Purification, Biochemical Characterization, Binding Activity, and Selectivity of a Glutamate Binding Protein from Bovine Brain

A glutamate binding protein was purified from bovine brain to apparent homogeneity. The procedure used for the purification of this protein involved extraction of a crude synaptic membrane fraction with Na-cholate, followed by solubilization of the binding protein from the membranes by Triton X-100, and, finally, affinity batch separation of the protein on L-glutamate-loaded glass fiber. The molecular characteristics of the purified protein were similar to those previously described for the glutamate binding protein from rat brain synaptic membranes and included the following: small Mr (14,000), acidic (pI = 4.7) protein with a single NH2-terminal amino acid (tyrosine), and significant absorption at wave-lengths greater than 300 nm. Complete amino acid analysis of the protein was not achieved, either because of destruction of some amino acids or of incomplete hydrolysis of the protein. The protein bound L-glutamate with high affinity (KD = 0.87 microM), exhibited one class of L-glutamate binding sites, and bound glutamate with a stoichiometry of 0.7 mol ligand/mol protein. The displacement of protein-bound L-glutamic acid by other neuroactive amino acids had characteristics similar to those observed for the displacement of L-glutamate from rat brain synaptic membrane or purified protein binding sites. Finally, the metal ligand formers KCN and NaN3 inhibited the activity of this protein just as they have been shown to do in rat brain synaptic membranes or the purified protein.     

Solubilisation of a Glutamate Binding Protein from Rat Brain

Rat brain synaptic plasma membranes were solubilised in either 1% Triton X-100 or potassium cholate and subjected to batch affinity adsorption on L-glutamate/bovine serum albumin reticulated glass fibre. The fibre was extensively washed, and bound proteins eluted with 0.1 mM L-glutamate in 0.1% detergent, followed by repeated dialysis to remove the glutamate from the eluted proteins. Aliquots of the dialysed extracts were assayed for L-[3H]glutamate binding activity in the presence or absence of 0.1 mM unlabelled L-glutamate (to define displaceable binding). Incubations were conducted at room temperature and terminated by rapid filtration through nitrocellulose membranes. Binding to solubilised fractions could be detected only following affinity chromatography. Binding was saturable and of relatively low affinity: KD = 1.0 and 1.8 microM for Triton X-100 and cholate extracts, respectively. The density of binding sites was remarkably high: approximately 18 nmol/mg protein for Triton X-100-solubilised preparations, and usually double this when cholate was employed. Analysis of structural requirements for inhibition of binding revealed that only a very restricted number of compounds were effective, i.e., L-glutamate, L-aspartate, and sulphur-containing amino acids. Binding was not inhibited significantly by any of the selective excitatory amino acid receptor agonists--quisqualate, N-methyl-D-aspartate, or kainate. The implication from this study is that the glutamate binding protein is similar if not identical to one previously isolated and probably is not related to the pharmacologically defined postsynaptic receptor subtypes, unless solubilisation of synaptic membranes resulted in major alterations to binding site characteristics. Since solubilisation with Triton X-100 is known to preserve synaptic junctional complexes, it seems likely that the origin of the glutamate binding protein may be extrajunctional, although its functional role is unknown.     

Rat Brain S100b Protein: Purification, Characterization, and Ion Binding Properties. A Comparison with Bovine S100b Protein

We purified to homogeneity rat brain S100b protein, which constitutes about 90% of the soluble S100 protein fraction. Purified rat S100b protein comigrates with bovine S100b protein in nondenaturant system electrophoresis but differs in its amino acid composition and in its electrophoretic mobility in urea-sodium dodecyl sulfate-polyacrylamide gel with bovine S100b protein. The properties of the Ca2+ and Zn2+ binding sites on rat S100b protein were investigated by flow dialysis and by fluorometric titration, and the conformation of rat S100b in its metal-free form as well as in the presence of Ca2+ or Zn2+ was studied. The results were compared with those obtained for the bovine S100b protein. In the absence of KCl, rat brain S100b protein is characterized by two high-affinity Ca2+ binding sites with a KD of 2 X 10(-5) M and four lower affinity sites with KD about 10(-4) M. The calcium binding properties of rat S100b protein differ from bovine S100b only by the number of low-affinity calcium binding sites whereas similar Ca2+-induced conformational changes were observed for both proteins. In the presence of 120 mM KCl rat brain S100b protein bound two Zn2+-ions/mol of protein with a KD of 10(-7) M and four other with lower affinity (KD approximately equal to 10(-6) M). The occupancy of the two high-affinity Zn2+ binding sites was responsible for most of the Zn2+-induced conformational changes in the rat S100b protein. No increase in the tyrosine fluorescence quantum yield after Zn2+ binding to rat S100b was observed.(ABSTRACT TRUNCATED AT 250 WORDS)     

