Abstract
Prostaglandin E2 (PGE2) in the hypothalamus is a principal mediator of the febrile response. However, the role of organic anion transporting polypeptide 2A1 (OATP2A1/SLCO2A1), a prostaglandin transporter, in facilitating this response is unknown. Here, we investigated the effect of Slco2a1 deficiency on the body core temperature (Tc) and on the PGE2 concentration in hypothalamus interstitial fluid (Cisf) and CSF (Ccsf) of lipopolysaccharide (LPS; 100 μg/kg, i.p.)-treated mice of both sexes. Slco2a1−/− mice did not develop a febrile response. Ccsf was increased in Slco2a1+/+ and Slco2a1−/− mice, and Ccsf of Slco2a1−/− mice was well maintained at 5 h after LPS injection (1160 pg/ml) compared with Slco2a1+/+ mice (316 pg/ml). A microdialysis study revealed that Cisf peaked at 2 h after LPS injection in Slco2a1+/+ mice (841 pg/ml), whereas the increase in Cisf was negligible in Slco2a1−/− mice. The PGE2 plasma concentration in Slco2a1−/− mice (201 pg/ml) was significantly higher than that in Slco2a1+/+ mice (54 pg/ml) at 1 h after LPS injection, whereas the two groups showed similar PGE2 concentrations in the hypothalamus. Strong Oatp2a1 immunoreactivity was observed in F4/80-positive microglia and perivascular cells and in brain capillary endothelial cells. The changes in Tc and Cisf seen in LPS-injected Slco2a1+/+ mice were partially attenuated in monocyte-/macrophage-specific Slco2a1−/− (Slco2a1Fl/Fl/LysMCre/+) mice. Thus, OATP2A1 facilitates the LPS-induced febrile response by maintaining a high level of Cisf, possibly by regulating PGE2 secretion from F4/80-positive glial cells and/or facilitating PGE2 transport across the blood–brain barrier. These findings suggest that OATP2A1 is a useful therapeutic target for neuroinflammation.
SIGNIFICANCE STATEMENT Fever is a physiological response caused by pyrogen-induced release of prostaglandin E2 (PGE2) in the hypothalamus, which plays a central role in regulating the set-point of body temperature. However, it is unclear whether the prostaglandin transporter OATP2A1/SLCO2A1 is involved in this response. We show here that LPS-induced fever is associated with increased PGE2 concentration in hypothalamus interstitial fluid (Cisf), but not in CSF (Ccsf), by means of a microdialysis study in global Slco2a1-knock-out mice and monocyte-/macrophage-specific Slco2a1-knock-out mice. The results suggest that OATP2A1 serves as a regulator of Cisf in F4/80-positive glial cells. OATP2A1 was detected immunohistochemically in brain capillary endothelial cells and, therefore, may also play a role in PGE2 transport across the blood–brain barrier.
Introduction
Fever is a defensive response of the host to infection and is caused by exposure to pyrogens, which cause the secretion of inflammatory cytokines, including tumor necrosis factor α (TNF-α) and interleukin-1β, from polymorphonuclear leukocytes (Roth and Blatteis, 2014). The role of prostaglandin E2 (PGE2) in this febrile response has been studied in mice lacking cyclooxygenase-2 (COX-2/PTGS2; Li et al., 1999; Steiner et al., 2005), microsomal PGE synthase-1 (PTGES; Engblom et al., 2003), or EP3 receptor (PTGER3; Oka et al., 2003). These mice showed no temperature increase after systemic challenge with pyrogens such as lipopolysaccharide (LPS); therefore, PGE2 is considered a key mediator of the febrile response.
In rodents, the response to LPS is polyphasic, suggesting that more than one mechanism is involved. Inflammatory cytokines induce the synthesis of Cox-2 protein in brain endothelial cells (Cao et al., 1995; Laflamme et al., 1999; Engström et al., 2012), but this process is relatively slow, and the body core temperature (Tc) rises quickly in animals given intravenous injections of LPS; therefore, the febrile response at the early stage may require a rapid supply of PGE2 across the blood–brain barrier (BBB; Steiner et al., 2006). Moreover, the increase in Tc is correlated with the PGE2 level in CSF. PGE2 does not undergo catabolism in the brain of adult mammals, because the level of 15-hydroxyprostaglandin dehydrogenase (HPGD) expression in the brain is low (Nakano et al., 1972; Alix et al., 2008). Hence, extraction of PGE2 from CSF to blood is another determinant of Tc (Schuster, 2015). These facts suggest a role of membrane transporters in PGE2 transfer across the BBB and elimination from CSF.
Organic anion transporters (Oat), organic anion transporting polypeptides (Oatp), and multidrug resistance associated protein (Mrp) can transport PGE2 in brain capillary endothelial cells (BCECs) of rats (Akanuma et al., 2011). Among them, OATP2A1/SLCO2A1, also known as prostaglandin transporter, is a high-affinity carrier for prostanoids (Kanai et al., 1995) and plays a major role in PGE2 disposition in vivo (Chang et al., 2010; Nakanishi et al., 2015), as well as in the febrile response (Ivanov et al., 2003; Ivanov and Romanovsky, 2004). Oatp2a1 is expressed in primary cultures of rat cerebral endothelial cells (Kis et al., 2006) and is also expressed at the blood–CSF boundary, such as in the apical membranes of the choroid plexus (CP) in mice (Tachikawa et al., 2012) and subarachnoidal blood vessels in rats (Hosotani et al., 2015). Thus, it presumably facilitates PGE2 elimination from CSF into blood. OATP2A1 is also localized in neurons, microglia, and astrocytes in the human brain (Choi et al., 2008). We have shown that OATP2A1 is involved in PGE2 secretion from several types of cells including macrophages (Shirasaka et al., 2013; Shimada et al., 2015; Kasai et al., 2016). Since microglia are brain-resident macrophages, they could be a source of PGE2 associated with brain inflammation and the febrile response.
