Refine
Year of publication
Document Type
- Article (47)
- Postprint (27)
- Conference Proceeding (2)
- Other (1)
- Preprint (1)
Language
- English (78) (remove)
Keywords
- NASH (6)
- inflammation (6)
- oxidative stress (6)
- cell-based assay (5)
- NAFLD (4)
- cholesterol (4)
- BoNT (2)
- BoNT/B uptake (2)
- Botulinum neurotoxin (2)
- E-2 (2)
Prostaglandin F(2alpha) (PGF(2alpha)), modulates hepatocyte functions via a heptahelical G(q)-coupled PGF(2alpha)-receptor (FP-R) which in liver is expressed exclusively in hepatocytes. The aim of the present study was to isolate the 5'-flanking region of the rat FP-R gene and to elucidate its basal and IL-6-modulated transcription control function in rat hepatocytes. The 5'-non-translated region of the rat hepatocyte FP-R mRNA differed from the corresponding region in rat fetal astrocyte or corpus luteum. It was encoded by exons 1a and 2 which were separated by a 1. 4 kb intron containing the exons 1b and 1c coding for the 5'-untranslated region of rat fetal astrocyte and corpus luteum FP-R mRNA, respectively. The transcription initiation site in hepatocytes was localized 263 bp upstream of the start ATG by 5'-RACE. A DNA-fragment covering the 5'-flanking region of the rFP-R gene from - 1 of the transcription initiation site to -2590 bp was cloned and sequenced. Its 3'-two thirds had a 65% sequence identity to the mouse FP-R promoter however no homology to the bovine FP-R promoter. In the overlapping sequence most of the putative transcription factor binding sites were conserved between mouse and rat. The rat promoter contained no classical TATA- or CAAT-boxes but putative binding sites for the transcription factors C/EBP, GATA-1, HNF-1, HNF-3beta, SP-1, and USF. Luciferase reporter gene constructs containing portions of the 5'-flanking region were transfected into rat hepatocytes. Luciferase expression ranked -181 >/= -608 < -1418 > -1821 >/= -2590. The strongest transcriptional activity was conferred by the region between -608 and -1418 containing a cluster of potential HNF-1 and HNF-3beta binding sites that might allow the exclusive expression of FP-R mRNA in hepatocytes. The amount of FP-R mRNA and the luciferase expression under control of the -2590 promoter fragment were reduced by IL-6 in hepatocytes. Copyright 2000 Academic Press.
Eight heptahelical receptors have been characterized for prostaglandin (PG) D(2), PGE(2), PGF(2alpha), prostacyclin and thromboxane A(2). They share a sequence identity of 40%. All of them have potential N-glycosylation sites. The current study analysed the role of the two N-glycosylation sites in the rat EP3beta-subtype PGE(2) receptor for protein folding and sorting. The N-glycosylation consensus sequences were eliminated by site-directed mutagenesis and receptors expressed in HEK-293 cells. Both potential N-glycosylation sites were used. Their joint elimination resulted in the synthesis of a receptor protein with full binding competence, biological activity and no reduction of affinity; however, the half-life of the non-glycosylated receptor was slightly reduced. Ligand binding to intact stably transfected cells and confocal laser microscopic immunocytochemistry showed that the glycosylated receptor was correctly inserted into the plasma membrane to a much larger extent than the non-glycosylated receptor, which tended to accumulate in the perinuclear zone of the endoplasmic reticulum. Inhibition of N-glycosylation with tunicamycin resulted in a similar perinuclear distribution of the wild-type receptor. Therefore, glycosylation of the EP3beta receptor seems not to be necessary for correct folding of the receptor protein but for the efficient transport of the receptor protein to the plasma membrane. This contrasts with a previous finding which described a reduction of the affinity for PGE(2) of the EP3alpha receptor by elimination of the distal glycosylation site when the receptor protein was expressed in insect cells.
