artificial cerebrospinal fluid
heterozygous cholecystokinin-1 receptor knockout
homozygous cholecystokinin-1 receptor knockout
dorsomedial hypothalamic nucleus
dorsal motor nucleus of the vagus
epididymal adipose tissue
nucleus tractus solitaries
Otsuka Long-Evans Tokushima Fatty
In mammals, satiety-controlling mechanisms develop before the switch from ingestive behaviour from milk suckling to regular food intake; however, the underlying mechanisms are currently unknown.
We show that absence of the cholecystokinin-1 (CCK-1) receptor gene significantly increases suckling pup weight, regardless of maternal genotype in mice.
CCK-1 receptor expression was observed in satiety-controlling regions such as the hypothalamus, caudal brainstem, nodose ganglion and pylorus, which is consistent with suggested receptor function in adults, whereas corresponding receptor expression was low to non-existent at pre-weanling stages.
Third ventricle ependymal tanycyte-like cells transiently expressed CCK-1 receptors at critical developmental stages, and gastrointestinal milk filling upregulated cFos expression in these cells.
Localized blocking of ependymal CCK-1 receptors was sufficient to produce overweight pups, suggesting that the ependymal CCK-1 receptor is an infant-specific satiety controller.
Abstract Cholecystokinin (CCK) is a hypothetical controller for suckling and infancy body weight, although the underlying mechanisms remain unclear. Therefore, the present study analysed the mechanisms using mice lacking the CCK-1 receptor (CCK1R–/–). Although CCK1R–/– mice displayed normal weights at birth and adulthood, CCK1R–/– pups had enlarged adipocytes and were overweight from the first to second week after birth, regardless of maternal genotype. The lacZ reporter gene assay and/or calcium imaging analysis demonstrated that CCK-1 receptors were abundant in satiety-controlling regions such as the hypothalamus, brainstem, nodose ganglion and pylorus in adults, whereas these signals were few to lacking at pre-weanling stages. At postnatal day (PD) 6, the increase in cFos expression in the medullary nucleus tractus solitarius was similarly triggered by gastrointestinal milk- or saline filling in both genotypes, further indicating immature CCK-1 receptor function in an ascending satiety-controlling system during infancy. Conversely, third ventricle ependymal tanycyte-like cells expressed CCK-1 receptors with expression peaking at PD6. At PD6, wild-type but not CCK1R–/– mice had increased cFos immunoreactivity in ependymal cells following gastrointestinal milk filling whereas the response became negligible at PD12. In addition, ependymal cFos was not increased by saline filling, indicating that these responses are dependent on CCK-1 receptors, developmental stage and nutrients. Furthermore, body weights of wild-type pups were transiently increased by blocking ependymal CCK receptor function with microinjection of a CCK-1 antagonist, but not a CCK-2 antagonist. Hence, we demonstrate de novo functions of ependymal CCK-1 receptors and reveal a new aspect of infant satiety-controlling mechanisms.
Cholecystokinin (CCK), which was originally described as an intestinal hormone (Ivy & Oldberg, 1928; Mutt & Jorpes, 1968), is currently recognized as a classic signalling peptide known to control satiety and food intake behaviour (Gibbs et al. 1973; Antin et al. 1975). CCK peptides secreted from the gastrointestinal tract in response to ingestion control hypothalamic satiety-control centres via a large-scale afferent network including the nervus vagus and the medullary nucleus tractus solitarius (NTS) (Rinaman et al. 1993; Reidelberger et al. 2003, 2004; Bi et al. 2004). In addition, CCK is a ubiquitous neuropeptide produced in the central nervous system (Vanderhaeghen et al. 1975; Raiteri et al. 1993), which may directly control satiety (Hirosue et al. 1993). CCK-1 receptors have been suggested to play a pivotal role for satiety control because (i) CCK-1 receptors are expressed in the nervus vagus and the hypothalamic satiety centres (Bi et al. 2004; Shimazoe et al. 2008), (ii) intraperitoneal (i.p.) injection of a CCK-1 agonist (CCK-8s), but not a CCK-2 agonist (CCK-4), significantly reduces ingestion (Moran et al. 1992; Hirosue et al. 1993) and (iii) larger meal sizes were observed in the Otsuka Long-Evans Tokushima Fatty (OLETF) mutant rat strain, which lacks the CCK-1 receptor gene (Takiguchi et al. 1997; Moran et al. 1998), and in mice lacking the CCK-1 receptor (CCK1R–/–) (Bi et al. 2004).
Among the wide variety of ingestive behaviours, suckling is a characteristic behaviour common to mammals. However, satiety-controlling neural circuits are immature during infancy (Rinaman et al. 2000; Rinaman 2006) and it is not yet fully understood how infants control suckling. Involvement of CCK in suckling indices such as nipple-attachment latencies and milk intake volumes was first studied in rats by Blass and Teicher (1980). They found that CCK administration reduced these indices at postnatal day (PD) 15–20 but had no effects before PD10. However, PD3 rats already display food-seeking behaviours when they are separated from their mothers (Hall, 1979), a behaviour called independent ingestion. Robinson et al. (1988) demonstrated that independent ingestion of milk was inhibited by CCK injections in neonatal rats (PD1–3). Moreover, injection of the CCK-1 receptor antagonist MK-329 has been shown to increase independent ingestion and block the inhibitory effects of CCK-8s on independent ingestion in PD9–12 rats (Weller et al. 1990; Smith et al. 1991). Although the mechanisms underlying native suckling and independent ingestion may not be identical, these results suggest that CCK-1 receptors are involved in satiety control during early postnatal life.