In the present study, we aimed to evaluate the role of OATP2A1 in PGE2 disposition in the brain and to establish whether it contributes to the development of the febrile response. For this purpose, we challenged globally Slco2a1-deficient (Slco2a1−/−) and monocyte-/macrophage-specific Slco2a1-knock-out (Slco2a1Fl/Fl/LysMCre/+) mice with LPS and monitored the resulting changes of colorectal temperature and PGE2 concentration in brain interstitial fluid (ISF) in the preoptic/anterior hypothalamic area. Our results indicate that OATP2A1 does influence the local disposition of PGE2 in the brain during the febrile response.
Materials and Methods
Materials.
LPS from Escherichia coli O111:B4 was purchased from Sigma-Aldrich. Paraformaldehyde and suramin sodium were purchased from Wako Pure Chemical Industries. All other compounds and reagents were commercial products of reagent grade. Microdialysis guide cannulas and probes were purchased from CMA Microdialysis. Anti-mouse Oatp2a1 guinea pig antibody was a gift from Prof. Ken-ichi Hosoya (University of Toyama, Toyama, Japan).
Animals.
Slco2a1+/+ and Slco2a1−/− mice used in this study had a mixed B6;129 genetic background. Slco2a1−/− mice were generated in our laboratory (Nakanishi et al., 2015; Gose et al., 2016). Slco2a1Fl/Fl/LysMCre/+ mice were generated by crossing Slco2a1Fl/Fl mice, in which Slco2a1 exon 1 is flanked by loxP sequences, with Lysozyme (Lys) MCre/+ mice, obtained from RIKEN Bioresource Center (RRID:IMSR_RBRC02302). Genotypes were confirmed by PCR using the following primers: for Slco2a1 exon 1: sense, 5′- AGGACCTGATAGGCAGCCAA-3′ and antisense, 5′- CACAGCAGAGACCCAACAGA-3′; for Lysozyme M-Cre: sense, 5′- GCATTGCAGACTAGCTAAAGGCAG-3′ and antisense, 5′- CCCAGAAATGCCAGATTACG-3′. Male mice (weighing 32.1 ± 5.6 g) and female mice (weighing 24.5 ± 4.3 g) age 8–49 weeks were used.
Male Wistar rats weighing 170–280 g were purchased from Japan SLC. All animals were allowed ad libitum access to food and water under a standard 12 h light/dark cycle (8:45 A.M. to 8:45 P.M. light) in a temperature- and humidity-controlled room (24 ± 1°C, 55 ± 5%). All animal experiments were performed in accordance with the requirements of the Kanazawa University Institutional Animal Care and Use Committee (permit numbers AP-143148, AP-153511, AP-163750, and AP-183957).
LPS injection and measurement of colonic temperature in mice.
Mice were administered LPS as a single intraperitoneal dose of 100 μg/kg in 150 μl of physiological saline, and changes of Tc were monitored by measuring the colonic temperature at 60 min intervals with a rectal thermometer, TD320 (Shibaura Electronics) or KN-91 (Natsume Seisakusho). The effect of handling stress on Tc was assessed at 30 min after intraperitoneal injection, as reported previously (Romanovsky et al., 2005). Each mouse was given injections of LPS between 11:15 and 11:25 A.M. to avoid the effect of diurnal fluctuation of thermoregulation. After a rest period of at least 1 week, basal Tc was measured in the same mice after injection of the vehicle alone. At 5 h after injection, the mice were exsanguinated under deep anesthesia (with pentobarbital), and brain tissues were excised for qRT-PCR, Western blotting, or immunohistochemistry. The hypothalamus and cerebral cortex tissue were collected according to Glowinski's method (Glowinski and Iversen, 1966).
Effect of suramin on Tc in rats.
Anesthetized rats were placed on a stereotaxic instrument. A hole was drilled in the skull, and a steel cannula (outside diameter, 0.5 mm; inner diameter, 0.26 mm) was implanted close to the lateral hypothalamus and fixed to the skull with dental cement. The coordinates for the hypothalamus were −0.1 mm anterior to bregma, 0.6 mm lateral, and 7.0 mm ventral to the skull (Paxinos and Watson, 2013). After 1 week, rats were given LPS (100 μg/kg, i.p., in 500 μl of physiological saline), and the colonic temperature was measured with a rectal thermometer (KN-91). A 30 gauge needle connected to a microsyringe was inserted into the implanted steel cannula 1 h before the LPS injection, and 2 μl of vehicle or suramin solution in physiological saline (700 μm) was injected using a microsyringe pump (CMA102; CMA Microdialysis AB) over 2 min.
Microdialysis.
Anesthetized mice were placed on a stereotaxic instrument. A hole was drilled in the skull, and a microdialysis guide cannula (CMA7 guide cannula) was implanted close to the lateral hypothalamus and fixed to the skull with dental cement. The coordinates for the hypothalamus were 0.9 mm anterior to bregma, 0.1 mm lateral, and 4.5 mm ventral to the skull (Paxinos and Franklin, 2012). The next day, a microdialysis probe (CMA7; membrane length, 1 mm; CMA Microdialysis) was inserted into the hypothalamus. Five days later, Krebs-Ringer phosphate buffer (KRPB; 120 mm NaCl, 2.4 mm KCl, 1.2 mm CaCl2, 1.2 mm MgSO4, 0.9 mm NaH2PO4, 1.4 mm Na2HPO4, pH 7.4) was perfused through the probe at a constant flow rate of 2 μl/min. After LPS treatment (100 μg/kg, i.p.), the dialysate was collected every 1 h. The experimental conditions were such that the animals could move freely. PGE2 in the dialysate was determined by liquid chromatography-mass spectrometry/mass spectrometry (LC-MS/MS) as described in our previous report (Gose et al., 2016). At the end of the experiment, the mice were exsanguinated under deep anesthesia, the microdialysis probe was removed, and brain tissues were collected. Placement of the inserted probe near the hypothalamus was confirmed by histological examination. The isolated probe was immediately placed in KRPB containing 10 ng/ml PGE2, and KRPB was perfused at the flow rate of 2 μl/min for 1 h. The in vitro relative recovery (RR) was determined as follows: RR (%) = Cd,vitro/Cr × 100, where Cd,vitro is the dialysate concentration of PGE2 and Cr is the reservoir concentration of PGE2 in KRPB. The concentration of PGE2 in the interstitial fluid (Cisf) was approximated by dividing the PGE2 concentration in the dialysate collected from mice (Cd,vivo) by RR, as follows: Cisf = Cd,vivo/RR.