Background/Aims: Prostaglandin E(2) (PGE(2)) is known to inhibit the lipopolysaccharide (LPS)-induced tumor necrosis factor alpha (TNFalpha) formation in Kupffer cells via an increase in cAMP. Four receptor-subtypes have been cloned for PGE(2) so far. Two of them, the EP2-receptor and the EP4-receptor are linked to stimulatory Gs-proteins and could mediate the inhibition by PGE(2) of TNFalpha-formation.Methods: The significance of both receptors for PGE(2)- dependent inhibition of LPS-induced TNFalpha-formation was studied using Kupffer cells of mice in which either one of the two receptors had been eliminated by homologous recombination.Results: The mRNAs of both receptors were expressed in wild type mouse Kupffer cells. Exogenous PGE(2) inhibited TNFalpha-formation in Kupffer cells lacking either EP2-receptor or EP4-receptor to a similar extent as in control cells, however, 10-fold higher PGE(2) concentrations were needed for half maximal inhibition in cells lacking the EP4-receptor than in control or EP2-receptor- deficient cells. The response to endogenous PGE(2) was blunted in EP4-receptor-deficient mice only and especially after prolonged incubation. Conclusions: The data indicate, that PGE(2) can inhibit TNFalpha-formation via both the EP2- and the EP4-receptor and that, however, the EP4-receptor appears to be physiologically more relevant in Kupffer cells since it conferred a high affinity response to PGE(2).
Prostanoid receptors belong to the class of heptahelical plasma membrane receptors. For the five prostanoids, eight receptor subtypes have been identified. They display an overall sequence similarity of roughly 30%. Based on sequence comparison, single amino acids in different subtypes of different species have previously been identified by site-directed mutagenesis or in hybrid receptors that appear to be essential for ligand binding or G-protein coupling. Based on this information, a series of mutants of the human FP receptor was generated and characterized in ligand- binding and second-messenger-formation studies. It was found that mutation of His-81 to Ala in transmembrane domain 2 and of Arg-291 to Leu in transmembrane domain 7, which are putative interaction partners for the prostanoid's carboxyl group, abolished ligand binding. Mutants in which Ser-263 in transmembrane domain 6 or Asp-300 in transmembrane domain 7 had been replaced by Ala or Gln, respectively, no longer discriminated between prostaglandins PGF(2alpha) and PGD(2). Thus distortion of the topology of transmembrane domains 6 and 7 appears to interfere with the cyclopentane ring selectivity of the receptor. PGF(2alpha)-induced inositol formation was strongly reduced in the mutant Asp-300Gln, inferring a role for this residue in agonist-induced G-protein activation.
Sensitization by interleukin-6 of rat hepatocytes to tumor necrosis factor alpha-induced apoptosis
(2003)
BACKGROUND/AIMS: Tumor necrosis factor (TNF) elicits hepatocyte apoptosis in toxic liver injury and is also central in hepatocyte proliferation after partial hepatectomy. In both circumstances interleukin (IL)-6 levels are also elevated. In mouse liver IL-6 attenuated Fas receptor-mediated apoptosis indicating its interference with pro-apoptotic signal chains. It was, therefore, the aim to examine the modulation by IL-6 of TNFalpha-induced apoptosis in rat hepatocytes. METHODS: Primary rat hepatocytes were treated with IL-6 prior to induction of apoptosis with TNFalpha/ actinomycin D or anti-Fas antibody M-20. Apoptosis was detected by determination of caspase-3 activation and bisbenzimide staining of condensed nuclei. Expression of TNFalpha receptors was analyzed by semi-quantitative polymerase chain reaction and ligand binding studies with [125I]-TNFalpha. RESULTS: IL-6 treatment doubled TNFalpha/actinomycin D- induced caspase-3 activity and significantly enhanced chromatin condensation. By contrast IL-6 inhibited Fas-induced increase in caspase-3 activity by 45% and significantly reduced chromatin condensation. IL-6 increased the mRNA level of TNF-R1 1.35-fold and augmented cell surface binding of [125I]-TNFalpha 3-fold. The latter and TNFalpha-mediated caspase activation was attenuated by prostaglandin E(2). CONCLUSIONS: IL-6 - in contrast to its anti-apoptotic modulation of the Fas-induced pathway - exerted a pro-apoptotic effect on the TNFalpha/actinomycin D-induced apoptosis by increasing the number of TNF-R on hepatocytes.
