Intragenic integration in DLC1 sustains factor VIII expression in primary human cells without insertional oncogenicity.

Techniques enabling precise genome modifications enhance the safety of gene-based therapy. DLC1 is a hot spot for phiC31 integrase-mediated transgene integration in vitro and in vivo. Here we show that integration of a coagulation factor VIII transgene into intron 7 of DLC1 supports durable expression of factor VIII in primary human umbilical cord-lining epithelial cells. Oligoclonal cells with factor VIII transgene integrated in DLC1 did not have altered expression of DLC1 or neighbouring genes within a 1-Mb interval. Only 1.9% of all expressed genes were transcriptionally altered; most were downregulated and mapped to cell cycle and DNA repair pathways. DLC1-integrated cells were not tumourigenic in vivo and were normal by high-resolution genomic DNA copy number analysis. Our data identify DLC1 as a locus for durable transgene expression that does not incur features of insertional oncogenesis, thus expanding options for developing ex vivo cell therapy mediated by site-specific integration methods.

Insulin expressed from endogenously active glucose-responsive EGR1 promoter in bone marrow mesenchymal stromal cells as diabetes therapy.

Advances in islet transplantation have encouraged efforts to create alternative insulin-secreting cells that overcome limitations associated with current therapies. We have recently demonstrated durable correction of murine and porcine diabetes by syngeneic and autologous implantation, respectively, of primary hepatocytes non-virally modified with a glucose-responsive promoter-regulated insulin transgene. As surgical procurement of hepatocytes may be clinically unappealing, we here describe primary bone marrow-derived mesenchymal stromal cells (BMMSC) as alternative insulin-secreting bioimplants. BMMSC are abundant and less invasively procured for clinical autologous transplantation. Electroporation achieved high transgene transfection efficiencies in human BMMSC (HBMMSC) and porcine BMMSC (PBMMSC). We transcriptomically identified an HBMMSC glucose-responsive promoter, EGR1. This endogenously active promoter drove rapid glucose-induced transgene secretions in BMMSC with near-physiological characteristics during static and kinetic induction assays simulating normal human islets. Preparatory to preclinical transplantation, PBMMSC transfected with the circular insulin transgene vector or stably integrated with the linearized vector were evaluated by intrahepatic or intraperitoneal xenotransplantation in streptozotocin-diabetic and non-diabetic NOD-SCID mice. Hyperglycemia, glucose tolerance and body weight were corrected in a dose-responsive manner. Hypoglycemia was not observed even in identically implanted non-diabetic mice. These results establish human EGR1 promoter-insulin construct-modified BMMSC as safe and efficient insulin-secreting bioimplants for diabetes treatment.

Plasmid-electroporated primary hepatocytes acquire quasi-physiological secretion of human insulin and restore euglycemia in diabetic mice.

We describe the durable correction of streptozotocin-induced murine diabetes by in vivo implantation of primary mouse hepatocytes electroporated ex vivo with a human proinsulin cDNA plasmid construct controlled by glucose and zinc regulatory elements. Transfected hepatocytes increased insulin transgene transcription and secretion within 10-20 min of exposure to 25 mM glucose or 60 microM zinc. Insulin release did not occur from secretory granules. Electroporated Rosa26 hepatocytes ( approximately 8 x 10(5) viable cells) were implanted in C57BL/6J diabetic mice in one of three sites: unresected liver, regenerating liver or mesentery. Control diabetic mice were implanted with untransfected hepatocytes. At 30 days after implantation, 8/15 control mice were alive, while 19/19 treated mice were alive. The ratio of body weight on day 30/nadir body weight was significantly higher for all treated groups compared with controls. All eight surviving control mice were hyperglycemic 30 days post-implantation, while 16/19 treated diabetic mice remained normoglycemic. Treated mice had lower mean glucose values (P< or =0.001) without fasting hypoglycemia and better glucose tolerance (P< or =0.0003) than untreated controls. All (6/6) diabetic mice implanted in regenerating liver and 71% (5/7) implanted in unresected liver were alive 77 days after implantation. Engrafted hepatocytes were identified, mainly around central veins, by staining for beta-galactosidase activity and with anti-human insulin antibody.