PGE2 quantification.
CSF, blood, hypothalamic tissue, and cerebral cortex tissue were collected under anesthesia from mice at 1 and/or 5 h after LPS injection (100 μg/kg or 20 mg/kg, i.p.). Blood was centrifuged at 2000 × g at 4°C for 10 min to obtain plasma. The collected samples were frozen and stored at −80°C until use. Brain tissues were homogenized by sonication (QSonica). Eicosanoids were extracted from the homogenate (10 mg), from CSF (5 μl), or from plasma (200 μl) containing d4-PGE2 as an internal standard with a solid-phase cartridge (Oasis MAX; Waters) according to the manufacturer's protocol. The eicosanoids were eluted with 0.1% formic acid/acetonitrile and quantified with an LC-MS/MS system consisting of an LCMS8050 triple quadrupole mass spectrometer (Shimadzu) coupled with an LC-30AD ultra-fast liquid chromatography system (Shimadzu). The flow rate of the mobile phase was 0.4 ml/min, and the injection volume was set as 30 μl. CAPCELL PAK IF2 (C18, 2.1 mm inner diameter × 100 mm, 2 μm; Shiseido) was used as an analytical column. Samples were kept at 4°C during the analysis. Electrospray negative ionization was used, and the mass transitions were monitored at m/z 351.1/271.1 for PGE2 and m/z 355.1/275.2 for d4-PGE2. Analyst software Lab Solution LCMS was used for data manipulation.
Immunohistochemistry.
Excised brain tissues were fixed overnight with 4% paraformaldehyde at 4°C. Frozen and paraffin-embedded sections were prepared from mouse brains. The sections were treated with 5% nonfat dry milk in PBS, pH 7.4, or Blocking One Histo (Nacalai Tesque) for 1 h or 10 min, respectively, at room temperature and incubated overnight at 4°C with guinea pig anti-Oatp2a1 IgG (1:150 dilution), rabbit anti-Oatp2a1 IgG (1:400 dilution; Bioss Antibodies), rat anti-F4/80 IgG (1:100 dilution; AbD Serotec; RRID:AB_2098196), rat anti-CD34 IgG (1:50 dilution; BD Biosciences; RRID:AB_395015), rabbit anti-Cox-2 IgG (1:100 dilution; Cell Signaling Technology; RRID:AB_2571729), or rabbit anti-Cox-1 IgG (1:100 dilution; Cell Signaling Technology; RRID:AB_10860249). The sections were further incubated with biotinylated secondary antibodies (1:200–400 dilution) and with horseradish peroxidase-conjugated streptavidin (Thermo Fisher Scientific). After 3,3′-diaminobenzidine staining, the sections were examined under a light microscope. For fluorescence immunohistochemistry, the primary antibody was labeled with the appropriate IgG linked to Alexa Fluor 594 or Alexa Fluor 488 (Life Technologies). Nuclei were counterstained with Hoechst 33342 (Life Technologies), and the sections were mounted with Vectashield (Vector Laboratories). The sections were examined with an LSM710 confocal laser microscope (Carl Zeiss) or a conventional BZ-9000 fluorescence microscope (Keyence).
Oatp2a1 was quantified in brain sections of Slco2a1Fl/Fl and Slco2a1FlFl/LysMCre/+, and results are shown as mean values of 32 randomly selected images from four independent mice in each group. Glial cells, including microglia and macrophage-derived perivascular cells, were identified based on F4/80 immunofluorescence.
qRT-PCR.
Total RNA was extracted from tissues and cells of interest using RNAiso Plus (Takara Bio) and reverse-transcribed with a high-capacity cDNA reverse transcription kit (Applied Biosystems). mRNA expression levels of murine Slco2a1, Cox-2/Ptgs2, prostaglandin E synthase (mPges-1/Ptges), Ep3/Ptger3, and C-X-C motif chemokine 10 (Cxcl10) were quantified by qRT-PCR with the following primers: for Slco2a1: sense, 5′-GGACGGTGCCCATTCAGCCA-3′ and antisense, 5′-AGGTTCACTGTAGCCGTGTCCA-3′; for Ptgs2: sense, 5′-ATGAGTACCGCAAACGCTTC-3′ and antisense, 5′-TCTGGACGAGGTTTTTCCAC-3′; for Ptges: sense, 5′-GCACACTGCTGGTCATCAAG-3′ and antisense, 5′-ATGAGTACACGAAGCCGAGG-3′; for Ptger3: sense, 5′-ACCATCAAAGCCCTGGTGTC-3′ and antisense, 5′-TGTGTCTTGCATTGCTCAACC-3′; for Cxcl10: sense, 5′-CAAGTGCTGCCGTCATTTTC-3′ and antisense, 5′-TCAACACGTGGGCAGGATAG-3′. mRNA expression was normalized with respect to hypoxanthine-guanine phosphoribosyltransferase (Hprt) or glyceraldehyde 3-phosphate dehydrogenase (Gapdh) expression as described in our previous report (Nakanishi et al., 2015). Ct values of <31 in the real-time PCR analysis were considered as representing significant expression of mRNA.
Western blot analysis.
Western blot analysis was performed as described previously (Nakanishi et al., 2015). Excised hypothalamic tissue was homogenized by sonication (QSonica) and lysed in RIPA buffer (150 mm sodium chloride, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, and 50 mm Tris, pH 8.0) containing protease inhibitor cocktail (Nacalai Tesque). Samples were centrifuged at 10,000 × g at 4°C for 10 min, and protein concentration was determined with a Bio-Rad protein assay kit. An aliquot of tissue homogenate (20 μg) was separated by SDS-PAGE (10%) and electrotransferred onto polyvinylidene difluoride membrane (EMD Millipore). The membrane was incubated with 5% nonfat dry milk in TBS-T [Tris-buffered saline containing 0.1% (v/v) Tween 20] for 1 h at room temperature; incubated overnight at 4°C with primary antibody against Cox-2 and Cox-1 (each 1:500 dilution; Cell Signaling Technology), 15-Pgdh (1:400 dilution; Cayman Chemical; RRID:AB_327883), or Gapdh (1:5000 dilution; Cell Signaling Technology); and further incubated with horseradish peroxidase-conjugated secondary antibody (1:1000–5000 dilution; Thermo Fisher Scientific) for 1 h at room temperature. Bands were detected by electrochemical luminescence assay (Wako Pure Chemical Industries) and quantified using a CS analyzer (ATTO).