hEP4-R (human prostaglandin E2 receptor, subtype EP4) is a G(s)-linked heterotrimeric GPCR (G-protein-coupled receptor). It undergoes agonist-induced desensitization and internalization that depend on the presence of its C- terminal domain. Desensitization and internalization of GPCRs are often linked to agonist-induced beta-arrestin complex formation, which is stabilized by phosphorylation. Subsequently beta-arrestin uncouples the receptor from its G-protein and links it to the endocytotic machinery. The C-terminal domain of hEP4-R contains 38 Ser/Thr residues that represent potential phosphorylation sites. The present study aimed to analyse the relevance of these Ser/Thr residues for agonist- induced phosphorylation, interaction with beta-arrestin and internalization. In response to agonist treatment, hEP4-R was phosphorylated. By analysis of proteolytic phosphopeptides of the wild-type receptor and mutants in which groups of Ser/Thr residues had been replaced by Ala, the principal phosphorylation site was mapped to a Ser/Thr-containing region comprising residues 370-382, the presence of which was necessary and sufficient to obtain full agonist-induced phosphorylation. A cluster of Ser/Thr residues (Ser-389-Ser-390-Thr-391-Ser-392) distal to this site, but not the principal phosphorylation site, was essential to allow agonist-induced recruitment of beta-arrestin1. However, phosphorylation greatly enhanced the stability of the beta-arrestin1-receptor complexes. For maximal agonist-induced internalization, phosphorylation of the principal phosphorylation site was not required, but both beta-arrestin1 recruitment and the presence of Ser/Thr residues in the distal half of the C-terminal domain were necessary.
More than any other organ, the liver contributes to maintaining metabolic equilibrium of the body, most importantly of glucose homeostasis. It can store or release large quantities of glucose according to changing demands. This homeostasis is controlled by circulating hormones and direct innervation of the liver by autonomous hepatic nerves. Sympathetic hepatic nerves can increase hepatic glucose output; they appear, however, to contribute little to the stimulation of hepatic glucose output under physiological conditions. Parasympathetic hepatic nerves potentiate the insulin-dependent hepatic glucose extraction when a portal glucose sensor detects prandial glucose delivery from the gut. In addition, they might coordinate the hepatic and extrahepatic glucose utilization to prevent hypoglycemia and, at the same time, warrant efficient disposal of excess glucose.
1 Two isoforms of the rat prostaglandin E-2 receptor, rEP3 alpha-R and rEP3 beta-R, differ only in their C- terminal domain. To analyze the function of the rEP3-R C-terminal domain in agonist induced desensitization, a cluster of Ser/Thr residues in the C-terminal domain of the rEP3 alpha-R was mutated to Ala and both isoforms and the receptor mutant (rEP3 alpha-ST341-349A-R) were stably expressed in HEK293 cells. 2 All rEP3-R receptors showed a similar ligand- binding profile. They were functionally coupled to Gi and reduced forskolin-induced cAMP-formation. 3 Repeated exposure of cells expressing the rEP3 alpha-R isoform to PGE(2) reduced the agonist induced inhibition of forskolin-stimulated cAMP-formation by 50% and led to internalization of the receptor to intracellular endocytotic vesicles. By contrast, Gi- response as well as plasma membrane localization of the rEP3 beta-R and the rEP3 alpha-ST341-349A-R were not affected by prior agonist-stimulation. 4 Agonist-stimulation of HEK293-rEP3 alpha-R cells induced a time- and dose-dependent phosphorylation of the receptor most likely by G protein-coupled receptor kinases and not by protein kinase A or protein kinase C. By contrast, upon agonist-stimulation the rEP3 beta-R was not phosphorylated and the rEP3 alpha-ST341-349A-R was phosphorylated only weakly. 5 These results led to the hypothesis that agonist-induced desensitization of the rEP3 alpha-R isoform is mediated most likely by a GRK-dependent phosphorylation of Ser/Thr residues 341 - 349. Phosphorylation then initiates uncoupling of the receptor from Gi protein and receptor internalization
The human FP-R (F2alpha prostaglandin receptor) is a Gq-coupled heptahelical ectoreceptor, which is of significant medical interest, since it is a potential target for the treatment of glaucoma and preterm labour. On agonist exposure, it mediates an increase in intracellular inositol phosphate formation. Little is known about the structures that govern the agonist-dependent receptor activation. In other prostanoid receptors, the C-terminal domain has been inferred in the control of agonist-dependent receptor activation. A DRY motif at the beginning of the second intracellular loop is highly conserved throughout the G-protein-coupled receptor family and appears to be crucial for controlling agonist-dependent receptor activation. It is replaced by an ERC motif in the FP-R and no evidence for the relevance of this motif in ligand-dependent activation of prostanoid receptors has been provided so far. The aim of the present study was to elucidate the potential role of the C-terminal domain and the ERC motif in agonist-controlled intracellular signalling in FP-R mutants generated by site-directed mutagenesis. It was found that substitution of the acidic Glu(132) in the ERC motif by a threonine residue led to full constitutive activation, whereas truncation of the receptor's C-terminal domain led to partial constitutive activation of all three intracellular signal pathways that had previously been shown to be activated by the FP-R, i.e. inositol trisphosphate formation, focal adhesion kinase activation and T-cell factor signalling. Inositol trisphosphate formation and focal adhesion kinase phosphorylation were further enhanced by ligand binding in cells expressing the truncation mutant but not the E132T (Glu132-->Thr) mutant. Thus C-terminal truncation appeared to result in a receptor with partial constitutive activation, whereas substitution of Glu132 by threonine apparently resulted in a receptor with full constitutive activity.