Naked plasmid-mediated gene transfer to skeletal muscle ameliorates diabetes mellitus.

BACKGROUND: The ability of tissues to take up naked plasmid DNA in vivo suggests an approach for reconstituting systemic metabolic deficiencies without the disadvantages of viral vectors and lipid-DNA complexes. Plasmid-mediated gene transfer into skeletal muscle was investigated as a means of providing a therapeutic source of insulin. METHODS: Four plasmid constructs, each bearing a mouse furin cDNA transgene and rat proinsulin cDNA (modified for processing by furin) driven by four different promoters were injected into the calf muscles of male Balb/c mice. Insulin and C-peptide concentrations were measured by radio-immunoassays having minimal crossreactivity for proinsulin and partially processed proinsulin. RESULTS: Intramuscular insulin concentrations increased by up to 3.6-fold over controls seven days after single injections of CMV, beta-actin, hsp70 and myoglobin promoter constructs. The optimal dose for most constructs was 100 micrograms plasmid DNA. Intramuscular plasmid injection into streptozotocin-induced diabetic Balb/c mice raised plasma insulin and C-peptide concentrations, and reduced hyperglycaemia. Two injections (100 micrograms plasmid DNA each) caused higher plasma insulin concentrations and significantly reduced hyperglycemia in diabetic mice than a single injection. Best results were obtained when plasmid injections preceded induction of diabetes by 14 days. CONCLUSIONS: Skeletal muscle is a potentially useful platform for ectopic secretion of insulin using naked plasmid as a gene transfer vector. Injection at two sites 14 days before the onset of severe hyperglycemia is optimal. This approach could protect Type I diabetics from fatal ketoacidosis and enhance the action of agents that sensitize tissues to insulin in type II diabetes.

Stable expression of human alpha-2,6-sialyltransferase in Chinese hamster ovary cells: functional consequences for human erythropoietin expression and bioactivity.

Hamster cell lines are common hosts for recombinant protein production, e.g. erythropoietin (Epo). Terminal sialylation of native human proteins is characteristically in both alpha-2,3 and alpha-2,6 linkage to galactose at the termini of N-linked oligosaccharides but only in alpha-2,3 linkage in recombinant proteins expressed in hamster cells which do not express alpha-2, 6-sialyltransferase (ST6GalI) (EC 2.4.99.1). This difference could alter the bioactivity of certain recombinant proteins. Chinese hamster ovary (CHO) cells stably transfected with human ST6GalI cDNA linked to the hamster metallothionein II promoter expressed highly inducible authentic ST6GalI activity. Untransfected CHO cells and CHO cells stably expressing ST6GalI cDNA when transfected with a human Epo cDNA expression cassette secreted immunoreactive Epo. Human Epo from singly transfected Pro-5 CHO cells induced significant reticulocytosis (7.00+/-1.58%; mean+/-S.D. % reticulocytes; control conditioned medium 3.04+/-1.29%; P<0.0024), whereas Epo from Pro-5 cells coexpressing ST6GalI elicited a more modest reticulocytosis (4.62+/-1.02%). Thus for recombinant human Epo, engineering CHO cells to express ST6GalI activity does not enhance Epo bioactivity in vivo in mice. The availability of CHO cells that express high levels of ST6GalI activity now enables systematic studies to determine the functional requirement for this form of sialylation in recombinant human proteins.

Human L-ficolin: plasma levels, sugar specificity, and assignment of its lectin activity to the fibrinogen-like (FBG) domain.