Quantitative PCR for genomic DNA.
Genomic DNA was isolated from excised tissues and peritoneal macrophages (collected from thioglycollate-treated mice) using phenol/chloroform/isoamyl alcohol (Wako Pure Chemical Industries). Slco2a1 allele was quantified by quantitative PCR using the following primers: sense, 5′-CCGCTCGGTCTTCAACAACA-3′ and antisense, 5′-ACTGGCCTCATCGACGTCCTA-3′.
Serum levels of inflammatory cytokines in mice.
Approximately 100 μl of blood was drawn from a tail vein at 2 h after LPS injection (100 μg/kg), allowed to clot at room temperature for 2 h, and centrifuged at 2000 × g at 4°C for 20 min to obtain serum. TNF-α was determined with an immunoassay kit (R&D Systems) according to the manufacturer's protocol. Optical density at 420 nm was measured with a micro-plate reader, ARVO MX (PerkinElmer Life Sciences). The procedure was repeated after injection of the vehicle alone in the same mice 1–4 weeks later.
Experimental design and statistical analysis.
Mice were assigned to experimental groups so that the groups were well matched for age and sex. The numbers of animals used in this study are given in the figure legends. All data are expressed as mean ± SEM of at least three experiments, as indicated, and statistical analysis was performed with Student's t test. The criterion of significance was p < 0.05.
Results
Impact of genetic deletion or pharmacological inhibition of Oatp2a1 on LPS-induced febrile response
To examine the effect of loss of Slco2a1 on the increase of Tc induced by LPS (100 μg/kg), we monitored colonic temperature in Slco2a1+/+ and Slco2a1−/− mice. There was no significant difference between the Tc values of Slco2a1+/+ and Slco2a1−/− mice treated with physiological saline (PS) at any time point examined, suggesting that OATP2A1 does not affect the circadian rhythm of basal body temperature. In LPS-treated Slco2a1+/+ mice, Tc began to increase at 2 h and reached a plateau by 5 h, whereas the Tc values of LPS-treated Slco2a1−/− mice remained close to basal levels (Fig. 1A). The difference in Tc between LPS-treated Slco2a1+/+ and Slco2a1−/− mice was greatest at 7 h, reaching 0.7°C (37.8°C vs 37.1°C, p = 0.009; Fig. 1A).
Contribution of OATP2A1 to the febrile response after LPS challenge. A, Tc was monitored in Slco2a1+/+ (open) and Slco2a1−/− (filled) mice treated with LPS (100 μg/kg; square) or vehicle alone (circle). n = 10 for each group. B, Tc was monitored in rats treated with LPS (100 μg/kg, i.p.; square) or PS (circle) and microinjected to the hypothalamus with suramin (filled) or PS (open). n = 4 for LPS/PS; n = 3 for LPS/suramin; n = 8 for PS/PS. Each point represents the mean value ± SEM. The asterisk indicates statistically significant differences (p < 0.05) between Slco2a1+/+:LPS and Slco2a1−/−:LPS (A) and between LPS/PS and LPS/suramin (B) by Student's t test.
We also examined the effect of intracerebral injection of suramin, which is a high-affinity inhibitor of OATP2A1 (Kamo et al., 2017), in rats. No significant increase in Tc was observed in the PS/PS control group, whereas in the LPS/PS group, Tc continued to rise for 5 h (Fig. 1B). In the LPS/suramin group, Tc rose initially and then fell to the basal level at 5 h. Rats in the LPS/suramin group exhibited significantly lower Tc (37.3°C) compared with those of the LPS/PS group (38.5°C) at 6 h after LPS injection (p = 0.001; LPS/PS vs LPS/suramin).
Expression of PGE2 signaling-related and inflammatory genes in the hypothalamus
To examine whether Slco2a1 deficiency affects the inflammatory response, we measured the mRNA levels of the inducible genes Ptgs2 and Cxcl10; Cxcl10 is an established inflammation marker in the hypothalamus (Wilhelms et al., 2014). The mRNA expression levels of both genes were significantly increased at 1 h after LPS treatment in both Slco2a1+/+ (Ptgs2, p = 0.002; Cxcl10, p = 0.006) and Slco2a1−/− (Ptgs2, p = 0.017; Cxcl10, p = 0.028; Fig. 2A,B). Cxcl10 mRNA was further elevated at 5 h. However, there was no significant difference in expression of these mRNAs between Slco2a1+/+ and Slco2a1−/− mice at any time during the test period, implying that both strains respond similarly to the inflammatory stimulus. In addition, no significant change was detected in the expression of Cox-2 (p = 0.419; Fig. 2C), Cox-1 (p = 0.745; Fig. 2D), or 15-Pgdh (p = 0.801; Fig. 2E) in the hypothalamus between Slco2a1+/+ and Slco2a1−/− mice at 5 h after LPS treatment. Loss of Slco2a1 also did not affect the mRNA expression of Ptges (p = 0.282; Fig. 2F) or Ptger3 (p = 0.823; Fig. 2G). These results suggest that the influence of Slco2a1 on the increase of Tc is not mediated by changes in the expression of any of these genes (Fig. 1A).
Effect of LPS on expression of PGE2 signaling-related and inflammatory genes in Slco2a1−/− mice. A, B, mRNA expression of Ptgs2 and Cxcl10 in hypothalamic tissue of LPS (100 μg/kg)-treated Slco2a1+/+ (n = 5–8) and Slco2a1−/− (n = 6–8) mice. C–G, Cox-2, Cox-1, and 15-Pgdh expression evaluated by Western blotting (C–E) and mRNA expression of Ptges and Ptger3 evaluated by qRT-PCR (F, G) in hypothalamic tissue from LPS-treated mice at 5 h after LPS injection (100 μg/kg). Band intensities of interest were normalized with that of Gapdh as shown in bar graphs at the bottom (C–E). Each bar represents the mean ± SEM (3 independent animals). The asterisk and dagger indicate statistically significant differences (p < 0.05) from LPS 0 h in Slco2a1+/+ and Slco2a1−/− mice by Student's t test.