Immunostaining studies have shown the presence of CCK peptides in diverse brain areas of neonatal rats (Cho et al. 1983; Kiyama et al. 1983). In addition, in situ hybridization studies showed that CCK mRNA is expressed in the paraventricular nucleus at embryonic day 17.5 (Giacobini & Wray, 2008). However, developmental changes in CCK-1 receptor expression and/or function in the brain have not been clearly demonstrated. It has been suggested that functional CCK-1 receptors may not be expressed in hypothalamic satiety centres during early development, because (i) CCK-binding in rat brains is low at PD1–2 but reached its maximum at PD12 (Hays et al. 1981) and (ii) CCK-8s-induced cFos expression is absent in rat hypothalamic neurons at PD2 (Rinaman et al. 1994). The observation of CCK-8s-induced cFos expression in the NTS in PD2 rats led to the hypothesis that medullary but not hypothalamic CCK-1 receptors may control satiety during early development (Rinaman et al. 1994, Rinaman, 2006).
Schroeder et al. (2006) demonstrated neonatal overgrowth in OLETF rats. This may be caused by mutations in the pups, because OLETF pups were heavier than the control strain pups even after their dams were reciprocally switched at PD10–12 (Schroeder et al. 2007). In addition, Blumberg et al. (2006) showed facilitated independent ingestion in OLETF pups. However, it is uncertain if these phenotypes are caused by a lack of CCK-1 receptors, because OLETF rats lack many other genes involved in metabolic control, such as pancreatic lipase (Muramatsu et al. 2005). Therefore, the present study used CCK1R–/– mice (Kopin et al. 1999; Takiguchi et al. 2002) and investigated the ontogeny of CCK-1 receptor function and localization, which may underlie infant body weight control.
All experiments were designed to minimize the number of animals used and their suffering based on international guidelines. All experiments were approved by the Committee of Animal Care from the National Kyushu Cancer Center and the University of Toyama, Japan (No. 20-7).
Mice were bred under a light–dark cycle (lights on 08:00–20:00 h) at a constant temperature (22 ± 1°C). Male CCK1R–/– mice and their wild-type control littermates were generated as described previously (Takiguchi et al. 2002; Shimazoe et al. 2008). In short, a targeting vector was designed to replace the Sal/I-BglII 1.9 kb genomic fragment of the mouse CCK-1 receptor gene with NLS-lacZ and pGK-neo cassette. The homologous recombination deleted the first 122 amino acids, including the first membrane-spanning region of the CCK-1 receptor. J1 embryonic stem (ES) cells were electroporated with the target vector and selected with G418 on embryonic fibroblast feeder cells. After Southern blot analysis, the successful ES clones were microinjected into blastocysts of C57BL/6J females. Two independent ES clones generated germline chimeras. The chimeras were bred with C57BL/6J mice to generate heterozygous (CCK1R+/–) mutant F1 mice. Finally, CCK1R–/– mice were generated by mating CCK1R+/– mice followed by sufficient backcross to C57BL/6J wild-type mice.
Body weight measurement
Body weights during PD3–21 were measured individually for the wild-type (n= 103) and CCK1R–/– (n= 110) mice. Because body weights of pups were not statistically different between the sexes at any point during PD3–21, data from male and female pups were pooled for the below analyses. First, body weights were measured for wild-type (n= 30) and CCK1R–/– (n= 40) pups grown by their own dams, which were fed regular chow ad libitum. Second, body weights were measured for wild-type (n= 40) and CCK1R–/– (n= 37) pups grown by their own dams, which were fed high-fat diets (% calories as fat 62.5, carbohydrate 25.5, protein 12; fat enhanced by soybean oil; Wako Pure Chemical Industry, Tokyo, Japan) ad libitum. The rest of the pups (4–6 pups per dam; total n= 33 for each genotype) were reared by substitutive mothers with reverse genotypes (n= 6 for each genotype). For the cross-fostering study, regular chow was supplied ad libitum to all dams and transferred pups were odorized with nesting chips to prevent refusal of sucking by the dams.
We also measured body weight in adult (5–10 weeks old; n= 5 for each genotype) and senior males (50–55 weeks old; n= 5 for each genotype), both of which were grown in our laboratory colony with access to regular chow ad libitum.
Animals were deeply anaesthetized with an i.p. injection of sodium pentobarbital (50 mg kg−1) and transcardially perfused with PBS for 5 min followed by ice-cold 4% (w/v) paraformaldehyde in 0.1 m phosphate buffer for 15 min. The brain, nodose ganglion and gastrointestinal axis were removed and further fixed in the same fixative (4°C, overnight). Then, the olfactory bulbs and/or cerebellum were removed from the brain in ice-cold PBS. Pre-trimmed brain tissues were immersed in 30% (w/v) sucrose and stored overnight at 4°C. The fixed and cryoprotected brain tissues were embedded with OCT compound (Sakura Finetek, Tokyo, Japan). Frozen sections of 50 μm thickness were cut using a cryostat microtome and washed three times with PBS in 24-well plates, after which they were mounted on glass slides. The cryostat brain sections, nodose ganglion and gastrointestinal axis were then stained with an X-gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) staining kit (K1465-01; Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. These samples were imaged using a colour CCD camera (Ds-5mc; Nikon, Tokyo, Japan) mounted on an inverted microscope (TE-2000, Nikon). Further details of X-gal staining techniques have been described previously (Shimazoe et al. 2008).