Ficolins are characterised by the presence of collagen-like and fibrinogen-like (FBG) sequences. Human L-ficolin is synthesised in the liver and secreted into blood circulation. In previous studies, it was shown to bind to N-acetyl-D-glucosamine (GlcNAc). In the present study, its detailed sugar specificity and binding site have been investigated. It was found to bind to GlcNAc and GalNAc (N-acetyl-D-galactosamine) while showing no significant affinity for the precursor sugars. The structure in these molecules which is recognised by L.-ficolin has been deduced to include an amide (-CO-NH-) or similar group. L-Ficolin was digested with collagenase and the collagenase resistant FBG domain was shown to bind to GlcNAc. Its levels in adult and cord blood-derived human plasma were also determined and showed that adult plasma contains approximately three times more L-ficolin than that of newborn babies.

Complexes of salicylaldehyde acylhydrazones: cytotoxicity, QSAR and crystal structure of the sterically hindered t-butyl dimer.

A series of acylhydrazones of salicylaldehyde and their transition metal complexes, predominantly copper(II), have been prepared and characterized. The crystal structure of the Cu(II) complex of the sterically hindered t-butyl derivative contains a phenolato bridged dimer with the ligand coordinated as a tridentate moiety. QSAR analyses of the cytotoxicity of the chelators and their Cu(II) complexes reveals that solubility is the dominant factor for activity. Compounds display a maximum with respect to lipophilicity, allowing optimization of the bioactivity for both the ligands and their complexes. Copper complexes are significantly more cytotoxic than the metal-free ligands and complexes of other metals: Cu > Ni > Zn = Mn > Fe = Cr > Cr > Co.

Purification and binding properties of a human ficolin-like protein.

Ficolin was initially identified from porcine uterus as a TGF-beta 1 binding protein and is considered to have an overall structure similar to that of the complement protein C1q and the collectins. Recent studies have shown that human ficolin is synthesized mainly by monocytes in peripheral blood and that it could potentially bind to sugar structures on microorganisms. The aim of the present investigations was to isolate ficolin from human plasma by affinity chromatography on immobilized sugars. A human serum protein was identified in the GlcNAc eluate from GlcNAc-Sepharose which migrated as a polypeptide of approx. 40 kDa on SDS-PAGE under reducing conditions and was, after further purification by FPLC on a mono-Q column, shown to have an identical N-terminal sequence, over the first 14 residues, to P35, a plasma protein having similar sequence and domain organisation to ficolin. This protein, named the ficolin-like protein, was shown to be sensitive to collagenase and similar to P35 in that it was also disulphide-linked into an oligomer of approx. 320 kDa. However, unlike P35, its binding to GlcNAc was independent of Ca2+. Gel-filtration studies showed that this ficolin-like protein also had a molecular weight of approx. 320 kDa under non-dissociating conditions. During the course of this study this ficolin-like protein was found to simply bind to CNBr-activated Sepharose which had been inactivated with Tris, and from which it could be eluted with GlcNAc. This ficolin-like protein was also shown to bind to GlcNAc, but not to mannose and maltose. The functional significance of the unusual binding property of this ficolin-like protein is not clear, but it has facilitated the development of a simple method for its purification.

Improvements on the purification of mannan-binding lectin and demonstration of its Ca(2+)-independent association with a C1s-like serine protease.

Mannan-binding lectin (MBL), previously called 'mannan-binding protein' or MBP, is a plasma C-type lectin which, upon binding to carbohydrate structures on micro-organisms, activates the classical pathway of complement. Purification of MBL relies on its Ca(2+)-dependent affinity for carbohydrate, but existing methods are susceptible to contamination by anti-carbohydrate antibodies. In the present study a sequential-sugar-elution method has been developed which can achieve a preparation of virtually pure MBL and its associated serine protease (MBL-associated serine protease, MASP) by two steps of affinity chromatography. In further separation of MASP from MBL, it was found that activated MASP was associated with MBL independent of Ca2+. Since MBL was found to bind to underivatized Sepharose 4B, the MBL-MASP complex was purified using Sepharose 4B and protease inhibitors were included to purify the complex with MASP in its proenzyme form. Analysis of thus-purified MBL-MASP complex by gel filtration on a Sephacryl S-300 column at pH 7.8 showed that the proenzyme MASP was also associated with MBL independently of Ca2+, but that the complex could be disrupted at a low pH (5.0). Therefore the mechanism of MBL-MASP-mediated complement activation appears to be significantly different from the C1-mediated classical pathway.