Brain disposition and plasma concentration of PGE2
Since the increase in Tc appears to be associated with raised PGE2 levels in CSF, we measured the PGE2 concentration in the CSF of Slco2a1+/+ (Ccsf,WT) and Slco2a1−/− (Ccsf,KO) mice. The PGE2 concentration was markedly increased at 1 h after LPS injection and reached 1118 and 1182 pg/ml, respectively (Fig. 3A). Ccsf,WT was decreased to the basal level at 5 h (316 pg/ml); however, Ccsf,KO remained almost unchanged (1160 pg/ml), being 3.7-fold higher than Ccsf,WT at 5 h (p = 0.048; Fig. 3A). Since Ccsf,KO was unrelated to changes in Tc of Slco2a1−/− mice (Fig. 1A), we measured the PGE2 concentration in ISF at the hypothalamus (Cisf) by means of a microdialysis technique. There was no significant difference in the RR of PGE2 between Slco2a1+/+ (1.8 ± 0.2%) and Slco2a1−/− (2.0 ± 0.2%, p = 0.389) mice. The time course of Cisf in Slco2a1+/+ mice (Cisf,WT) was slightly biphasic. The maximum PGE2 concentration of Slco2a1+/+ mice (Cisf,max,WT) was 841 pg/ml, and Cisf,WT was significantly higher than Cisf of Slco2a1−/− mice (Cisf,KO) for 5 h after LPS injection (p < 0.05; Fig. 3B). In contrast, Cisf,KO was only slightly increased in Slco2a1−/− mice, with Cisf,max,KO of 380 pg/ml, which was not significantly different from the basal level (Cisf,KO,5 h, 246 pg/ml, p = 0.213). Moreover, the PGE2 concentration in the hypothalamus (Chyp) was similar in Slco2a1+/+ and Slco2a1−/− mice; Chyp,WT and Chyp,KO were 12.9 and 12.1 pg/mg tissue, respectively (p = 0.833; Fig. 3C). Chyp,WT was unaffected by LPS treatment, whereas Chyp,KO was slightly decreased at 1 h (6.5 pg/mg tissue) but was similar to Chyp,WT at 5 h (9.9 pg/mg tissue). Next, the PGE2 concentration in the cerebral cortex (Ccor) was determined. Basal Ccor,WT (13.1 pg/mg tissue) and Ccor,KO (14.3 pg/mg tissue) were similar to basal Chyp,WT (12.9 pg/mg tissue) and Chyp,KO (12.1 pg/mg tissue), respectively. As with Chyp, LPS (100 μg/kg) had no marked effect on Ccor (Fig. 3D); however, a higher dose of LPS (20 mg/kg) elevated Ccor,5 h and Chyp,5 h despite the Slco2a1 deficiency (Fig. 3C,D), in agreement with a previous report (Nomura et al., 2011). These results suggest that Slco2a1+/+ and Slco2a1−/− mice have similar abilities to produce PGE2. Finally, the PGE2 plasma concentration (Cp) in Slco2a1−/− mice (201 pg/ml) was significantly higher than that of Slco2a1+/+ mice (54 pg/ml) at 1 h after LPS treatment (p = 0.007; Fig. 3E).
Effect of Slco2a1 deficiency on PGE2 disposition in the brain. A, Time course of Ccsf in Slco2a1+/+ (open, n = 3–4) and Slco2a1−/− (filled, n = 3–4) mice treated with LPS (100 μg/kg, i.p.). B, Time course of Cisf in Slco2a1+/+ (open, n = 7) and Slco2a1−/− (filled, n = 6) mice treated with LPS (100 μg/kg, i.p.). C, D, Chyp (C) and Ccor (D) in Slco2a1+/+ (open, n = 3–6) and Slco2a1−/− (filled, n = 3–6) mice treated with LPS (100 μg/kg or 20 mg/kg, i.p.). Tissue concentration was normalized by wet weight of tissue. E, Cp in Slco2a1+/+ (open, n = 4) and Slco2a1−/− (filled, n = 4) mice at 1 h after LPS treatment (100 μg/kg, i.p.). Each point or bar represents the mean ± SEM. The asterisk indicates statistically significant differences (p < 0.05) by Student's t test.
Immunostaining of Oatp2a1 in mouse brain
Oatp2a1 localization in the brain was studied by means of immunohistochemistry in mice challenged with LPS (100 μg/kg). As previously reported, Oatp2a1 was detected in the epithelial cells of the CP (Fig. 4A) in Slco2a1+/+ mice. Parenchymal cells, including glial cells and nerve cells, in the region of the hypothalamus were stained (Fig. 4B), and immunoreaction was also detected around blood vessels and at the axial fibers of nerve cells in the cerebral cortex (Fig. 4C,D). Staining was negligible in Slco2a1−/− mice, confirming the specificity of the antibody for Oatp2a1 (Fig. 4A–D). Furthermore, Oatp2a1 was detected in BCECs of Slco2a1+/+ mice but not in BCECs of Slco2a1−/− mice (Fig. 4E). Since perivascular macrophages and endothelial cells are major sources of PGE2 in the brain (Saper et al., 2012), we examined the colocalization of Oatp2a1 with macrophage marker F4/80 and endothelial cell marker CD34 in the perivascular region by means of fluorescence immunohistochemistry of frozen brain sections. Fluorescence attributable to anti-Oatp2a1 antibody was partially colocalized with that caused by anti-CD34 antibody (Fig. 4F), as well as with that attributable to F4/80 in perivascular cells (Fig. 4G) and in glial cells (presumably microglia; Fig. 4H). Oatp2a1 was not detected in corresponding sections derived from Slco2a1−/− mice (Fig. 4F–H). Cox-2 staining was detected not only along blood vessels (Fig. 4I) but also on brain parenchymal cells (Fig. 4J) in mice challenged with LPS (100 μg/kg). No staining was observed in the absence of the primary antibody (Fig. 4I,J). Moreover, Cox-1/Ptgs1 was detected in F4/80-positive cells in the hypothalamus (Fig. 4K).