Brain sections (30 μm thickness) prepared as above were also used for immunofluorescence staining. The fixed sections were rinsed three times with PBS and then incubated in a blocking solution containing 10% (v/v) donkey serum (Jackson Immuno Research Laboratories, West Grove, PA, USA) and 0.1% (v/v) Triton-X 100 (Sigma RBI, St Louis, MO, USA) for 24 h at 4°C. For double labelling of β-galactosidase (β-gal) and vimentin, samples were incubated with 1:2000 FITC-conjugated rabbit anti-β-gal (Abcam, Cambridge, UK) dissolved in PBS containing 5% (v/v) donkey serum and 0.05% (v/v) Triton-X 100 for 48 h at 4°C. For antigen retrieval, samples were then mounted on glass slides and incubated in Tris-HCl buffer (200 mm, pH 9.0) for 15 h at 60–65°C. After three washes with PBS, these samples underwent secondary staining procedures with 1:100 rabbit anti-vimentin (GenScript Corp., Piscataway, NJ, USA) as a primary antibody and 1:200 Cy3-conjugated donkey anti-rabbit IgG (Jackson) as a secondary antibody, each incubated overnight at 4°C. For cFos immunostaining, brain sections were prepared from PD6 pups that were placed on a disposable body warmer and received an i.p. injection of 10 μg kg−1 CCK-8s (Sigma) or gastrointestinal infusion of milk (Pet milk SP; Morinyu Co., Tokyo, Japan) or saline (5% of body weight) using a fine polyethylene tube (300 μm outer diameter) 75–90 min before brain sampling. Following blocking procedures as described above, sections were incubated with 1:5000 rabbit anti-cFos (Calbiochem, La Jolla, CA, USA) and 1:200 Cy3-conjugated donkey anti-rabbit IgG as a secondary antibody overnight at 4°C. These samples were thoroughly washed with PBS and then embedded with Elvanol polyvinyl alcohol. Fluorescent images were acquired using a cooled monochrome CCD camera (Ds-2MBWc; Nikon), 100 W xenon lamp, and standard FITC/Cy3 filter sets mounted on an inverted microscope (TE-2000; Nikon). Digital images were processed using PhotoShop CS5 software (Adobe Systems, San Jose, CA, USA).
Confocal microscopy of epididymal adipose tissue (EAT)
The weight of EAT isolated from wild-type and CCK1R–/– pups and the size of adipocytes in the EAT were analysed at PD7 or PD17. For measurements, mice were deeply anaesthetized with pentobarbital (50 mg kg−1, i.p.) and EAT was removed along with the testis. Wet weights of EAT were measured following removal of the testis in PBS. The EAT weight at PD17 following high-fat diets were compared. The size of EAT in pups bred under normal conditions was too small for quantitative weight measurement. Alternatively, the size of adipocytes was analysed in detail using plasma membrane staining and confocal microscopy. The EAT of PD7 or PD17 pups was isolated as above and fixed in 10% (w/v) paraformaldehyde in PBS (4°C, 48 h) and cryoprotected in 30% (w/v) sucrose in PBS (4°C, overnight) following trimming of tissue to an approximately 5 × 5 mm square. EAT was further immersed overnight at room temperature in a 1:2 mixture of 30% (w/v) sucrose and OCT compound. Finally, tissues were embedded with OCT compound. Frozen sections of 15 μm thickness were cut using a microtome and mounted on glass slides. The sections were washed three times with PBS and then stained with 1:2000 CellMask™ deep red plasma membrane stain (Invitrogen) dissolved in PBS (37 °C, 15 min). These samples were thoroughly washed with PBS and then embedded with glycerol. Immunofluorescence images were viewed using a confocal imaging system equipped with an inverted microscope, UPLSAPO 40 × NA0.90 objective lens and multicolour lasers (Fluoview 1000; Olympus, Tokyo, Japan). Scanning parameters were unified across specimens. Digital images were processed using PhotoShop CS5 (Adobe).
Ca2+ imaging in developing brain slices
Coronal hypothalamic slices (200–220 μm) containing the dorsomedial hypothalamic nucleus (DMH) were prepared from PD3–12 mice following deep pentobarbital anaesthesia using a vibrating blade microtome in ice-cold high-Mg2+ artificial cerebrospinal fluid (ACSF) containing 138.6 mm NaCl, 3.35 mm KCl, 21 mm NaHCO3, 0.6 mm NaH2PO4, 9.9 mm d-glucose, 0.5 mm CaCl2 and 4 mm MgCl2, and bubbled with 95% O2/5% CO2. Two sequential slices were prepared from each mouse brain and placed separately on 7 × 7 mm 12 μm membrane filters (Nuclepore, Cambridge, MA, USA). The slices were incubated at room temperature for 1–8 h in regular ACSF (2.5 mm CaCl2 and 1 mm MgCl2) bubbled with 95% O2/5% CO2. For Ca2+ imaging, the slices were placed in a 0.40 μm filter cup (30 mm diameter; Millicell-CM, Millipore, Bedford, MA, USA) and immersed for 45–60 min in 1.5 ml of regular ACSF containing 10 μm fura-2 AM (Dojindo Laboratories, Kumamoto, Japan) and 0.01% (v/v) Pluronic (Invitrogen). During the entire staining procedure, the staining solution was gently bubbled with 95% O2/5% CO2 through a stainless steel pipe located outside the cup filter. After washing with ACSF and incubating for an additional 30 min, the slice was gently removed from the holding membrane filter and placed in a glass-bottomed microscope stage chamber (0.5 ml) for optical measurements.
Fluorescence images were obtained using an upright microscope (Axioskop; Carl Zeiss, Thornwood, NY, USA) with a water-immersion objective (Achroplan × 40, NA 0.75; Carl Zeiss). The wavelength of the excitation UV light (340 nm or 380 nm pulse; 100 ms) was switched using a filter wheel (Lambda 10–2; Sutter Instruments, Novato, CA, USA). The UV light was generated by a full-spectrum 175 W xenon bulb (Lambda LS; Sutter), conducted to the microscope through a liquid light guide and reflected using a dichroic mirror (FT 395 nm; Carl Zeiss). The pair of fluorescence images was processed using a band-pass filter (BP 485–515 nm; Carl Zeiss) and exposed to a multiple format cooled CCD camera (CoolSnap-fx; Photometrics, Tokyo, Japan) at 6 s intervals. The filter wheel and the CCD camera were controlled using digital imaging software (MetaFluor ver. 6.0; Japan Molecular Devices, Tokyo, Japan). The background fluorescence was also subtracted using the software. During recording, slices were placed in a 0.5 ml bath chamber and perfused with ACSF containing tetrodotoxin (0.5 μm) at a flow rate of 2.5 mL min−1. The 10 nm CCK-8s, 100 nm CCK-4 (Peptide Institute, Osaka, Japan), 100 nm lorglumide (LGM; LKT Laboratories Inc., St. Paul, MN, USA), 300 nm CI-988 (Tocris, Bristol, UK) and 60 mm potassium (high K+) ACSF were applied by switching the perfusate. Further details for Ca2+ imaging techniques have been described elsewhere (Ikeda et al. 2003).