Biosynthesis of human ficolin, an Escherichia coli-binding protein, by monocytes: comparison with the synthesis of two macrophage-specific proteins, C1q and the mannose receptor.

Ficolin is characterized by the presence of both collagen-like and fibrinogen-like sequences, and potentially has a similar overall structure as the complement protein C1q and the collectins. Previous studies have reported the presence of human ficolin mRNA predominantly in peripheral blood leucocytes. In the present study, the cellular origin of human ficolin was investigated in further detail. Preliminary studies using reverse transcriptase-polymerase chain reaction (RT-PCR) showed that ficolin mRNA was synthesized by U937 cells, a human monocyte cell line. This finding suggested that blood monocytes also normally synthesize human ficolin. Peripheral blood monocytes from adult human donors were harvested at serial time-points (0-20 hr) after adhesion to tissue culture plates, and total RNA was isolated and assayed for ficolin mRNA by RT-PCR. Ficolin mRNA was highly expressed in monocytes throughout the first 20 hr of adhesion. In contrast, C1q and mannose receptor mRNA were not detectable during the first 8 hr of adhesion, but were highly expressed by 20 hr. Cells were harvested at longer time intervals (1, 2, 4, 6 and 8 days) to determine whether ficolin expression was temporally regulated at later stages of monocyte differentiation. Ficolin mRNA levels decreased sharply from day 1 to day 6. In contrast, the levels of both C1q and mannose receptor mRNA showed no changing trend. These results are consistent with the absence of ficolin expression in many macrophage-rich tissues previously reported. The origin of ficolin from monocytes, together with its structural similarity to C1q and the collectins, raises the possibility that ficolin may be another plasma protein capable of binding to surface structures of micro-organisms. Escherichia coli was therefore incubated with human serum, and bound proteins, after elution with sugars, were analysed by Western blotting using an antiserum raised against a synthetic ficolin peptide. The antiserum identified a polypeptide of approximately 42000 MW, which is similar in size to that of ficolin as predicted from its cDNA-derived sequence.

Human ficolin: cDNA cloning, demonstration of peripheral blood leucocytes as the major site of synthesis and assignment of the gene to chromosome 9.

Pig ficolins and a number of other proteins contain sequences that are homologous to the C-terminal halves of fibrinogen beta- and gamma-chains. To clone the cDNA for human ficolin, two degenerate oligonucleotide primers were synthesized, based on two stretches of protein sequence that were highly conserved among those proteins, and used for PCR with cDNA from a human uterus lambda gt11 library as a template. A PCR product with a predicted size of 300 bp was obtained and this was used to screen a uterus cDNA library. Of the positive clones isolated, two (U1 and U2), containing inserts of 1.7 and 1.1 kb respectively, were found to encode human ficolin. The cDNA-derived amino acid sequence of human ficolin has approx. 75% identity with, and a similar domain organization to, the two pig ficolin sequences, which are characterized by the presence of a leader peptide, a short N-terminal segment followed by a collagen-like region and then by a C-terminal fibrinogen-like domain. The 1.1 kb insert of clone U2 was used in Northern-blot analysis, and a very strong signal for a 1.4 kb mRNA species was detected in mRNA from human peripheral blood leucocytes. This showed that, despite the initial characterization of pig ficolin as a putative receptor on uterine cells for transforming growth factor beta 1, blood leucocytes are probably the major site of human ficolin synthesis. Much weaker signals of the same size were also detected in spleen, lung and thymus and may be due to the presence of tissue macrophages or trapped blood in these tissues. An mRNA species of approx. 1.3 kb in human liver also weakly hybridized to the U2 probe, indicating the presence of a sequence that was distinct from, but related to, ficolin. The gene for human ficolin has been mapped to chromosome 9.