Localization of Oatp2a1 and Cox in mouse brain. A–E, Immunostaining of Oatp2a1 in the CP region (A), parenchyma of the hypothalamus (B, C), cerebral cortex (D), and capillary endothelial cells of the hypothalamus (E) in paraffin-embedded sections of Slco2a1+/+ and Slco2a1−/− mice. F–H, Dual immunostaining of Oatp2a1 and CD34 (blood vessels; F) or F4/80 [blood vessels (G) and parenchymal cells (H)] in frozen brain sections of Slco2a1+/+ and Slco2a1−/− mice. I, J, Cox-2 immunostaining in paraffin-embedded sections of the hypothalamus of Slco2a1+/+ mice showing blood vessels (I) and brain parenchymal cells (J). K, Dual immunostaining of Cox-1 and anti-F4/80 in frozen sections of brain parenchyma. Nuclei were counterstained with hematoxylin (A–E, I, J) or Hoechst33342 (F–H, K). Sections were prepared from at least three individual Slco2a1+/+ and Slco2a1−/− mice at 4 h after LPS treatment (100 μg/kg). Arrows indicate expression of the indicated protein.
Contribution of Oatp2a1 in macrophages to the LPS-induced febrile response
To assess the contribution of Oatp2a1 in F4/80-positive cells to the febrile response, we used Slco2a1Fl/Fl/LysMCre/+ mice. In Slco2a1Fl/Fl/LysMCre/+-derived peritoneal macrophages, floxed alleles of Slco2a1 were reduced to ∼10% of those in Slco2a1Fl/Fl mice (p = 0.001), whereas no significant change was observed in other tissues (Fig. 5A). Slco2a1 mRNA expression was significantly decreased in peritoneal macrophages (p = 0.006) and spleen (p = 0.036) of Slco2a1Fl/Fl/LysMCre/+ mice compared with Slco2a1Fl/Fl mice (Fig. 5B). Reduced Oatp2a1 protein expression was confirmed in F4/80-positive cells by means of fluorescence immunohistochemistry (Fig. 5C). The ratio of Oatp2a1-positive cells to F4/80-positive cells was significantly reduced in brain sections of Slco2a1Fl/Fl/LysMCre/+ mice (0.29 ± 0.04) compared with Slco2a1Fl/Fl mice (0.71 ± 0.06, p < 0.001). In contrast, Oatp2a1 staining was observed in axial fibers of nerve cells and in the CP of Slco2a1Fl/Fl/LysMCre/+ mice (Fig. 5D,E). Therefore, the effect of Slco2a1 deficiency in F4/80-positive cells on febrile response was further examined in LPS-injected Slco2a1Fl/Fl/LysMCre/+ mice.
Expression of Oatp2a1 in macrophages of Slco2a1Fl/Fl/LysMCre/+ mice. A, B, Slco2a1 allele (A) and mRNA expression of Slco2a1 (B) in tissues and peritoneal macrophages of Slco2a1Fl/Fl and Slco2a1Fl/Fl/LysMCre/+ mice. Each bar represents the mean with SEM of three mice (for abundance of Slco2a1 allele), eight mice for Slco2a1Fl/Fl, and four mice for Slco2a1Fl/Fl/LysMCre/+ mice (for mRNA expression). The asterisk indicates statistically significant differences (p < 0.05) by Student's t test. C, Dual immunostaining of Oatp2a1 and F4/80 in frozen brain sections of Slco2a1Fl/Fl and Slco2a1Fl/Fl/LysMCre/+ mice at 4 h after LPS treatment (100 μg/kg). D, E, Oatp2a1 immunostaining of paraffin-embedded sections of Slco2a1Fl/Fl/LysMCre/+ mice. 3,3′-Diaminobenzidine staining of nerve cells (D) and the CP region (E). Nuclei were counterstained with Hoechst33342 (C) or hematoxylin (D, E). Arrows indicate expression of the indicated protein. The illustrations are representative of the results of at least three individual mice in each group.
The time courses of Tc were similar in PS-injected Slco2a1Fl/Fl (control group) and Slco2a1Fl/Fl/LysMCre/+ mice, and no significant change in Tc was observed during the test period. On the other hand, Tc in Slco2a1Fl/Fl mice was elevated at 2 h and reached a plateau at 5 h after LPS treatment (35.9°C; Fig. 6A), which is similar to the result in Slco2a1+/+ mice (Fig. 1A). In Slco2a1Fl/Fl/LysMCre/+ mice, Tc was increased at 2 h but did not reach the level of Slco2a1Fl/Fl mice. The Tc of Slco2a1Fl/Fl/LysMCre/+ mice was significantly lower than that of Slco2a1Fl/Fl at 6 h (p = 0.0044), 7 h (p = 0.010), and 9 h (p = 0.048) after LPS treatment (Fig. 6A). We further examined Cisf of Slco2a1Fl/Fl (Cisf,Fl) and Slco2a1Fl/Fl/LysMCre/+ (Cisf,Fl/LysM) mice using a microdialysis technique. Cisf,Fl and Cisf,Fl/LysM peaked at 2 h after LPS injection, reaching 1122 and 512 pg/ml, respectively. Cisf,Fl/LysM was significantly lower than Cisf,Fl throughout the test period (p < 0.05; Fig. 6B). No significant difference was observed in Cox-2 (p = 0.783), Cox-1 (p = 0.319), or 15-Pgdh (p = 0.340) protein expression, or in Ptges (p = 0.437) or Ptger3 (p = 0.540) mRNA expression, between Slco2a1Fl/Fl and Slco2a1Fl/Fl/LysMCre/+ mice (Fig. 6C–G).