A tight hormonal control of energy homeostasis is of pivotal relevance for animals. Recent evidence suggests an involvement of the nuclear receptor NR1i3 (CAR). Fasting induces CAR by largely unknown mechanisms and CAR-deficient mice are defective in fasting adaptation. In rat hepatocytes CAR was induced by WY14643, a PPARalpha-agonist. A DR1 motif in the CAR promoter was necessary and sufficient for this control. The PPARalpha-dependent increase in CAR potentiated the phenobarbital-induced transcription of the prototypical CAR-dependent gene CYP2B1. Since free fatty acids are natural ligands for PPARalpha, a fasting-induced increase in free fatty acids might induce CAR. In accordance with this hypothesis, CAR induction by fasting was abrogated in PPARalpha-deficient mice.
Human anaphylatoxin C3a had previously been shown to increase glycogenolysis in perfused rat liver and prostanoid formation in rat liver macrophages. Surprisingly, human C5a, which in other systems elicited stronger responses than C3a, did not increase glycogenolysis in perfused rat liver. Species incompatibilities within the experimental system had been supposed to be the reason. The current study supports this hypothesis: (1) In rat liver macrophages that had been maintained in primary culture for 72 h recombinant rat anaphylatoxin C5a in concentrations between 0.1 and 10 pg/ml increased the formation of thromboxane A₂, prostaglandin D₂, E₂ and F₂α6- to 12-fold over basal within 10 min. In contrast, human anaphylatoxin C5a did not increase prostanoid formation in rat Kupffer cells. (2) The increase in prostanoid formation by recombinant rat C5a was specific. It was inhibited by a neutralizing monoclonal antibody. (3) In co-cultures of rat hepatocytes and rat Kupffer cells but not in hepatocyte mono-cultures recombinant rat C5a increased glycogen phosphorylase activity 3-fold over basal. This effect was inhibited by incubation of the co-cultures with 500 μM acetylsalicyclic acid. Thus, C5a generated either locally in the liver or systemically e.g. in the course of sepsis, may increase hepatic glycogenolysis by a prostanoid-mediated intercellular communication between Kupffer cells and hepatocytes.
Rat serum, in which the complement sytem had been activated by incubation with zymosan, increased the glucose and lactate output, and reduced and redistributed the flow in isolated perfused rat liver clearly more than the control serum. Heat inactivation of the rat serum prior to zymosan incubation abolished this difference. Metabolic and hemodynamic alterations caused by the activated serum were dose dependent. They were almost completely inhibited by the cyclooxygenase inhibitor indomethacin and by the thromboxane antagonist 4-[2-(4-chlorobenzenesulfonamide)-ethyl]-benzene-acetica cid (BM 13505), but clearly less efficiently by the 5’-lipoxygenase inhibitor nordihydroguaiaretic acid and the leukotriene antagonist N-{3-[3-(4-acetyl-3-hydroxy-2-propyl-phenoxy)-propoxy]-4-chlorine-6-methyl-phenyl}-1H-tetrazole-5-carboxamide sodium salt (CGP 35949 B). Control serum and to a much larger extent complement-activated serum, caused an overflow of thromboxane B₂ and prostaglandin F₂α into the hepatic vein. It is concluded that the activated complement system of rat serum can influence liver metabolism and hemodynamics via release from nonparenchymal liver cells of thromboxane and prostaglandins, the latter of which can in turn act on the parenchymal cells.