Microinjection of CCK receptor antagonists
The wild-type (n= 37) and CCK1R–/– (n= 29) pups were marked by a permanent marker to identify individuals and anaesthetized with an i.p. injection of sodium pentobarbital (30 mg kg−1) at PD5. A glass pipette filled with 5 μl of drug solution was stereotaxically inserted at AP –0.5 mm, ML +0.7 mm and DV + 2.0 mm from bregma. The CCK-1 receptor antagonist (LGM; 100 μm), CCK-2 receptor antagonist (CI-988; 100 μm) or vehicle (saline containing 0.1% (v/v) DMSO) was injected into the lateral ventricle using a microinjector (Model IM-108; Narishige, Tokyo, Japan; net injectant volume ≅ 2 μl). Pups were then returned to their original cages. Dams were fed high-fat diets after delivery. Pup body weights were measured at PD5, PD6 and PD10. To visualize cerebral distribution of LGM following microinjections, fluorescent 4-(N,N-dimethylaminosulfonyl)-7-piperazinobenzofurazan (DBD-PZ) linked to the carboxyl terminal of LGM (LGM/DBD-PZ; gifts from Dr Naoki Toyooka at the University of Toyama) was microinjected as above in the wild-type pups at PD5 (n= 3). As a control, unlinked DBD-PZ (Tokyo Chemical Industry, Co., Tokyo, Japan) was microinjected in a separate group (n= 3). Brains were collected for cryostat sections 24 h after injection, and LGM labelling was visualized by the fluorescence intensity of LGM/DBD-PZ, which was subtracted from non-specific DBD-PZ background levels.
Data are presented as the means with standard error. Two-way analysis of variance (ANOVA) or one-way ANOVA followed by Duncan's multiple range tests were used for statistical comparison across multiple means. The two-tailed Student's t test was used for pairwise comparisons. A 95% confidence level was considered to indicate statistical significance.
Overgrowth in CCK1R–/– pups
CCK1R–/– mice had normal body weights at PD3 (n.s.; Fig. 1Aa), but they grew faster during the subsequent pre-weanling stage. Consequently, CCK1R–/– mice were significantly heavier than wild-type mice at PD5 (P < 0.05; Fig. 1A). The tendency for overgrowth was sustained until PD14, when body weights were 12% higher than the wild-type pups (F1,11= 41.4, P < 0.01; Fig. 1A). Body weight curves remained parallel until PD21. The tendency for overgrowth in CCK1R–/– pups was enhanced by high-fat diets supplied to the dams (Fig. 1Ab). A statistical difference was first observed at PD7 (P < 0.05; Fig. 1A) and the body weights of CCK1R–/– mice reached a maximum 20% increase over wild-type mice at PD19 (F1,16= 51.7, P < 0.01; Fig. 1A).
Pups from CCK1R–/– mice raised from birth by wild-type dams also displayed overgrowth, becoming heavier than normal wild-type pups at PD7 (F3,132= 4.4, P < 0.05) or later (Fig. 1Ba). The CCK1R–/– pups raised by wild-type dams were also heavier than wild-type pups raised by CCK1R–/– dams at PD7 (F3,132= 4.4, P < 0.05) or later. However, wild-type pups raised by CCK1R–/– dams did not display overgrowth. In fact, they were lighter than normal CCK1R–/– pups at PD7 (F3,132= 4.4, P < 0.05) or later (Fig. 1Bb) and were not statistically different from normal wild-type pups. These results indicate that pups from CCK1R–/– mice upregulate suckling at/before PD5–7 and become overweight during pre-weanling stages, regardless of dam genotype.
Body weights at young adult and senior stages were not statistically different between genotypes (n= 5 for each genotype; Fig. 1C).
Adipocyte enlargement in CCK1R–/– pups
EAT from CCK1R–/– pups at PD17 from dams fed a high-fat diet were 1.7-fold heavier than wild-type mice (P < 0.01; Fig. 2A). EAT weights from pups bred under normal conditions were not measureable at PD17 because of size, whereas the average size of adipocytes in the EAT was larger in CCK1R–/– than wild-type mice (P < 0.05 at PD7 and P < 0.01 at PD17; Fig. 2B). These results indicated that adipose enlargement in CCK1R–/– pups was facilitated by suckling during pre-weanling stages.
CCK-1 receptor expression in the satiety-controlling axis
To analyse CCK-1 receptor expression in the ascending satiety-controlling system at pre-weanling stages, the present study measured CCK-1 receptor expression in the caudal brainstem, nodose ganglion, stomach and gut of CCK1R+/– mice using the X-gal staining method (Fig. 3). In the caudal brainstem, blue staining was distributed within the hind brain sub-region containing the NTS and dorsal motor nucleus of the vagus (NTS/DMV) at adult stages (Fig. 3A). This result is similar to the distribution reported using in situ hybridization (Honda et al. 1993). However, the corresponding hindbrain signal was lacking at PD6 (Fig. 3A). Also, X-gal staining signals were dense in ganglion nodes at adult stages, whereas corresponding signals were lacking at PD6 (Fig. 3B). In the gastrointestinal axis, similar to CCK expression (Lay et al. 1999), X-gal staining signals were localized to the upper duodenum, and markedly in the pylorus in adult mice, but the signals were also undetectable in PD6 mice (Fig. 3C). It should be noted that in PD6 mice, X-gal signals appear to be sparsely distributed in enteroendocrine cells in the inner layer of the descending duodenum, although the X-gal signal was too faint to allow identification of specific cell types. These results suggest limited expression of peripheral CCK-1 receptors during infancy.