Contribution of Oatp2a1 in macrophages to the febrile response after LPS challenge. A, Tc was monitored in Slco2a1Fl/Fl (open, n = 6 for PS and 10 for LPS) and Slco2a1Fl/Fl/LysMCre/+ (filled, n = 6 for PS and 10 for LPS) mice after treatment with LPS (100 μg/kg, i.p.; square) or vehicle alone (circle). B, Time course of Cisf in Slco2a1Fl/Fl (open, n = 4) and Slco2a1Fl/Fl/LysMCre/+ (filled, n = 4) mice treated with LPS (100 μg/kg, i.p.). C–G, Cox-2, Cox-1, and 15-Pgdh expression was evaluated by Western blotting (C–E) and mRNA expression of Ptges and Ptger3 was evaluated by qRT-PCR (F, G) in hypothalamic tissue from Slco2a1Fl/Fl and Slco2a1Fl/Fl/LysMCre/+ mice at a 5 h after injection of LPS (100 μg/kg). Band intensities were normalized with that of Gapdh as shown by bar graphs at the bottom (n = 3 for each group). Each bar represents the mean ± SEM. The asterisk indicates statistically significant differences (p < 0.05) from Slco2a1Fl/Fl and Slco2a1Fl/Fl/LysMCre/+ mice by Student's t test.
Serum concentration of TNF-α
Since TNF-α is a major pyrogenic cytokine, we measured its serum level in LPS (100 μg/kg)-treated Slco2a1-global knock-out mice. LPS increased the serum concentration of TNF-α in all groups, but the concentration in Slco2a1−/− mice (652 pg/ml) was significantly lower than that in Slco2a1+/+ mice (1084 pg/ml, p = 0.011; Fig. 7A). Similar results were obtained for serum TNF-α in Slco2a1Fl/Fl and Slco2a1Fl/Fl/LysMCre/+ mice (675 pg/ml vs 473 pg/ml, p = 0.048; Fig. 7B), suggesting that Oatp2a1 expressed in macrophages is associated with the release of TNF-α from activated inflammatory cells.
Effect of global and monocyte-/macrophage-specific Slco2a1 deficiency on serum levels of TNF-α. A, B, Serum levels of TNF-α of global (A) and monocyte-/macrophage-specific (B) Slco2a1-knock-out mice. Animals were treated with LPS (100 μg/kg, i.p.) or vehicle only, and serum was collected 2 h later. Four Slco2a1+/+ or Slco2a1−/− (A) and 10 Slco2a1Fl/Fl or Slco2a1Fl/Fl/LysMCre/+ (B) mice were used. Each result represents the mean ± SEM. N.D., Not detected. The asterisk indicates statistically significant differences (p < 0.05) by Student's t test.
Discussion
PGE2 is a major chemical mediator of the febrile response, and membrane transporters are thought to be involved in regulating its concentration in the hypothalamus, which plays a key role in regulating body temperature. However, the details remain unclear. Therefore, in the present study, we examined the role of a major prostaglandin transporter, OATP2A1/SLCO2A1, in the febrile response by using global and monocyte-/macrophage-specific Slco2a1-knock-out mice. Our findings indicate that OATP2A1 serves to maintain an increased interstitial PGE2 concentration in the hypothalamus during the febrile response.
Impact of OATP2A1 on PGE2 disposition and thermoregulation
The LPS-induced febrile response was attenuated in Slco2a1−/− mice (Fig. 1A). In accordance with this finding, the OATP2A1 inhibitor suramin attenuated the LPS-induced increase of Tc in rats (Fig. 1B). It has been shown that Ccsf is positively associated with Tc, and therefore we assessed Ccsf in these mice. As shown in Figure 3A, the time course of Ccsf,WT was consistent with changes in LPS-induced Tc in Slco2a1+/+ mice. Ccsf,KO rose as quickly as Ccsf,WT but then surprisingly remained high throughout the test period. Since Oatp2a1 is expressed in the CP (Fig. 4A), this result presumably reflects Oatp2a1-mediated PGE2 uptake at the apical side of CP (Tachikawa et al., 2012). Namely, high Ccsf,KO could be attributed to reduced clearance of PGE2 from the CSF. However, the negligible change of Tc in Slco2a1−/− mice cannot be explained by this observation. This led us to evaluate the kinetics of Cisf in free-moving mice by means of a microdialysis technique. Cisf ranged from 200 to 1200 pg/ml (Figs. 3B, 6B), which is similar to the reported ranges of PGE2 concentration in brain ISF of guinea pigs (20–1200 pg/ml; Sehic et al., 1996) and rabbits (200–3500 pg/ml; Chang et al., 2013) treated with LPS (2 μg/kg, i.v.). These observations suggest the surgical procedure may not have had a marked effect on Cisf levels in our experiment. In contrast to Ccsf,KO, Cisf,KO did not rise and remained at the basal level. Because ISF–CSF communication at the inner ventricle surface is supposed to occur freely, concentrations of solutes in brain ISF are generally equivalent to those in CSF (Lei et al., 2017). This was not the case in Slco2a1−/− mice (Cisf, 380 pg/ml; Ccsf, 1182 pg/ml). This suggests that independent mechanisms operate in the regulation of PGE2 concentration in the two regions.
Moreover, Chyp was estimated as ∼12,000 pg/ml in mice before LPS injection, and Cisf,max was, at most, 7% of this value in Slco2a1+/+ mice (Fig. 3B,C). The high tissue concentration of PGE2 could be attributed to a high tissue-bound fraction, although we could not determine the bound fraction in this study. Brain interstitial space corresponds to a few percent of the volume of brain parenchyma, and intracellular PGE2 concentration is not necessarily associated with Cisf. It was not established whether newly synthesized or pre-existing PGE2 mediates the febrile response. Considering the antipyretic effect of COX inhibitors, it seems likely that newly synthesized PGE2 is involved; however, intracerebrally injected suramin, which does not inhibit PGE2 production (Kamo et al., 2017), partially attenuated the rise of Tc in rats. Also, Slco2a1+/+ and Slco2a1−/− mice showed similar ability to respond to inflammation (Fig. 2). Therefore, we hypothesized that Cisf in the hypothalamus is regulated by OATP2A1-mediated transport, and we next examined which cells secrete PGE2.