In perfused rat livers, infusion of prostaglandin F₂α (PGF₂α) or noradrenaline increased glucose and lactate output and reduced flow. Glucagon increased glucose output and decreased lactate output without influence on flow. Infusion of phorbol 13-myristate 14-acetate (PMA) for 20 min prior to these stimuli strongly inhibited the metabolic and hemodynamic effects of noradrenaline, reduced the metabolic actions of PGF₂α but did not alter the effects of glucagon. In isolated rat hepatocytes PGF₂α, noradrenaline and glucagon activated glycogen phosphorylase but only PGF₂α and noradrenaline increased intracellular inositol 1,4,5-1risphosphalc (InsP₃). The noradrenaline- or PGF₂α-elicited activation of glycogen phosphorylase and increase in InsP₃ were largely reduced after preincubation of the cells for 10 min with PMA, whereas the glucagon-mediated enzyme activation was not affected. In contra\t to PMA, the phorbol ester 4a-phorbol 13,14-didecanoate. which does not activate protein kinase C, did not attenuate the PGF₂α- and noradrenaline-elicited stimulation of glucose output, glycogen phosphorylase and InsP, formation. Stimulation of InsP₃ formation by AlF₄⁻, which activates phospholipase C independently of the receptor, was not attenuated by prior incubation with PMA. Plasma membranes purified from isolated hepatocytes had both a high-capacity, low-affinity and a low-capacity, high-affinity binding site for PGF₂α. The Kd of the high-capacity, low-affinity binding site was close to the concentration of PGF₂α that increased glycogen phosphorylase activity halfmaximally. Binding to the high-capacity, low-affinity binding site was enhanced by guanosine 5'- 0-(3-thio)triphosphate (GTP[S]). This high-capacity, low-affinity site might thus represent the receptor. The Bmax and Kd of the high-capacity site, as well as the enhancement by GTP[S] of PGF₂α binding to this site, remained unaffected by PMA pretreatment. It is concluded that, in hepatocytes, activation of protein kinase C by PMA interrupted the InsP₃-mediated signal pathway from PGF₂α via a PGF₂α receptor and phospholipase C to glycogen phosphorylase at a point distal of the receptor prior to phospholipase C.
Human placenta is surprisingly rich in post-proline dipeptidyl peptidase activity. Among various cell fractions, microsomes have the highest specific activity. A homogeneous enzyme preparation is obtained in a six-step purification procedure. The final preparation appears homogeneous upon dodecyl sulfate electrophoresis, but analytical isoelectric focussing reveals various active bands with isoelectric points in the range of pH 3 - 4. The enzyme is a glycoprotein containing about 30% carbohydrate. Treatment with neuraminidase lowers the isoelectric points but does not reduce the heterogeneity of the band pattern. The subunit molecular weight is 120000 as estimated by dodecyl sulfate electrophoresis, whereas Mr of the native enzyme is > 200000, as can be concluded from gel filtration experiments. The purified dipeptidyl peptidase cleaves various synthetic and natural peptides, including substance P, kentsin, casomorphin and a synthetic renin inhibitor. In general, the specificity of the placenta peptidase is similar to that of post-proline dipeptidyl peptidase from other sources. Phenylalanylprolyl-P-naphthylamide (Km = 0.02 mM, I/ = 92 Ujmg) is the best substrate among various synthetic peptide derivatives. Only peptides with a free N-terminal amino group and proline, hydroxyproline, or alanine in position 2 of the N-terminal sequence are cieaved. However, X-Pro-Pro- . . . structures, e. g. as in bradykinin, are not attacked. 1 mM bis-(6nitrophenyI)phosphate or 1 mM diisopropylfluorophosphate completely inactivate the peptidase within 30 min at 30°C (pH 8). The peptidase is also completely inhibited by 1 mM Zn²⁺ and by other heavy metals.