Purification and Molecular Characterization of the Brain Synaptic Membrane Glutamate-Binding Protein

A glutamate-binding protein from rat brain synaptic plasma membranes has been purified to apparent homogeneity. This protein has a Mr of 14,300 based on amino acid and carbohydrate analyses. The protein is enriched with tryptophan residues, which contribute substantially to its hydrophobic nature. It also has a relatively high content of acidic amino acids, which determine is low isoelectric point (4.82). The protein exhibits either a single, high-affinity class of sites for L-[3H]glutamate binding (KD = 0.13 microM) when binding is measured at low protein concentrations, or two classes of sites with high (KD = 0.17 microM) and low affinities (KD = 0.8 microM) when binding is measured at high protein concentrations. These observations suggest preferential binding of L-glutamate to a self-associating form of the protein. The displacement of protein-bound L-[3H]glutamic acid by other neuroactive amino acids has characteristics similar to those observed for displacement of L-glutamate from membrane binding sites. Chemical modification of the cysteine and arginine residues results in an inhibition of glutamate binding activity. The possible function of this protein in the physiologic glutamate receptor complex of neuronal membranes is discussed.     

Cross-Reaction of Anti-Rat B-50: Characterization and Isolation of a "B-50 Phosphoprotein" from Bovine Brain

Abstract: Antibodies to the phosphoprotein B-50 of rat brain were used to trace cross-reacting brain proteins of vertebrates. With the SDS-gel-immunoperoxidase method, a cross-reacting protein (CP) of apparent M_r 53,000 was demonstrated in the homogenate and the synaptic plasma membrane fraction of bovine brain. Sequence 1–24 of adrenocorticotropin (ACTH_1-24) (10⁻⁵ M and 10⁻⁴ M ) inhibited endogenous phosphorylation of CP in synaptic plasma membranes. The protein was partially characterized and purified to homogeneity from bovine brain by procedures previously described for rat B-50. CP was enriched in ammonium sulfate precipitated protein (ASP) fractions and phosphorylated by an endogenous protein kinase. Two-dimensional gel analysis of bovine and rat ASP showed that the cross-reacting protein had an isoelectric point less acidic than B-50. Limited proteolysis by Staphylococcus aureus protease yielded a "peptide map" analogous to B-50. Two major fragments of M_r 30,000 and 17,000 were produced. In addition, CP exhibited other similarities to rat B-50: phosphorylation by rat brain protein kinase C, microheterogeneity observed after isoelectric focusing, and possibly degradation by endogenous proteolysis. Cross-reaction of proteins in brain homogenates of other mammalian species and of chicken was demonstrated: the M_r of the proteins ranged from 47,000 to 53,000. We conclude that (1) the cross-reacting bovine protein is a "B-50 protein," and (2) the M _r of the "B-50 protein" varies from species to species.     

Highly Sensitive Immunoassay for the α Subunit of the GTP-Binding Protein Go and Its Regional Distribution in Bovine Brain

Antisera were raised in rabbits against the alpha subunit of a GTP-binding protein, Go. Because the antisera cross-reacted weakly with the alpha subunit of inhibitory GTP-binding protein of adenylate cyclase (Gi), they were purified with a Go alpha-coupled Sepharose column. Purified antibodies reacted only with Go alpha and did not cross-react with the Gi alpha subunit or beta gamma subunits in an immunoblot assay. Using these purified antibodies, a highly sensitive enzyme immunoassay method for the quantification of bovine brain Go alpha was developed. The assay system consisted of polystyrene balls with immobilized antibody F(ab')2 fragments and the same antibody Fab' fragments labeled with beta-D-galactosidase from Escherichia coli. The minimal detection limit of the assay was 0.1 fmol, or 4 pg. The assay was specific for Go alpha, and it did not cross-react with Gi alpha or beta gamma. Samples from various regions of bovine brain were solubilized with 2% sodium cholate and 1 M NaCl, and the concentrations of Go alpha were determined. Go alpha was detected in all the regions, and the highest concentration was observed in the cerebral cortex. The immunohistochemical study showed that the neuropil was rich in Go alpha.     

