, C, F, G); 500 μm (D, E). CA1, Ammon's horn, field CA1.

Figure 8.

ZnT3-immunoreactivity in the dentate gyrus and CA3 of wild-type (WT) (A–C), seizure-prone GALR1-KO mice (D–F) and GALR1-KO mice without seizures (nsz) (G–I). Sprouting of mossy fibers (mf) is seen between the pyramidal cells (pyr) in CA3 in seizure-prone GALR1-KO mice (arrowheads in E, Φ). Σχαλɛβαρσ: 20 μm (A, B, D, E, G, H); 10 μm (C, F, I). CA1, CA3, Ammon's horn, fields CA1, CA3; gr, granule cell layer; mf, mossy fibers; mo, molecular layer; po, polymorph layer; sg, supragranular layer; slm, stratum lacunosum moleculare; sr, stratum radiatum; so, stratum oriens.

In GALR1-KO mice that did not display seizures, immunohistochemical detection of galanin (Fig. 2I), enkephalin (Fig. 3I), NPY (Fig. 4G), CCK (Fig. 5I), substance P (Fig. 6E), and dynorphin (Fig. 6H) revealed normal distribution of these neuropeptides in the hippocampal formation without apparent differences in the intensity of staining when compared with WT mice.

Galanin

A moderately dense galanin-positive fiber network was fairly evenly distributed in the hippocampal formation of WT mice (Figs. 1A and 2A). A marked increase was seen in intensity of the supragranular layer in GALR1-KO mice, whereas a general decrease was noted of galanin-positive fibers in the hippocampal formation, including the granule cell layer (cf. Figs. 1B; 2B, D, F with Figs. 1A and 2A, C, E). Numerous large neurons with intense galanin immunoreactivity (-ir) were present, especially in the caudal parts of the polymorph layer of the GALR1-KO mice (Fig. 2D), whereas fewer such neurons appeared in the WT mice (Fig. 2C). These galanin-ir cells were morphologically similar in both groups of mice (Fig. 2G and H). Moreover, double-labeling experiments showed that the galanin cells also are CCK positive (Fig. 1C–E), and that the intense bands of galanin- and CCK-ir strongly overlap (Fig. 1A and B). In the confocal microscope, coexistence of galanin- and CCK-ir also could be observed in nerve endings in the supragranular layer. Additionally, galanin-positive polymorph neurons did not display GABA-ir. In situ hybridization showed a strong increase of the galanin mRNA signal in cells of the polymorph layer in GALR1-KO mice (Fig. 7A; cf. Fig. 7B and C).

Enkephalin

A prominent enkephalin expression was present in the mossy fiber system in both WT and GALR1-KO mice (Fig. 3A and B), but in the KO mice, enkephalin-ir was markedly increased in the hilar region and in stratum lucidum, reflecting its high content in mossy fibers (cf. 3B and D with 3A and C). Because of the strong and diffuse enkephalin-ir in the hilar region, it was not clear if any of the polymorph cells were enkephalin positive. However, we could not detect a distinct enkephalin mRNA signal in the polymorph layer. In contrast, the granule cells, particularly in the upper blade, displayed an increased enkephalin-ir, these cells being predominantly located in the external part of the granule cell layer (Fig. 3F). In accordance with the peptide change, enkephalin mRNA expression was significantly increased in granule cells in GALR1-KO mice, being most prominent in the ventral part of the dentate gyrus (Fig. 7A; cf. Fig. 7D with 7E). Increased transcript levels were also noted in the CA1 region and in the entorhinal cortex (cf. 7D with 7E).

Neuropeptide Y

NPY-positive fibers and interneurons were evenly distributed in the WT mouse hippocampal formation (Fig. 4A and C). In contrast, in GALR1-KO mice, NPY-ir was increased in several hippocampal/dentate regions, such as the hilus and mossy fibers in stratum lucidum, in the external part of the molecular layer, and in strata lacunosum moleculare, radiatum, and oriens (Fig. 4B and D). An increase in the number of NPY-positive fibers was seen in the granule cell layer (cf. Fig. 4F with 4E). These changes in NPY-ir were corroborated by significantly elevated levels of NPY mRNA in both the granule and polymorph cells (Fig. 7A).