Selective antiproliferative effects of thymidine.

A substance with antiproliferative bioactivity in an aqueous extract of Cordyline terminalis was purified and identified by mass spectrometry to be the natural nucleoside, thymidine. 10(-5) M Thymidine inhibited EL4 cell replication and decreased cell viability after 12-24 h. The effect was highly specific for this nucleoside. Treated cell cultures showed a significant increase in S phase cells and a corresponding decrease in G1 phase cells. Nitrobenzylthioinosine (which prevented facilitated entry of thymidine) protected cells from the antiproliferative action of thymidine. A human breast cancer cell line (MCF7) was also growth-inhibited by 10(-5) M thymidine but a murine lymphoma cell line (K36) was not. Thus, submillimolar thymidine has effects on cell proliferation which are selective both with respect to specificity for the compound and for different tumour cell lines.

Synthesis of 2-(p-chlorobenzyl)-3-aryl-6-methoxybenzofurans as selective ligands for antiestrogen-binding sites. Effects on cell proliferation and cholesterol synthesis.

A series of nonsteroidal compounds, 2-(p-chlorobenzyl)-3-aryl-6- methoxybenzofurans derived from the 2-(p-chlorobenzyl)-6-methoxy-3(2H)-benzofuranones has been synthesized. The key steps in the synthesis were reactions of 2-(p-chlorobenzyl)-6-methoxy-3(2H)-benzofuranones with the arylorganometallic reagents followed by dehydration of the resulting carbinols. The benzofurans are ligands for antiestrogen-binding sites (AEBS) and display no significant interaction with the estrogen receptor (ER). All bind to AEBS with equivalent or greater affinity than tamoxifen. These compounds decrease [3H]thymidine incorporation in AEBS-containing EL4 lymphoid cells and MCF7 breast cancer cells in a concentration-dependent manner between 10(-8) and 10(-6) M and are generally more inhibitory than tamoxifen. In contrast, they have no effect on [3H]thymidine incorporation by an AEBS-deficient variant of the MCF7 cell line, RTx6. The present findings of (1) selective and high affinity binding of the benzofurans to AEBS, (2) their concentration-dependent inhibition of [3H]thymidine incorporation in AEBS-containing cells, and (3) their lack of antiproliferative effect in an AEBS-deficient cell line suggest a functional role for AEBS in mediating the antigrowth effect of these compounds. Two of the more active benzofuran compounds also significantly inhibited de novo cholesterol biosynthesis in EL4 cells which lack ER. This effect could be obtained after 5 h of treatment and preceded significant loss of cell viability. This is the first demonstration that selective ligands of AEBS (other than the known nonsteroidal antiestrogens) interfere with cholesterol biosynthesis-an action that may contribute to their antigrowth effect.

Fatty acid modulation of antiestrogen action and antiestrogen-binding protein in cultured lymphoid cells.