Effect of Free Malonate on the Utilization of Glutamate by Rat Brain Mitochondria

Malonate is an effective inhibitor of succinate dehydrogenase in preparations from brain and other organs. This property was reexamined in isolated rat brain mitochondria during incubation with L-glutamate. The biosynthesis of aspartate was determined by a standard spectrofluorometric method and a radiometric technique. The latter was suitable for aspartate assay after very brief incubations of mitochondria with glutamate. At a concentration of 1 mM or higher, malonate totally inhibited aspartate biosynthesis. At 0.2 mM, the inhibitory effect was still present. It is thus possible that the natural concentration of free malonate in adult rat brain of 192 nmol/g wet weight exerts an effect on citric acid cycle reactions in vivo. The inhibition of glutamate utilization by malonate was readily overcome by the addition of malate which provided oxaloacetate for the transamination of glutamate. The reaction was accompanied by the accumulation of 2-oxoglutarate. The metabolism of glutamate was also blocked by inclusion of arsenite and gamma-vinyl-gamma-aminobutyric acid but again added malate allowed transamination to resume. When arsenite and gamma-vinyl-gamma-aminobutyric acid were present, the role of malonate as an inhibitor of malate entry into the mitochondrial interior could be determined without considering the inhibition of succinate dehydrogenase. The apparent Km and Vmax values for uninhibited malate entry were 0.01 mM and 100 nmol/mg protein/min, respectively. Malonate was a competitive inhibitor of malate transport (Ki = 0.75 mM).     

Deimination of Human Myelin Basic Protein by a Peptidylarginine Deiminase from Bovine Brain

Abstract: A peptidylarginine deiminase (PAD; EC 3.5.3.15) has been isolated from bovine brain and some of its characteristics have been studied. The enzyme showed an absolute requirement for Ca²⁺, a temperature optimum at ～50°C, and two K_mvalues when benzoylarginine ethyl ester was used as substrate, 0.78 mMand 11.2 mM.The higher K_mhas not been reported previously. Protein substrates for the enzyme included polyarginine and myelin basic protein but not histones. Because one of the components of MBP contains six citrullinyl residues per mole, enzymic deimination appeared to be a likely mechanism. When the most cationic component (C-1) was subjected to PAD in solution, 17 of the 19 arginyl residues were modified. From sequence analyses we concluded that the nature of the amino acid residues adjacent to the deiminated arginine were not modifiers of the reaction as arginyl residues in a variety of environments were deiminated. This deimination was reflected in a large increase in random structure, as measured by [θ]₂₀₀. At 5°C, the [θ]₂₀₀of the deiminated protein was -70 × 10³ compared with -30 × 10³ deg cm²/dmol for the native protein. When the temperature was increased to 70°C, the [θ]₂₀₀ was -44 × 10³ for the deiminated protein and -20 × 10⁷ deg cm²/ dmol for the native C-1. When plotted as a function of temperature, [θ]₂₀₀ decreased linearly from 5°C to 50°C for both proteins and did not change from 50°C to 70°C. PAD provides a mechanism for deimination of arginyl residues of myelin basic protein. The selective deimination of the six arginyl residues that are consistently found deiminated in C-8 may be determined by the orientation of the protein in the membrane and/or the more complex lipid composition of myelin may affect the selectivity of the deimination.     

Sequential Limited Proteolysis of Myelin Basic Protein by Neutral Protease Activities of Bovine Brain

Acid extracts of delipidated white matter of bovine brain were prepared, and their proteolytic activities toward myelin basic protein (MBP) were evaluated at pH 3 and pH 7. This was done by measuring changes in total protein using a selective dye-binding assay, and by evaluating peptide patterns by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and densitometry. At pH 7 greater than 50% of total protein and about 75% of MBP were degraded after 48 h, whereas at pH 3 it was less than 20% altogether. Neutral proteolysis of MBP entailed up to 12 different proteolytic peptide fragments in the molecular weight range of 17.5 to 6 kd. Its enzymatic nature was verified using protease inhibitors, including N-ethylmaleimide, phenylmethylsulfonyl fluoride, o-phenanthroline, and EDTA, as well as pepstatin A and alpha 2-macroglobulin. Both transient changes in percentages of some intermediate peptides and differential effects of individual inhibitors on electrophoretic peptide patterns strongly suggest a sequential type of limited proteolysis. The results also indicate that acid extracts contained several endopeptidases of which a cysteine protease appears to initiate the breakdown of MBP.     