Cholecystokinin

The WT mice displayed a wide expression of CCK-ir in the hippocampal formation (Figs. 1A and 5A), whereas GALR1-KO mice showed a mostly similar but quantitatively different distribution, with an increase in the molecular layer and in the stratum lacunosum moleculare (Figs. 2B and 5B). The most conspicuous change, however, was a depletion of CCK-ir in the mossy fiber system, as could be most clearly seen in the stratum lucidum (cf. Fig. 5D with 5C). The decrease in the density of mossy fiber staining in the hilar region of GALR1-KO mice allowed easy visualization of CCK-positive cells in the polymorph layer (cf. Fig. 5H with 5G; see also Fig. 1C–E). These cells were also present in the WT mice but more weakly fluorescent and, again, masked by dense CCK-positive mossy fiber network and could be visualized by adjusting the conditions of laser scanning in confocal microscope. Also the supragranular layer in GALR1-KO mice appeared to have a somewhat increased CCK-ir (cf. Fig. 5F with 5E); CCK mRNA levels were not changed significantly in the polymorph cell layer, but it was decreased in the granule cell layer (Fig. 7A).

Substance P

Dot-like substance P-ir was detected in the granule cell layer of both WT (Fig. 6A and C) and GALR1-KO (Fig. 6B and D) mice, presumably in fibers surrounding granule cells, because no substance P–positive mossy fibers were found in either type of mouse (Fig. 6A and B). The intensity of this immunostaining was, however, stronger in GALR1-KO mice (cf. Fig. 6D with C), and in this mouse, a substance P–positive terminal network also was found in the molecular layer (Fig. 6B), whereas it was hardly detected in the WT mice (Fig. 6A). In accordance with the peptide change, the substance P (preprotachykinin) mRNA level was increased in cells in the polymorph layer, but was not detected in the granule cell layer (cf. Fig. 7G with F).

Dynorphin

In both WT and GALR1-KO mice, the only prominent dynorphin-ir in the hippocampal formation was detected in the mossy fiber system (Fig. 6E and F). Even though the intensity of dynorphin immunostaining appeared unchanged in GALR1-KO mice, in situ hybridization revealed a decrease of the dynorphin mRNA signal in the granule cells in KO mice (Fig. 7A).

ZnT3

Immunohistochemical analysis detected ZnT3 in the mossy fibers as previously reported (32). No apparent difference in ZnT3-ir granules density in the granule cell layer was seen between WT and GALR1-KO mice, suggesting none or a very low level of collateral sprouting of mossy fibers (Fig. 8D). However, in the CA3 area, more numerous and continuous ZnT3-ir dots were found in seizure-prone GALR1-KO mice (Fig. 8E,F), but not in WT or GALR1-KO mice without seizures, indicating the presence of terminal sprouting of mossy fibers in seizure-prone GALR1-KO mice.

DISCUSSION

It recently was reported that deletion of a single neuropeptide receptor, the GALR1, a receptor for a peptide normally present at comparatively low levels in the hippocampal formation, can trigger complex and severe pathologic seizure behavior (18). The present results demonstrate that this absence of the normal, full-length transcript encoding GALR1 in mice with spontaneous seizures is paralleled by major changes in distribution and expression levels of several neuropeptides in the hippocampal formation. These changes are very similar but not fully overlapping with what has been described in several epilepsy models (19–21) and could perhaps be called an epilepsia peptidergic profile (EPP). Some differences between the models could probably arise from the fact that the GALR1-KO mice analyzed in the present experiment exhibit milder stages of epilepsy than do established models of SE.

Overall, four different types of changes were found in hippocampal neuropeptide expression in seizure-prone GALR1-KOs distinctly different from the pattern seen in WT or GALR1-KOs mice without seizures: (a) strong upregulation of a peptide normally expressed at very low/low/moderate levels, such as galanin-ir in mossy cells and the supragranular cell layer and of enkephalin-ir in the granule cells/mossy fibers; (b) expression of a peptide in cells that normally do not appear to display detectable levels, such as NPY-ir in the granule cells/mossy fibers; (c) a strong downregulation of a peptide, such as CCK-ir in the mossy fibers; and (d) a modest change in the level of peptide expression, such as an upregulation of CCK-ir in the stratum lacunosum moleculare and of substance P-ir in the fibers in the granule cell and molecular layers.

Because we did not observe any abnormality in neuropeptide distribution in GALR1-KO mice without seizures, it is clear that EPP is not a direct consequence of lack of transmission via GALR1 on expression of various peptides, but is secondary to GALR1 deletion–induced seizures, which in turn activate the various peptide systems. The fact that the most dramatic changes occurred in the granule cells, which do not seem to express GALR1 (41), supports this hypothesis. It is likely that glutamatergic transmission may play a crucial role, glutamate being the main excitatory transmitter in the principal hippocampal neurons and thought to be critical for development of SE (42). Thus it is conceivable that GALR1 receptors are part of a natural “braking system,” because galanin has been shown to inhibit glutamate release (10). Why this braking system appears critical only in one fourth of GALR1-KO mice to produce the spontaneous-seizure phenotype is difficult to answer, but these data, which strongly associate EPP and seizures, suggest that specific changes in neuropeptide expression in the hippocampal neurons not only may be result of seizures but also could predispose neurons for development of seizures, as discussed later. Peptidergic plasticity in GALR1-KO mice seemed also to be accompanied, albeit in a very limited fashion, by terminal sprouting of mossy fibers, which is characteristic for kindling and may contribute to seizure formation (43).