To further explore the developmental processes of CCK-1 receptor gene expression in the brain, developing brains were also analysed for X-gal staining (Fig. 4). At neonatal stages (PD3–6), signals in the DMH, the ventral medial hypothalamus and the arcuate nuclei (ARC) were low or undetectable (Fig. 4A). However, signals in these neuronal nuclei increased thereafter until juvenile stages (PD9–20). Interestingly, dense signals were found at the third ventricular (3V) ependyma at PD3–6 (Fig. 4Ba, b). These signals were gradually lost and were replaced by signals in the above neuronal nuclei during PD9–12 (Fig. 4Bc).
Characteristics of 3V ependymal cells were further analysed by double immunostaining for β-gal (corresponding to the aforementioned reporter gene expression) and vimentin (tanycyte or radial glial cell marker in the ependyma) in CCK1R+/– mice at PD6. All β-gal-positive ependymal cells co-localized with vimentin staining (total number of cells observed = 36 in three separate slices; Fig. 5). Together with the morphology of β-gal-positive ependymal cells, the results indicate that CCK-1 receptor expression in 3V tanycytes occurs at a critical pre-weanling stage.
Function of CCK-1 receptors in the 3V ependyma in developing mice
Functionality of 3V CCK-1 receptors during early developmental stages was studied using Fura-2-based Ca2+ imaging in acutely isolated hypothalamic slices (Figs 6 and 7). Bath application of CCK-8s (10 nm) evoked Ca2+ transients in about half of the 3V ependymal cells at PD6 (48 ± 5%; number of slices = 7; number of responsive cells = 59; Fig. 6A). Because this response was absent in slice preparations from CCK1R–/– mice (Fig. 6B), the ependymal CCK-8s response was theoretically dependent on CCK-1 receptors. This response was also observable, but in smaller responsive cell populations, at PD3 (21 ± 3%; number of slices = 7; number of responsive cells = 31) and PD9 (17 ± 4%; number of slices = 6; number of responsive cells = 17; F3,23= 26.8, P < 0.01 compared with the responsive cell populations at PD6; Fig. 6D). In addition, the ependymal CCK-8s response was almost undetectable at PD12 (3 ± 1.5%; number of slices = 7; number of responsive cells = 3; F3,23= 26.8, P < 0.01; Fig. 6C and D), suggesting that CCK-1 receptors function within a particular developmental window. However, the same bath application of CCK-8s evoked Ca2+ transients in 25 ± 5% of DMH cells at PD12 (number of slices = 8; number of responsive cells = 54; Fig. 6B and E), and this response was significantly smaller at earlier developmental times (F3,28= 12.6, P < 0.01; Fig. 6E). Also, amplitudes of CCK-8s-induced Ca2+ mobilizations in 3V ependymal cells (mean F ratio changes = 0.10 ± 0.01) were much larger than those in DMH cells at PD6 (mean F ratio changes = 0.03 ± 0.004, number of cells = 12; P < 0.01).
We also analysed the expression of CCK-2 receptors in 3V ependymal cells using Ca2+ imaging experiments (Fig. 7). When the CCK-2 receptor agonist CCK-4 (100 nm) was applied to PD6 slices in combination with CCK-8s (10 nm), all 3V ependymal cells that responded to CCK-8s also responded to CCK-4 (40 of 40 cells in four slices, Fig. 7A). The CCK-2 receptor antagonist CI-988 (300 nm) blocked only the CCK-4-induced responses, whereas the CCK-1 receptor antagonist LGM (100 nm) blocked only CCK-8s-induced responses (Fig. 7A and B). This suggests co-expression of CCK-1 and CCK-2 receptors in 3V ependymal cells. However, the function of CCK-2 receptors in 3V ependymal cells does not appear to depend on developmental stage because CCK-4-induced Ca2+ mobilizations were observed in 3V ependymal cells both at PD6 (responsive cell ratio = 73 ± 5% in five slices; number of cells = 69) and PD12 (responsive cell ratio = 67 ± 6% in seven slices; number of cells = 71; n.s.; Fig. 7C). These results indicate functional CCK-1 receptor expression in 3V ependymal cells within a specific time window during postnatal development.
cFos expression in response to satiety-controlling signals
The i.p. injection of CCK-8s (10 μg kg−1) at PD6 transiently reduced body weight in wild-type pups but not in CCK1R–/– pups (1 day weight gain after injection, 0.49 ± 0.04 g in the wild-type pups and 0.85 ± 0.06 g for CCK1R–/– pups, n= 6–7 for each group; P < 0.01). Therefore, to determine the role of CCK-1 receptors in satiety control during early postnatal life, cFos immunofluorescent staining in the developing hypothalamus and caudal brainstem was performed following i.p. injection of CCK-8s (10 μg kg−1) at PD6. In the caudal brainstem, cFos expression in the NTS/DMV of wild-type mice was 7-fold higher than after saline injection (P < 0.01; Fig. 8A and B). Also, increases in cFos expression were observed in the area postrema (AP) of wild-type mice after CCK-8s injection (Fig. 8A) whereas the expression levels were more variable. In the hypothalamus, cFos expression in DMH cells was only 1.7-fold higher after CCK-8s injection than after saline injection (P < 0.05; Fig. 8C and D). On the other hand, cFos expression in 3V ependymal cells of wild-type mice was 10-fold higher after CCK-8s injection than after saline injection (P < 0.05; Fig. 8C and E). CCK-8s had a negligible effect (<200 cells mm−2) on cFos expression in all loci in CCK1R–/– mice, and the expression levels were not statistically different from the corresponding saline-injected controls.