牛脑中神经特异性S-100蛋白的纯化及鉴定

牛脑充分匀浆后经三次硫酸铵分级沉淀,再通过一次DEAE-Sepharose CL-6B层析柱,线性梯度洗脱后共收集4个峰洗脱液。PAGE分析(7.5％凝胶)显示第3峰为单一区带;免疫双扩散证实该洗脱液中蛋白为S-100蛋白。SDS-PAG E分析显示S-100蛋白分子量约为10kD;非还原条件下,凝胶过滤(Sephadex G-75)显示S-100蛋白位于MW为20kD区域。认为该纯化方法简便、快速,可获得较高纯度的S-100蛋白,活性高达1∶128以上,完全能满足进一步研究之用。     

Effect of Monocular Deprivation on Uptake and Binding of [3H]Glutamate in the Visual System of Rat Brain

Abstract: [³H]Glutamate uptake and binding studies were performed in the visual cortices, lateral geniculate nuclei (LGN), and superior colliculi of 3-month-old rats with one eyelid surgically closed from postnatal day 10 (monocular deprivation). Uptake and binding were highest in the lateral geniculate nucleus followed by the visual cortex (69% and 15%, respectively compared to LGN values) and the superior colliculus (32% and 59% of LGN values). Monocular deprivation did not affect [³H]glutamate uptake in any of the visual regions examined. However, a 46% decrease in [³H]glutamate binding in the lateral geniculate nucleus ipsilateral to the sutured eye was detected. Binding levels in other regions were not affected.     

Glutamate Dehydrogenase in Human Brain: Regional Distribution and Properties

Glutamate dehydrogenase (GDH) activity was studied in 17 regions of six human brains. Duration and conditions of the postmortem period did not affect enzyme activity. Specific activity ranged between 103 and 377 nmoles/min/mg protein at 25 degrees C and it was 10-fold higher than that found in leukocytes. Apart from exclusively white matter regions (corpus callosum and centrum ovale), there was a moderate regional distribution (2.5-fold variation), with highest values in the inferior olive and hypothalamus, and lowest in the cerebellum and lenticular nucleus. With alpha-ketoglutarate (alpha-KG), NADH, or NH4+ as variable substrate, the apparent Km values in human brain were Km alpha-KG = 1.9 X 10(-3) M, KmNADH = 0.21 X 10(-3) M, and KmNH4+ = 28 X 10(-3) M, and in leukocytes they were Km alpha-KG = 1.7 X 10(-3) M, KmNADH = 0.24 X 10(-3) M, and KmNH4+ = 28 X 10(-3) M. The effects of cofactors, inhibitor, and pH were similar in brain and leukocyte GDH.     

Glutamate Binding to Brain Membranes Is Increased in Pentylenetetrazole-Kindled Rats

Abstract: The specific binding of L-[³H]glutamate to its receptors was investigated on crude membrane preparations from different brain regions of pentylenetetrazole-kindled rats using a binding assay technique. Pentylenetetrazole kindling induced by 10 intraperitoneal applications of 45 mg/kg over a period of 20 days resulted in a significant increase of both the convulsive susceptibility of animals to the convulsant and the specific L-[³H]glutamate binding in hippocampus and in motor, frontal, and inferiotemporal (acoustic) cortex tested with a L-[³H]glutamate concentration of 50 n M . No differences were observed in the other brain structures studied. Kinetic studies indicated that the enhanced L-[³H]glutamate binding to hippocampal membranes from kindled rats reflects changes in the density of the glutamate binding sites rather than an increase in receptor affinity. To study the effect of acute generalized convulsions on L-[³H]glutamate binding to synaptosomal membranes of hippocampus and visual cortex, rats were treated 24 h before the experiment with 60 mg/kg of pentylenetetrazole, i.p. Under these conditions, no differences between treated and control rats were observed. From these findings, it is concluded that the increase in glutamate receptor density demonstrated in hippocampus and several neocortical brain structures of pentylenetetrazole-kindled rats may be the expression of a specific enhancement of susceptibility of glutamatergic systems to this excitatory amino acid developing in the course of formation of pentylenetetrazole-induced kindling.     