Unfortunately, the exact distribution of the galanin receptors in the mouse (and rat) brain is incompletely understood, because, in addition to binding studies (44,45), only the transcripts have been localized with in situ hybridization, even in the rat (41). Nevertheless, both binding and in situ hybridization studies have provided evidence for a high density of galanin-binding sites and GALR1 mRNA in ventral cortices and ventral hippocampus including entorhinal cortex, whereas in the dorsal hippocampal formation, no galanin (1–29) binding but only GALR2 mRNA has been found in granule cells. It is therefore possible that the initial triggering of seizures emanates from the entorhinal cortex and the perforant pathway(s) to the granule cells in the dentate gyrus. If this circuitry is “released” because of lack of inhibitory GALR1 receptors, seizure activity can spread into hippocampus, resulting in the peptide remodeling described here. A further potential source for GALR1 is represented by the noradrenergic locus coeruleus neurons, which are known to express both galanin (46) and GALR1 (and GALR2) mRNA (41) and which project to cortical and hippocampal regions.

In GALR1-KO mice, we observed a strong upregulation of galanin expression only in neurons in the polymorph layer. They are presumably mossy cells sending fibers to the supragranular layer, because they, as shown here, also express CCK, but not GABA, and because Fredens et al. (47) have identified the polymorph layer CCK neurons as mossy cells. These mossy cells are glutamatergic and should normally have a stimulatory effect on granule cell dendrites in the supragranular layer. Galanin released from mossy cell terminals may therefore act on granule cells via GALR2 to modulate glutamate release. However, the significance of an increased galanin synthesis in the mossy cells for development of seizures is not clear and can be ambivalent. Galanin-overexpressing mice aberrantly produce galanin in the granule cells/mossy fibers, which possibly contribute to increased seizure threshold (17). They do not express detectable galanin in the mossy cells/supragranular layer, but this expression appears after hippocampal kindling. Appearance of galanin-positive cells in the polymorph layer were also reported in a rat model of SE (15). Moreover, in contrast to inhibitory GALR1, GALR2 can be an excitatory receptor acting via the IP3 pathway to elevate calcium concentration (4,5). A presynaptic excitatory role of GALR2 has been suggested to modulate nociceptive transmission in the spinal cord (48), and it is possible that in the granule cells, GALR2 has a similar presynaptic effect. However, activation of GALR2 also can decrease intracellular cyclic adenosine monophosphate (cAMP) (4,5). Thus if galanin has a generally antiepileptic effect in the hippocampal formation, so far no evidence exists that galanin in the mossy cells is preventing seizure development, and its true role in these cells requires further clarification.

Another component of EPP formation in GALR1-KO mice is peptidergic plasticity of the principal dentate neurons, the granule cells, such as an appearance of strong NPY expression as reflected by an increased NPY transcript in the granule cells and peptide in the mossy fibers. A transient increase of NPY expression in the granule cells was found in kindling models (20,49,50), whereas it was long lasting after electrically or kainic acid (KA)-induced SE in rats (51,52) and mice (53). In GALR1-KO mice, NPY also was upregulated in the hilar neurons and possibly in their projections in the outer part of the molecular layer, a finding that also was shown in kindled rats (20,50,51). Thus the pattern of NPY expression found in GALR1-KO is rather similar to one that follows the kindling than to changes associated with SE, where an increased NPY-ir in the inner part of the molecular layer was observed (51). Because both granule cells and CA3 pyramidal cells express the Y2 receptor, NPY released from the mossy fibers can inhibit glutamate release acting both presynaptically (54) and postsynaptically (55). Considering the important antiseizure effect of NPY (53,56), our finding of increased NPY expression in the granule cell/mossy fibers of GALR1-KOs suggests that this NPY plasticity rather represents an adaptive mechanism against seizures. Regarding the functional significance of an increased NPY expression in hilar neurons, it is important to discriminate between the increased expression of NPY in the hilus after an acute seizure event, which may contribute to the protective antiseizure mechanisms and the ectopic expression, such as in rats with spontaneous seizures (57), which probably contribute to epileptic state. NPY hilar neurons may modulate activity of the granule cells via the Y1 receptor, which was shown to mediate a proepileptic effect, when tested with a specific Y1-receptor antagonist (58). Additionally, anti-NPY antibodies were shown to reduce seizure susceptibility to metrazol in KA-treated rats (59). Thus the functional significance of an increased NPY expression by hilar interneurons in GALR1-KO mice requires further clarification.