Subsequently, cFos expression was analysed following gastrointestinal filling with milk or saline in PD6 (Fig. 9) and PD12 (Fig. 10) mice. In wild-type pups, milk significantly upregulated cFos expression in NTS/DMV cells compared with fasting controls (22-fold at PD6, F2,7= 115.4, P < 0.01; Fig. 9A and B and 42-fold at PD12, F2,7= 45.4, P < 0.01; Fig. 10A and B). Saline also upregulated cFos expression in the wild-type NTS/DMV (16-fold at PD6, F2,7= 115.4, P < 0.01; Fig. 9A, B and 29-fold at PD12, F2,7= 45.4, P < 0.01; Fig. 10A and B). Similar to the wild-type responses, milk upregulated cFos expression in NTS/DMV cells compared with fasting controls in CCK1R–/– mice (37-fold at PD6, F2,8= 44.3, P < 0.01; Fig. 9A and B and 24-fold at PD12, F2,11= 22.1, P < 0.01; Fig. 10A and B). In addition, saline also upregulated cFos expression in the NTS/DMV of CCK1R–/– mice (30-fold at PD6, F2,8= 44.3, P < 0.01; Fig. 9A and B and 15-fold at PD12, F2,11= 22.1, P < 0.01; Fig. 10A and B). These results suggest that hindbrain responses did not depend on CCK-1 receptors at both PD6 and PD12.
Within the hypothalamus, the present study quantified cFos expression in the DMH and the 3V ependyma. The DMH displayed similar but significantly lower increases in cFos expression than the NTS/DMV following milk or saline infusions. At PD6, maximal cFos expression in the DMH following milk filling was 31% of the NTS/DMV peak for wild-type mice and 15% of the NTS/DMV peak for CCK1R–/– mice (Fig. 9C and D). Also, at PD12, maximal cFos expression following milk filling was 28% of the NTS/DMV peak for wild-type mice and 46% of the NTS/DMV peak for CCK1R–/– mice (Fig. 10C and D). However, unlike the DMH, only milk filling at PD6 upregulated cFos expression in 3V ependymal cells of wild-type mice (F2,13= 188.4, P < 0.01; Fig. 9C and E). cFos expression in 3V ependymal cells was not observed in fasting conditions, while it reached 43% of the NTS/DMV peak in wild-type mice at PD6. More importantly, both milk and saline failed to increase cFos expression in 3V ependymal cells in CCK1R–/– mice at PD6 (F2,13= 1.5, n.s.; Fig. 9C and E). Furthermore, milk filling did not upregulate cFos expression in 3V ependymal cells of both genotypes at PD12 (20 ± 14.6 cells mm−2, n= 5; F2,9= 0.8, n.s. for the wild-type and 8.3 ± 5.3 cells mm−2, n= 6; F2,10= 0.9, n.s. for CCK1R–/–; Fig. 10C). These results demonstrate that the ependymal cell response was dependent on CCK-1 receptors, nutrients and developmental stage.
Blocking 3V CCK-1 receptor function increases body weight in pups
To estimate the function of ependymal CCK-1 receptors, pups received intracerebroventricular (i.c.v.) injection of LGM or CI-988 on PD5. First, we estimated the distribution of LGM in the brain following i.c.v. injection using LGM/DBD-PZ. We found that a 2 μl injection at a concentration of 100 μm was sufficient for LGM to reach 3V ependymal cells and that labelling lasted for a day after injection (Fig. 11A). Neither hypothalamic neuronal nuclei such as the paraventricular nucleus and the DMH, nor brainstem nuclei were labelled by the LGM/DBD-PZ injection, enabling specific blocking of ependymal CCK-1 receptors in mouse pups (Fig. 11A and B).
After microinjection of LGM, CI-988 or vehicle in the wild-type and CCK1R–/– mice on PD5, body weights were analysed. In the wild-type pups, weight gain during the first 24 h (PD5–6) was significantly greater in the LGM-injected group than in the CI-988- or vehicle-injected groups (F2,34= 11.4, P < 0.01; Fig. 11C). This transient increase in body weight following the microinjection of LGM was not observed in CCK1R–/– mice (Fig. 11C). Also, in the following days (PD6–10), the effects of LGM injections decreased to below statistical significance in wild-type mice (F2,34= 2.3, n.s.; Fig. 11D). These results indicate that localized blocking of CCK-1 receptors in ependymal cells is sufficient to produce overweight pups at PD5–6.
The present study demonstrated that (i) CCK1R–/– pups are overweight, (ii) little or no CCK-1 receptor expression in the hypothalamic neuronal nuclei, caudal brainstem, nodose ganglion and pylorus is observed at pre-weanling stages, and (iii) there is alternative CCK-1 receptor expression in 3V ependymal cells. In addition, cFos expression was upregulated in ependymal cells following gastrointestinal milk filling in wild-type pups, but not in CCK1R–/– pups. Furthermore, gastrointestinal saline filling did not affect cFos expression in ependymal cells, but increased cFos in the caudal brainstem. Therefore, nutrient complacency, but not physical stimulation of gastrointestinal systems, activated ependymal CCK-1 receptors. Because i.c.v. injection of the CCK-1 antagonist but not the CCK-2 antagonist produced a transient overweight phenotype, blocking of ependymal CCK-1 receptors may be sufficient to induce weight gain at pre-weanling stages. Taken together, these results suggest that ependymal CCK-1 receptors are nutrient-specific satiety controllers at pre-weanling stages.
CCK-1 receptor as a common satiety controller for pups
Schroeder et al. (2006) analysed the body weights of OLETF rats from PD1 to PD65, and demonstrated that OLETF rats were heavier than controls throughout the developmental process. However, the difference was more evident at pre-weanling stages (PD15–30) than at newborn stages (PD1–10). The present study demonstrated that body weights of CCK1R–/– mice were not different from those of wild-type mice at PD3, but they became heavier at PD5–7 or later. Because suckling pups do not display significant locomotion and their basal body temperatures are under the influence of dams, it is appropriate to interpret that observation of overweight pups could be due to facilitated suckling, but not to the differential metabolic rates in pups lacking CCK-1 receptors. In the rat pylorus, CCK-1 receptor expression peaks at PD10, and neonatal CCK-1 receptor function has been reported (Robinson et al. 1987; Schwartz et al. 1990). Here, we observed small CCK-1 receptor expression in the mouse pylorus at PD6. Together with a 1 week shift in the developmental schedule for CCK-1 receptor-dependent body weight changes in mice and rats, the principal machineries and pathways underlying CCK-dependent satiety systems during pre-weanling stages may differ across animal species.