Contribution of OATP2A1 in macrophages to development of the febrile response
Oatp2a1 was detected in many cells, including microglia, astrocytes, nerve cells, and BCECs (Fig. 4), in agreement with previous work (Kis et al., 2006; Choi et al., 2008). Astrocytes (Mishra et al., 2016) and nerve cells (Lacroix et al., 2015) in the brain constitutively synthesize PGE2. Moreover, Cox-2 is expressed not only in blood vessels but also in brain parenchyma (Fig. 4I,J). Perivascular macrophages and microglia secrete PGE2 under inflammatory conditions (Elmquist et al., 1997; Schiltz and Sawchenko, 2002). Previous studies, including ours, suggest that OATP2A1 is dominantly expressed in cytoplasm of astrocytes and macrophages (Gordon et al., 2008; Shimada et al., 2015). We recently showed that OATP2A1 facilitates PGE2 exocytosis from murine macrophages (Shimada et al., 2015). Moreover, the LPS-induced febrile response was significantly attenuated by monocyte-/macrophage-specific Slco2a1 deficiency (Fig. 6A), in association with decreased Cisf (Fig. 6B). Accordingly, these results suggest that OATP2A1 is involved in PGE2 secretion from macrophages during the febrile response. Cox-2 is not a determinant of LPS-induced fever generation in LysM-positive cells (Nilsson et al., 2017). However, Cox-1 was expressed in F4/80-positive cells (Fig. 4K), in agreement with the previous report of Cox-1 expression in microglia and perivascular cells (García-Bueno et al., 2009). There is also evidence of the involvement of microglial COX-1 in immune challenge-induced neuroinflammation in various models (Candelario-Jalil et al., 2007; García-Bueno et al., 2009). Therefore, we cannot rule out the possibility that PGE2 produced via Cox-1 and secreted from microglia and perivascular cells via OATP2A1 is essential for the febrile response. Another possibility is that the increase of Cisf under inflammatory conditions is attributable to altered OATP2A1 activity, since LPS-induced transcriptional upregulation of rat Oatp2a1 has been reported (Hosotani et al., 2015). Further studies are needed to establish the underlying mechanism.
Tc and Cisf were partially decreased in Slco2a1Fl/Fl/LysMCre/+ mice (Fig. 6A,B); therefore, Oatp2a1 in cells other than macrophages may also contribute to the febrile response. As the blood PGE2 concentration rises concomitantly with body temperature, the entry of peripherally synthesized PGE2 is suggested to trigger the response (Steiner et al., 2006; Wilhelms et al., 2014). Here, we found that Oatp2a1 is expressed in BCECs in vivo (Fig. 4E); this result is consistent with its expression at the apical membranes of primary-cultured rat BCECs (Kis et al., 2006). In contrast to the remarkable elevation of Cp in Slco2a1−/− mice (0.60 nm; Fig. 3E), basal Cisf was not affected (Fig. 3B), although OATP2A1 is unlikely to be saturated (Km for PGE2 is about 100 nm); therefore, our data suggest a role of OATP2A1 in PGE2 uptake by BCECs, as in endothelial cells of other tissues (Nakanishi et al., 2017; Nakanishi and Tamai, 2018). Thus, OATP2A1 may contribute to PGE2 transport into the brain.
OATP2A1 is a PGE2/lactate exchanger (Chan et al., 2002). Considering that the physiological lactate concentration in the brain is higher than that in plasma of mammals (Abi-Saab et al., 2002), it is reasonable to suppose that OATP2A1 facilitates PGE2 efflux from endothelial cells into the brain interstitial space. BCECs are accepted to be a major source of PGE2 during the febrile response (Yamagata et al., 2001; Engström et al., 2012); however, the mechanism of PGE2 release from these cells has been little explored. We speculate that Oatp2a1 contributes to the febrile response by participating in PGE2 transport across and/or secretion from BCECs in exchange for extracellular lactate. Furthermore, OATP2A1 may serve to maintain an increased level of Cisf during the febrile response.
Impact of Slco2a1 deficiency on serum TNF-α level and its effect on the febrile response
Serum TNF-α was also significantly increased by LPS in all test animals, but the increase was significantly smaller in Slco2a1−/− and Slco2a1 Fl/Fl/LysMCre/+ mice than in their respective counterparts (Fig. 7A,B). TNF-α induces protein expression of Cox-2 in the brain during the febrile response (Cao et al., 1998); nevertheless, our present data suggest that Slco2a1 deficiency did not affect the Cox-2 protein level in the hypothalamus after LPS injection (Figs. 2C, 6C). This is also supported by the observation that PGE2 concentration in the hypothalamus was similar in Slco2a1+/+ and Slco2a1−/− mice at 5 h after LPS treatment (Fig. 3C). Furthermore, it was reported that LPS-induced fever was not affected in rats given injections of a nonsteroidal anti-inflammatory drug, nimesulide, despite a substantial decrease in serum TNF-α (from 3237 to 564 pg/ml; Dogan et al., 2002). Therefore, the difference in TNF-α levels may have had little impact on the LPS-induced increase in Tc in our experiment.
In conclusion, our findings indicate that OATP2A1 mediates PGE2 secretion from F4/80-positive glial cells in the hypothalamus, as well as PGE2 transport across the BBB and secretion from BCECs. Thus, OATP2A1 appears to be a key determinant of the PGE2 concentration in hypothalamus interstitial fluid, which is the proximal mediator of the LPS-induced febrile response. We believe these findings on the multiple roles of OATP2A1 contribute to our understanding of the mechanisms of inflammatory response in the brain and may provide a clue to develop new therapeutic strategies for treating refractory neuroinflammatory diseases.
Footnotes
This work was supported by Grants-in-Aid for Scientific Research (KAKENHI, 15H04755 and 15K15181) from the Japan Society for the Promotion of Science. We thank Richard Steele for careful reading of this manuscript. We thank Drs. Ken-ichi Hosoya and Shin-ichi Akanuma at the University of Toyama for constructive comments and technical advice and for providing antibodies to detect Oatp2a1. We also thank Dr. Eiichi Hinoi and Maika Okamoto for assistance with measuring colorectal temperature.
The authors declare no competing financial interests.
- Correspondence should be addressed to Dr. Takeo Nakanishi, Faculty of Pharmaceutical Sciences, Institute of Medical, Pharmaceutical, and Health Sciences, Kanazawa University, Kakuma-machi, Kanazawa 920-1192, Japan. nakanish{at}p.kanazawa-u.ac.jp