An opioid peptide component of the EPP in GALR1-KO mice generally fits what was found in other epilepsy models (23,60–63), showing changes in opposite directions (that is, a clear increase of enkephalin expression and a modest decrease of dynorphin in the granule cells). Dynorphin has been shown to cause presynaptic inhibition of neighboring mossy fibers (64,65) and is considered an anticonvulsant peptide. The interpretation of the changes in enkephalin expression is more complex, because inhibition of glutamate release from mossy fibers via the δ receptor was shown (66); but enkephalin also can inhibit GABA interneurons (67), and a δ agonist was found to facilitate SE initiation (21). Thus changes of opioid peptide expression in granule cells of GALR1-KO mice suggest a rather proconvulsive effect.

In GALR1-KO mice, increased expression of substance P mRNA in the hilar neurons cells and substance P-ir in the molecular layer suggest that these changes in substance P interneurons may have effect on activity of granule cells. We have, however, not seen a detectable increase in substance P content in the principal dentate neurons, as has been reported for other epilepsy models (68,69), and this difference may be related to less severe seizures in our GALR1-KO mice. Considering that substance P enhances glutamate release from hippocampal slices and has proepileptogenic properties (69,70), an upregulation of substance P in the GALR1-KO mice may contribute to seizure activity.

Both CCK mRNA and peptide levels were decreased in the granule cells/mossy fibers in GALR1-KO mice. These changes are similar to those seen after hippocampal lesion–induced seizures (71) and those found in the mouse model of megalencephaly, where the mouse displays shakiness in gait and seizure-like events (72). In kindled rats, another epilepsy model, an increase in CCK immunore-activity fibers in the supragranular layer was found (20); in GALR1-KO mice, we also observed some tendency for increased CCK-ir in the supragranular layer as well as in the polymorph layer, although changes in CCK mRNA in the hilar neurons were not significant. CCK was shown to block the excitatory effect of KA in mossy fiber CA3 transmission (73), and has, in some studies, been shown to act as an anticonvulsive peptide (74,75). However, CCK is generally considered a stimulatory messenger, and its excitatory effects on granule and CA1 pyramidal cells were reported (76,77). Thus CCK may have both anti- and proconvulsive effects.

It is difficult to judge the significance of the studied neuropeptides for seizure development in GALR1-KO mice, because our results suggest both pro- and antiepileptic effects. Nevertheless, if the EPP can be the key setting in the hippocampal circuitries to form a spontaneous-seizure phenotype, it is not clear why it appears only in 25% of GALR1-KO mice. It was suggested by Wasterlain et al. (21) that maladaptive neuropeptide changes in epilepsy models may represent a liability of the brain to enhance excitation and decrease inhibition in response to repetitive firing of neurons. This hypothesis also could probably be applied to GALR1-KO mice, among which some animals may respond to an enhanced glutamate release by building circuitries that will more readily respond to the excitatory neurotransmission.

In conclusion, an intact galaninergic system, including the galanin peptide itself and the GALR1, appears important for maintaining a normal level of neuronal excitability. The spontaneous-seizure phenotype observed in 25% of GALR1-KO mice was found to be accompanied by complex changes in neuropeptide expression in some principal hippocampal neurons and interneurons. These changes, which we call EPP, were dramatic for galanin in mossy cells (up), for enkephalin and NPY in granule cells/mossy fibers (up), and for CCK in granule cells/mossy fibers (down) and also occur in most other epilepsy models. However, a more complete understanding of the role of neuropeptides in seizures and seizure models will require detailed knowledge of the localization of, in particular, the peptide receptors and their transduction mechanisms.

Acknowledgments

Acknowledgment:  This study was supported by the Swedish MRC (04X-2887; 03X-10350), Marianne and Marcus Wallenberg's Foundation, Knut and Alice Wallenberg's Foundation, an Unrestricted Bristol-Myers Squibb Neuroscience Grant, and the European Union (QLK3-CT-2000-00237). S.F. was supported by the Wenner-Gren Foundations. For the generous supply of antisera, we thank Drs. I. Christensson-Nylander, Uppsala University, Uppsala, Sweden (substance P); P. Frey, Novartis Institute of Biomedical Research, Basel, Switzerland (CCK); Helen Wong and the late J.H. Walsh, CURE, Digestive Diseases Research Center, Antibody/RIA Core (HIH grant DK41301), Los Angeles, CA, U.S.A. (NPY); L. Terenius, Karolinska Institutet, Stockholm, Sweden (enkephalin, substance P); E. Theodorsson, Linköping University, Linköping, Sweden (galanin); E. Weber, CoCensys, Irvine, CA, U.S.A. (dynorphin); and R. Palmiter, University of Washington, Seattle, WA, U.S.A. (ZnT3).

Ancillary