On the other hand, we would rather emphasize the commonality in CCK-1 receptor-mediated satiety control across animal species during pre-weanling stages, because CCK-1 receptor function for body weight control in adult rodents currently remains controversial. For example, CCK1R–/– mice do not develop hyperphagia and obesity when maintained on regular chow (Kopin et al. 1999; Takiguchi et al. 2002) despite the obvious obese phenotype of OLETF rats. Bi et al. (2004) demonstrated that OLETF rats had elevated neuropeptide Y mRNA expression in the DMH, whereas this was not seen in CCK1R–/– mice. Because the CCK-1 receptor co-localizes with neuropeptide Y in the DMH of control strain rats (Bi et al. 2004) and knockdown of neuropeptide Y in OLETF rats results in a significant reduction in body weight and food intake (Yang et al. 2009), the phenotypic discrepancy between rats and mice has been explained by differential receptor expression in the DMH and resultant uncoupling from neuropeptide Y signalling. However, Blevins et al. (2009) demonstrated that CCK-1 receptor knockout rats (F344,Cck1r–/–) are not obese, as observed in CCK1R–/– mice. Thus, it is possible that the phenotype of OLETF rats depends on genes other than the CCK-1 receptor gene, which has been demonstrated previously (Muramatsu et al. 2005).
The present study shows that there are few to no CCK-1 receptors in the DMH during pre-weanling stages in mice, but that they are expressed in the adult mouse DMH. Because mouse pups were overweight even though the DMH lacked CCK-1 receptor expression, satiety controllers such as those discussed above for adults may not be applicable to mice at pre-weanling stages. Because we recently observed over-expression of CCK-2 receptors in parvocellular PVN neurons of CCK1R–/– mice at adult stages (Mohammad et al. 2012), such compensatory gene expression may mask intrinsic CCK-1 receptor function for adult satiety control. Thus, commonality in pups overweight in OLETF rats and CCK1R–/– mice may be explained by the lack of compensatory mechanisms during the early stage of satiety control.
CCK-1 receptor expression and function at pre-weanling stages
The present study revealed that CCK-1 receptors in the gastrointestinal axis, vagal sensory nerves, caudal brainstem and hypothalamus, which have been described to control satiety responses (Rinaman et al. 1993, 2000; Reidelberger et al. 2003, 2004; Bi et al. 2004), were present in PD20 mice, whereas corresponding signals were negligible in PD6 mice. Nevertheless, the fact that body weight and fat increased faster in sucking CCK1R–/– pups raised the question of where in the body the CCK-1 receptors enroll their critical function.
CCK-8s injection (10 μg kg−1, i.p.) in adult rats elevated cFos in the satiety centres in the hypothalamus and caudal brainstem (Rinaman et al. 1993). On the other hand, the same CCK-8s injection increased cFos expression in the caudal brainstem (NTS and AP), but not in the hypothalamus in PD2 rats (Rinaman et al. 1994). These former studies indicate development of brainstem but not hypothalamic satiety controls via CCK-1 receptors in neonatal rats. Consistent with this, the present study observed similar cFos expression patterns in PD6 wild-type mice following the same CCK-8s injections and corresponding signals were completely lacking in CCK1R–/– pups. However, it should be noted that the cFos signal does not fit CCK-1 receptor expression patterns analysed by the lacZ reporter gene assay, where almost negligible signals were observed in the NTS/DMV and AP at PD6. Thus, it is reasonable to interpret that CCK-8s-induced cFos elevations in the infant brainstem may be caused by remote activation via CCK-1 receptors outside the field.
We analysed possible site(s) affecting infancy weight gain and discovered uncharacterized CCK-1 receptor gene expression in the 3V ependyma. Furthermore, using Ca2+ imaging techniques in developing brain slices and immunofluorescence double-labelling analyses, we determined that vimentin-positive ependymal cells express functional CCK-1 receptors within a limited time window during early postnatal life. The CCK-1 receptor-positive cell sends lateral projections to the brain parenchyma forming a typical tanycyte structure. The lateral projections may interact with surrounding neurons as intracellular Ca2+ moderately increases in DMH neurons at PD6 following CCK-8s stimulation. We also demonstrated that well-documented satiety-controlling centres in the hypothalamus lacked CCK-1 receptor gene expression at birth, but that CCK-1 expression gradually increased between PD9 and PD20. This may explain why hypothalamic satiety centres failed to express cFos in response to peripheral CCK administration (Rinaman et al. 1994), while CCK-1 receptor antagonists and/or gene deletion of CCK-1 receptors upregulated ingestion in neonates (Weller et al. 1990; Blumberg et al. 2006). On the other hand, dramatic shifts in X-gal signals from ependymal cells to the brain parenchyma during postnatal development raised the possibility that these cells could be neuronal progenitors, whereas it is still a matter of debate whether vimentin-positive glial cells can differentiate into neurons during postnatal development (Alvarez-Buylla et al. 2001).
A former study using retrograde neuronal tracers has shown that projections from the hypothalamic neuronal nuclei (such as DMH and PVN) to the medullary dorsal vagal complex (such as the NTS) already exist at birth but are significantly increased during pre-weanling stages in rats (Rinaman et al. 2000). Thus, remote control of hypothalamic neuronal nuclei by ependymal CCK-1 receptors could be a potential output for infancy satiety controls. On the other hand, failure to characterize ependymal CCK-1 receptors in former studies (Rinaman et al. 1994) is reasonable, because conventional cFos immunostaining using an avidin–biotin complex produces high background staining at the ventricular ependyma, which should not be incorporated in the data analyses.