Binding of ATP to Brain Glutamate Decarboxylase as Studied by Affinity Chromatography

Abstract: The interactions of two forms of porcine brain glutamate decarboxylase (β-GAD and γ-GAD) with the effector ATP were studied by affinity chromatography. A third form, γk-GAD, was only slightly retarded by the affinity matrix and was eluted in the buffer wash. The interaction of GAD with the ATP affinity matrix was qualitatively similar to its interaction with free ATP as reported in previous kinetic studies. The rank order of adenine nucleotides as eluting agents and affinity ligands was ATP > ADP > AMP. GAD was also eluted by its cofactor, pyridoxal 5'-phosphate, and this was enhanced by 1 mM P_i In contrast, a high concentration (140 mM) of P_i by itself was required to elute the enzyme. GAD remained active while bound to the affinity column and was eluted in the holoenzyme form by ATP, indicating that the affinity ligand did not bind in the active site and did not displace catalytically active cofactor from the enzyme.     

Characterization of a Sulfoglucuronyl Carbohydrate Binding Protein in the Developing Nervous System

Abstract: The developmentally regulated and stage-specifically expressed HNK-1 carbohydrate found on sulfoglucuronylglycolipids (SGGLs) and certain glycoproteins has been proposed to be involved in neural cell adhesion and recognition processes through its interaction with protein "receptors." We have isolated and purified a ∼30-kDa SGGL-binding protein (SBP-1) from neonatal rat brain. SBP-1 specifically bound to SGGLs and sulfatide both in solid-phase immunobinding and high-performance thin-layer chromatography-immunooverlay assays. N-terminal sequence analysis showed that SBP-1 is similar to an adhesive neurite outgrowth promoting protein amphoterin. Desulfation of SGGLs resulted in abolition of SBP-1 binding. However, chemical modification of glucuronic acid moiety by either esterification or reduction of the carboxyl group had no effect, suggesting requirement of the carbohydrate-linked sulfate group for SBP-1 binding. The binding of SBP-1 to SGGLs was specifically inhibited by HNK-1 antibody but not by other IgM antibodies. The binding of SBP-1 to sulfatide, however, was not inhibited by HNK-1 antibody. Heparin, fucoidan, and dextran sulfate (50K) also inhibited the binding of SBP-1 to SGGLs. During development of the rat cerebral cortex, the level of SBP-1 decreased after embryonic day 18 to an almost undetectable level by postnatal day 10; whereas in the cerebellum, the expression of SBP-1 was maximal at postnatal day 7. SBP-1 also bound specifically to the HNK-1 glycoproteins isolated from rat brain by HNK-1 immunoaffinity chromatography. Proteins without HNK-1 carbohydrate did not bind SBP-1. The binding to HNK-1 glycoproteins was inhibited by HNK-1 antibody, but not by other IgM antibodies, indicating that the binding was mediated through the HNK-1 carbohydrate moiety of the proteins. The interaction and coexpression of SBP-1 with SGGLs and HNK-1 glycoproteins, during the perinatal brain development, suggest a functional role for this protein.     

Macrophage Migration Inhibitory Factor: Identification of the 30-kDa MIF-Related Protein in Bovine Brain

Macrophage migration inhibitory factor (MIF) is a ubiquitous protein playing various immunologic, enzymatic, and hormonal roles. MIF was originally identified for its capacity to inhibit the random movement of macrophages in vitro. MIF is widely expressed in many tissues with particularly high levels in the nervous system. Using the reversed-phase HPLC, N-terminal microsequence analysis, and database searching, we have identified in bovine brain several MIF-like proteins. According to mass spectral analysis, the molecular masses for three of them were determined as 12,369.2, 12,299.7, and 9,496.2 Da. In addition, we have identified another MIF-related protein (29,568.9 Da) by Western blotting using anti-MIF antibody raised to MIF (having an apparent molecular weight of 12 kDa) isolated to homogeneity from bovine brain cytosol. The modified purification procedure was mainly based on exclusion- and ion-exchange chromatography. Using p-hydroxyphenylpyruvic acid as a substrate, we have demonstrated tautomerase activity of the isolated MIF. The N-terminal sequences for all MIF-like proteins were found to be identical. Several other higher molecular weight putative MIF-related proteins were also revealed in the bovine brain cytosol extract. A multifunctional nature of MIF is suggested to be a result of its occurrence in different oligomerization states in a wide variety of tissues and cells.     