Nonsteroidal antiestrogens reversibly and specifically inhibited the proliferation of two estrogen receptor-negative lymphoid cell lines (EL4 and Raji) in a dose-dependent manner. [3H]Thymidine incorporation of concanavalin A-stimulated primary splenocytes was also inhibited by 10(-6) M clomiphene (1-[4-(2-diethylaminoethoxy)phenyl]-1,2-diphenyl-2-chloroethylene). The antiproliferative effect could be prevented by the simultaneous presence in the growth medium of 10(-5) M linoleic acid or 10(-5) M arachidonic acid but not by 10(-6) M estradiol. Both lymphoid cell lines had high affinity antiestrogen-binding sites whose affinity could be altered by conditions of growth. Growth of EL4 cells in RPMI 1640 medium supplemented with charcoal-pretreated 5% fetal calf serum (charcoal-stripped medium) resulted in significantly higher affinity (Kd 0.54 nM +/- 0.11 nM; n = 6) than growth in medium supplemented with untreated serum (complete medium) (Kd = 1.68 nM +/- 0.48 nM; n = 6) (p less than 0.001). This change in affinity was partly due to removal of fatty acids from the growth medium by charcoal pretreatment, since addition of 10(-5) M linoleic acid or 10(-5) M gamma-linolenic to charcoal-stripped medium decreased the affinity of the antiestrogen-binding protein. In contrast, growth in 10(-5) M stearic acid or 10(-5) M oleic acid did not significantly alter the affinity of the antiestrogen-binding protein, whereas 10(-5) M palmitic acid significantly increased its affinity. The same fatty acids were also tested for their intrinsic effects on EL4 cell proliferation. Oleic, linoleic and gamma-linolenic acids were growth stimulatory while stearic and palmitic acids were not. Thus linoleic and gamma-linolenic acids whose presence in the growth medium was associated with decreased affinity of [3H]tamoxifen (1-[4-(2-dimethylaminoethoxy)phenyl]-1,2-diphenylbut-1(Z)-ene) binding to the intracellular antiestrogen-binding protein were also growth stimulatory. Unsaturated fatty acids have previously been shown to inhibit binding of [3H]tamoxifen to the antiestrogen-binding protein in a cell-free system. The present observations demonstrate that unsaturated fatty acids also modify the affinity of the antiestrogen-binding protein in intact cells.

Cytostatic effect of antiestrogens in lymphoid cells: relationship to high affinity antiestrogen-binding sites and cholesterol.

The antiproliferative effect of 10(-6) M antiestrogens in an estrogen receptor-negative lymphoid cell line (K36) was enhanced in lipoprotein-poor growth medium. The enhancement was not due to increased bioavailability because cellular uptake of [3H]tamoxifen was not increased and the lipoprotein fraction of serum had negligible [3H]tamoxifen-binding capacity. Cholesterol and lipoproteins, but not mevalonate, reversed the cytostatic effect of antiestrogens. Reversal by cholesterol was dose-related (10(-7) M to 10(-5) M), while that by lipoproteins could also be demonstrated in medium undepleted of lipoproteins. The cytostatic efficacy of a series of ten compounds correlated well with their relative binding affinities for solubilized antiestrogen-binding sites from K36 cells when log IC50 values (concentration required to reduce [3H]thymidine incorporation by 50%) were plotted against log RBA50 values (concentration required to reduce [3H]tamoxifen binding by 50%) (correlation coefficient 0.94). Transmission electron microscopy of antiestrogen-treated cells showed evidence of disordered cytokinesis which was partially reversed by cholesterol. These observations implicate the antiestrogen-binding protein in the antiproliferative effect of antiestrogens in nonestrogen target cells.

Estrogens, antiestrogens and cell proliferation.

The classical estrogen receptor model does not sufficiently account for the tumor-promoting activity of estrogens or for the antiproliferative effect of anti-estrogens in estrogen-dependent tumors. Particular difficulties not readily accommodated within the model are that hormonal autonomy can supervene without loss of the estrogen receptor and that antiestrogen effects are highly context-dependent, without apparent differences in the estrogen receptor itself or in metabolic transformation of antiestrogens. Recent studies suggest that estrogens may promote cell proliferation, in part, through the mediation of growth factors and that antiestrogens may exert some of their effects by mechanisms unrelated to the estrogen receptor.

Murine antiestrogen-binding protein: characterization, solubilization and modulation by lipids.