Differential satiety-controlling mechanisms between adults and pups
Leptin and its receptor have been identified as the products of the obese (ob) gene (Zhang et al. 1994) and the diabetic (db) gene (Lee et al. 1996) in monogenic mutant mice. Leptin is a peptide hormone secreted from adipocytes. The results of the present study demonstrated faster enlargement of adipocytes in CCK1R–/– pups and thus involvement of leptin signalling in overweight pups should be discussed. Indeed, several functions of CCK-1 receptors in satiety control have been explained by synergic interactions with leptin-mediated signalling in adults (Wang et al. 2000; Peters et al. 2006; Bi et al. 2007). Leptin has been shown to work as a neurotrophic factor, promoting the development of neural projections from the ARC at neonatal stages (Bouret & Simerly, 2004, 2007). Thus, it may be associated with the formation of satiety-controlling circuits. Interactions between CCK and leptin-mediated signalling in neonatal satiety control were suggested because gastric CCK and leptin additively elevated electrical activities in the NTS of 1- to 5-day-old rats (Yuan et al. 2000). Indeed, leptin administration to neonatal mice and rats upregulates the metabolic rate, although it does not influence food intake (Stehling et al. 1996; Mistry et al. 1999). However, it has also been shown that leptin in neonates is primarily via transport from the placenta (Hassink et al. 1997). This may rule out the possible involvement of leptin as an essential cause for infancy overgrowth in CCK1R–/– mice, because the phenotype did not depend on maternal genotype.
The ontogeny of hindbrain-hypothalamic circuits involved in feeding behaviours has been studied in detail and one possible interpretation underlying CCK-mediated satiety control during infancy is due to remote hypothalamic control via the NTS (Rinaman et al. 2000; Rinaman, 2006). The present results showing that gastrointestinal milk or saline filling both enhanced cFos expression in the NTS/DMV of PD6–12 mice demonstrate activation of the NTS/DMV independent of nutrient complacency at these stages. It has been suggested that satiety response in rats at birth is triggered by nutrient-independent mechanical filling of the stomach (Houpt & Epstein, 1973) and thus such neonatal type satiety responses may be mediated via NTS activation. However, results showing that gastrointestinal filling similarly enhanced cFos expression in the NTS/DMV of both wild-type and CCK1R–/– mice suggests that remote hypothalamic controls, if any, via the NTS/DMV may not explain the overweight phenotype of CCK1R–/– mice at pre-weanling stages.
Ependymal CCK-1 receptor function as a satiety controller during infancy
The results of the present study demonstrated that milk but not saline filling increased cFos in 3V ependymal cells at PD6 whereas corresponding signals were significantly reduced at PD12. Thus, it seems likely that ependymal CCK-1 receptors specifically function in nutrient-dependent satiety signalling in mice pups at specific pre-weanling ages. To further analyse the specific function of ependymal CCK-1 receptors in suckling and energy balance during infancy, we examined i.c.v. microinjection of LGM at the critical developmental time. An LGM concentration sufficient to bind to ependymal cells, but insufficient to permeate into the brain parenchyma, produced transient overweight wild-type mice. In addition, this transient effect of LGM was absent in CCK1R–/– controls. Thus, we conclude that ependymal CCK-1 receptors function as critical regulators for suckling and energy balance at pre-weanling stages. Interestingly, CI-988 microinjection failed to increase pup weight despite its similar inhibitory effect on CCK-mediated Ca2+ mobilizations. Because tanycyte-like fibres are widely distributed in the brain parenchyma and fura-2 imaging studies do not identify subcellular localization of CCK-1 and -2 receptors, a possible explanation of this discrepancy is that there are differential spatial distributions and functions (e.g. differential CCK inputs) between CCK-1 and -2 receptors. Results showing the presence of CCK-4-induced Ca2+ mobilization independent of the developmental time support this hypothesis.
The mechanisms behind the involvement of tanycyte-like cells in the regulation of infant satiety control raise questions because the characteristics of these cells are poorly understood at any developmental stage. Sanders et al. (2004) examined i.c.v. injections of a glucokinase inhibitor in adult rats and observed (i) reversible destruction of 3V tanycytes, (ii) reversible impairment in glucose counter regulatory responses and (iii) temporal inhibition of food intake. Also, Frayling et al. (2011) recently visualized glucose-induced Ca2+ mobilizations in 3V tanycytes. These former results pointed to a possible role of tanycyte signalling in the regulation of glucose homeostasis and metabolism in adults. Inhibitory control of independent ingestion by glucose has been shown in rats at PD9–15 (Swithers & Hall, 1989). However, Weller et al. (1997) showed that the CCK-1 receptor antagonist devazepide failed to reduce the inhibitory effects of glucose preload on rat-independent ingestion at PD12. Thus, the function of tanycytes in satiety control in pups may be related to, but not simply explained by, the proposed functions for glucose homeostasis in adults.
In conclusion, we found expression of CCK-1 receptors in 3V tanycyte-like cells during early postnatal life and demonstrated de novo functions in infant satiety control. This mechanism was activated before the formation of well-documented hypothalamic satiety control centres, but was disabled at adult stages, bridging neonatal to adult-type ingestions. The results from the present study shed light on the mechanisms underlying infant obesity and delectation; however, the precise molecular mechanisms and source of CCK peptides involved remain unclear and require further research.
T.O. performed the experiments and co-analysed the data. S.M. co-analysed the data and co-wrote the manuscript. E.M. co-analysed the data and co-designed the study. S.T. supplied experimental materials and co-designed the study. M.I. co-designed and supervised the study, co-analysed the data, and co-wrote the manuscript.
This work was supported in part by a Grant-in-Aid for scientific research (22300108) from the Ministry of Education, Culture, Sports, Science, and Technology, Japan, to M.I. We thank S. Miyakoshi (University of Toyama) for her assistance with body weight measurements, Dr N. Toyooka (University of Toyama) for donating LGM/DBD-PZ and Dr T. Shimazoe (Kyushu University) for valuable discussion.