BSA载体蛋白抗体的去除

为了去除抗血清中BSA载体蛋白产生的抗体,一根对BSA载体蛋白抗体高度特异性的亲和柱被构建。结果表明：所构建的亲和柱对BSA载体蛋白抗体具高度特异性和亲和力,能有效地去除BSA载体蛋白产生的抗体。     

Thyrotropin-Releasing Hormone Receptor Binding Sites: Autoradiographic Distribution in the Rat and Guinea Pig Brain

Thyrotropin-releasing hormone (TRH) binding sites were labeled in vitro in mounted brain tissue sections from rat and guinea pig brains with [3H]methyl TRH and localized autoradiographically using 3H-sensitive film. Regional densities of TRH binding sites were measured by computer-assisted microdensitometry. The distribution of sites in both species was highly heterogeneous. In both guinea pig and rat brains, the highest densities of binding sites were seen in the amygdaloid nuclei and the perirhinal cortex. In contrast, in other brain areas, a clear difference between the distribution of sites in rat and guinea pig was found. The temporal cortex, pontine nuclei, and interpeduncular nucleus, which contained high densities of binding in the guinea pig, were scarcely labeled in the rat. The accessory olfactory bulb and the septohippocampal area presented in the rat higher concentrations of binding sites than in the guinea pig. Other brain areas showing intermediate to low densities in both species were accumbens nucleus, bed nucleus of the stria terminalis, dentate gyrus, facial and hypoglossal nuclei, and gelatinosus subnucleus of the trigeminal nerve, among others. The anterior pituitary also presented low to intermediate concentrations of receptors. The distribution of TRH sites here described does not completely correlate with that of endogenous TRH, but is in good agreement with previous biochemical data. The results are discussed in correlation to the physiological effects that appear to be mediated by TRH.     

Developmental Study of Proteolipids in Bovine Brain: A Novel Proteolipid and DM-20 Appear Before Proteolipid Protein (PLP) During Myelination

In a developmental study, we have shown that DM-20 is present before proteolipid protein (PLP) in the fetal bovine cerebral hemispheres. When the white matter appears (27-30 weeks of gestation), the amount of DM-20 drastically increases. DM-20 remains the major proteolipid until birth. PLP is detected only 2-4 weeks after the appearance of white matter, that is, more than 4 weeks after the appearance of DM-20. The early appearance of DM-20 at the beginning of myelination raises the question of its particular function. In the adult bovine cerebral hemispheres, PLP is the major proteolipid but DM-20 remains quantitatively important because the PLP/DM-20 ratio ranges from 1.5 to 1.7. In the same developmental study we have, in the fetal cerebral hemispheres, isolated and characterized a novel proteolipid (apparent Mr 20,000), which appears even before DM-20 and is not detected in the adult brain. It is structurally related to PLP and DM-20 because the first 31 N-terminal amino acid residues are the same. However, in immunoblot, it did not react either with the antitridecapeptide 117-129 antiserum of PLP or with the anti-C-terminal hexapeptide antiserum of PLP.     

Regional Development of Glutamate Dehydrogenase in the at Brain

The development of glutamate dehydrogenase enzyme activity in rat brain regions has been followed from the late foetal stage to the adult and through to the aged (greater than 2 years) adult. In the adult brain the enzyme activity was greatest in the medulla oblongata and pons greater than midbrain = hypothalamus greater than cerebellum = striatum = cortex. In the aged adult brain, glutamate dehydrogenase activity was significantly lower in the medulla oblongata and pons when compared to the 90-day-old adult value, but not in other regions. The enzyme-specific activity of nonsynaptic (free) mitochondria purified from the medulla oblongata and pons of 90-day-old animals was about twice that of mitochondria purified from the striatum and the cortex. The specific activity of the enzyme in synaptic mitochondria purified from the above three brain regions, however, remained almost constant.