The properties of the antiestrogen-binding protein have been examined in mouse tissues, a species in which nonsteroidal antiestrogens are virtually pure agonists. As in other species studied, this protein was distributed in all tissues - highest levels being in the liver. Subcellular fractionation of mouse liver showed that 82% of the antiestrogen-binding protein was associated with the rough endoplasmic reticulum where it was confined to the membranous component. The antiestrogen-binding protein was also present in smooth endoplasmic reticulum, nuclei and cytosol. Its concentration in intact nuclei was at least 10-times higher than levels previously reported in intact rat liver nuclei. Binding of [3H]tamoxifen to the murine antiestrogen-binding protein was of high affinity (Kd = 1 nM) and was inhibited by unsaturated fatty acids and 7-ketocholesterol. In general, cis-isomers of unsaturated fatty acids were more effective binding inhibitors than trans-isomers. The antiestrogen-binding protein solubilized from rough endoplasmic reticulum membranes by the zwitterionic detergent CHAPS, had a molecular mass of approx. 700 kDa and a sedimentation coefficient of about 19 S. [3H]Tamoxifen binding capacity of the solubilized protein was abolished by trypsin and nonspecific proteinases but not by clostripain or Staphylococcus aureus V8 proteinase, suggesting that lysine residue(s) may be involved in [3H]tamoxifen binding.

Human endometrial antioestrogen-binding protein: characterization and levels during the menstrual cycle.

Human endometrial cytosol (90,000 g, 60 mins) contains a protein that binds non-steroidal anti-oestrogens with high specificity and high affinity (Kd = 10(-10)M). It binds oestrogens and other steroid hormones poorly, and can be physically distinguished from the oestrogen receptor protein by gel filtration and ion exchange chromatography. The endometrial anti-oestrogen binding protein is larger and less negatively charged than the oestrogen receptor. In this regard, it resembles the human myometrial anti-oestrogen binding protein. Unlike the myometrial protein, however, levels of the endometrial anti-oestrogen binding protein vary during the menstrual cycle, being higher in the proliferative phase (mean 29 fmol/mg protein, SD 13 fmol/mg protein) than in the secretory phase (mean 13 fmol/mg protein, SD 7 fmol/mg protein). These findings suggest that the concentration of the anti-oestrogen binding protein in human endometrium may be regulated, at least in part, by the cyclical pattern of hormonal secretion that characterizes the menstrual cycle.

Antiestrogen binding sites in rat liver nuclei.

Isolated, intact rat liver nuclei have high-affinity (Kd = 10(-9) M) binding sites that are highly specific for nonsteroidal antiestrogens, especially for compounds of the triphenylethylene series. Nuclear [3H]tamoxifen binding capacity is thermolabile, being most stable at 4 degrees C and rapidly lost at 37 degrees C. More [3H]tamoxifen, however, is specifically bound at incubation temperatures of 25 degrees C and 37 degrees C than at 4 degrees C although prewarming nuclei has no effect, suggesting exchange of [3H]tamoxifen for an unidentified endogeneous ligand. Nuclear antiestrogen binding sites are destroyed by trypsin but not by deoxyribonuclease I or ribonuclease A. The nuclear antiestrogen binding protein is not solubilized by 0.6 M potassium chloride, 2 M sodium chloride, 0.6 M sodium thiocyanate, 3 M urea, 20 mM pyridoxal phosphate, 1% (w/v) digitonin or 2% (w/v) sodium cholate but is extractable by sonication, indicating that it is tightly bound within the nucleus. Rat liver nuclear matrix contains high-affinity (Kd = 10(-9) M) [3H]tamoxifen binding sites present in 5-fold higher concentrations (4.18 pmol/mg DNA) than in intact nuclei (0.78 +/- 0.10 (S.D.) pmol/mg DNA). Low-speed rat liver cytosol (20 000 X g, 30 min) contains high-capacity (955 +/- 405 (S.D.) fmol/mg protein), low-affinity (Kd = 10.9 +/- 4.5 (S.D.) nM) antiestrogen binding sites. In contrast, high-speed cytosol (100 000 X g, 60 min) contains low-capacity (46 +/- 15 (S.D.) fmol/mg protein), high-affinity (Kd = 0.61 +/- 0.20 (S.D.) nM) binding sites. Low-affinity cytosolic sites constitute more than 90% of total liver binding sites, high-affinity cytosolic sites 0.3%-3.2%, and nuclear sites less than 0.5% of total sites.