Prostaglandin (PG)F₂α has previously been shown to increase glucose output from perfused livers and isolated hepatocytes, where it stimulated glycogen phosphorylase via an inositol-trisphosphatedependent signal pathway. In this study, PGF₂α binding sites on hepatocyte plasma membranes, that might represent the putative receptor, were characterized. Binding studies could not be performed with intact hepatocytes, because PGF₂α accumulated within the cells even at 4°C. The intracellular accumulation was an order of magnitude higher than binding to plasma membranes. Purified hepatocyte plasma membranes had a high-affinity/low-capacity and a low-affinity/highcapacity binding'site for PGF₂α. The respective binding constants for the high-affinity site were Kd = 3 nM and Bmax = 6 fmol/mg membrane protein, and for the low-affinity site Kd = 426 nM and Bmax = 245 fmol/mg membrane protein. Specific PGF₂α binding to the low-affinity site, but not to the high-affinity site, could be enhanced most potently by GTP[γS] followed by GDP[ϐS] and GTP, but not by ATP[γS] or GMP. PGF₂α competed most potently with [³H]PGF₂α for specific binding to hepatocyte plasma membranes, followed by PGD₂ and PGE₂. Since the low-affinity PGF₂α-binding site had a Kd in the concentration range in which PG had previously been shown to be half-maximally active, and since this binding site showed a sensitivity to GTP, it is concluded that it might represent the receptor involved in the PGF₂α signal chain in hepatocytes. A biological function of the high-affinity site is currently not known.
Prostaglandin E₂ has been reported both to stimulate glycogen-phosphorylase activity (glycogenolytic effect) and to inhibit the glucagon-stimulated glycogen-phosphorylase activity (antiglycogenolytic effect) in rat hepatocytes. It was the purpose of this study to resolve this apparent contradiction and to characterize the signalling pathways and receptor subtypes involved in the opposing prostaglandin E₂ actions. Prostaglandin E₂ (10 μM) increased glucose output, glycogen-phosphorylase activity and inositol trisphosphate formation in hepatocyte cell culture andor suspension. In the same systems, prostaglandin E₂ decreased the glucagon-stimulated (1 nM) glycogen-phosphorylase activity and cAMP formation. The signalling pathway leading to the glycogenolytic effect of PGE₂ was interrupted by incubation of the hepatocytes with 4P-phorbol 12-myristate 13-acetate (100 nM) for 10 min, while the antiglycogenolytic effect of prostaglandin E₂ was not attenuated. The signalling pathway leading to the antiglycogenolytic effect of prostaglandin E₂ was interrupted by an incubation of cultured hepatocytes with pertussis toxin (100 ng/ml) for 18 h, whereas the glycogenolytic effect of prostaglandin E₂ was enhanced. The EP₁/EP₃ prostaglandin-E₂-receptor-specific prostaglandin E₂ analogue Sulproston had a stronger glycogenolytic potency than the EP₃ prostaglandin-E₂-receptor-specific prostaglandin E₂ analogue Misoprostol. The antiglycogenolytic potency of both agonists was equal. It is concluded that the glycogenolytic and the antiglycogenolytic effects of prostaglandin E₂ are mediated via different signalling pathways in hepatocytes possibly involving EP₁ and EP₃ prostaglandin E₂ receptors, respectively.
In the isolated rat liver perfused in situ, stimulation of the nerve bundles around the hepatic artery and portal vein caused an increase of glucose and lactate output and a reduction of perfusion flow. These changes could be inhibited completely by α-receptor blockers. The possible involvement of inositol phosphates in the intracellular signal transmission was studied. 1. In cell-suspension experiments, which were performed as a positive control, noradrenaline caused an increase in glucose output and, in the presence of 10 mM LiCl, a dose-dependent and time-dependent increase of inositol mono, bis and trisphosphate. 2. In the perfused rat liver 1 μM noradrenaline caused an increase of glucose and lactate output and in the presence of 10 mM LiCl a time-dependent increase of inositol mono, bis and trisphosphate that was comparable to that observed in cell suspensions. 3. In the perfused rat liver stimulation of the nerve bundles around the portal vein and hepatic artery caused a similar increase in glucose and lactate output to that produced by noradrenaline, but in the presence of 10 mM LiCl there was a smaller increase of inositol monophosphate and no increase of inositol bis and trisphosphate. These findings are in line with the proposal that circulating noradrenaline reaches every hepatocyte, causing a clear overall increase of inositol phosphate formation and thus calcium release from the endoplasmic reticulum, while the hepatic nerves reach only a few cells causing there a small local change of inositol phosphate metabolism and thence a propagation of the signal via gap junctions.