Quantitative Electron Microscopic Autoradiography Research Articles

Internalization of ( 125I) alpha-bungarotoxin into rat suprachiasmatic nucleus neurons and dendrites

The distribution and rate of internalization of ( 125I) alpha-bungarotoxin (α-BuTX) was evaluated by quantitative electron microscopic autoradiography in rat suprachiasmatic nucleus. The ultrastructural distribution of silver grains was studied by both line source and 50% probability circle analyses. Line source analysis demonstrated that with increasing time following intracerebroventricular infusion, ligand is internalized from neuronal membranes. At all time points, real grain distribution was significantly different from randomly generated hypothetical grains. Quantification of silver grain localization by probability circle analysis indicated that membrane bound sources were mainly associated with axo-dendritic appositions, regardless of the length of time the tissue was exposed to radioligand. The compartment containing synaptic terminals was the most enriched when comparing real to hypothetical grains. By eight hours after intracerebroventricular infusion of specifically labelled α-BuTX binding sites were most likely to be within neurons and dendrites. These studies demonstrate that the majority of α-BuTX binding sites remain membrane bound with respect to time and may be associated with synaptic transmission; a significant proportion of silver grains are internalized within SCN neurons and dendrites.

Selective Uptake of [3H]Arachidonic Acid Into the Dense Tubular System of Human Platelets

We have used quantitative electron microscopic autoradiography to characterize the subcellular distribution of arachidonoyl phospholipids following brief (5 minutes) exposure of unstimulated human platelets to [3H]arachidonic acid. Labeled arachidonate was taken up rapidly and incorporated into phospholipids. Phospholipid radioactivity was preserved and spatially fixed during tissue processing for electron microscopy. Analysis of autoradiographs showed that following a brief exposure to 750 nmol/L [3H]arachidonate, there is selective labeling of an internal membrane compartment composed of the dense tubular system and the open canalicular system. The plasma membrane, platelet granules, and nonmembranous cytoplasm were not labeled. Since the open canalicular system is continuous with the plasma membrane and since phospholipids in continuous membranes are freely diffusible, our observations indicate that [3H]arachidonate was incorporated into phospholipids within the dense tubular system and not the open canalicular system. Thus, the dense tubular system, known to contain cyclooxygenase activity, incorporates arachidonate selectively following brief exposure to this fatty acid, presumably to concentrate it in proximity to enzymes for icosanoid synthesis.

Selective uptake of [3H]arachidonic acid into the dense tubular system of human platelets

We have used quantitative electron microscopic autoradiography to characterize the subcellular distribution of arachidonoyl phospholipids following brief (5 minutes) exposure of unstimulated human platelets to [3H]arachidonic acid. Labeled arachidonate was taken up rapidly and incorporated into phospholipids. Phospholipid radioactivity was preserved and spatially fixed during tissue processing for electron microscopy. Analysis of autoradiographs showed that following a brief exposure to 750 nmol/L [3H]arachidonate, there is selective labeling of an internal membrane compartment composed of the dense tubular system and the open canalicular system. The plasma membrane, platelet granules, and nonmembranous cytoplasm were not labeled. Since the open canalicular system is continuous with the plasma membrane and since phospholipids in continuous membranes are freely diffusible, our observations indicate that [3H]arachidonate was incorporated into phospholipids within the dense tubular system and not the open canalicular system. Thus, the dense tubular system, known to contain cyclooxygenase activity, incorporates arachidonate selectively following brief exposure to this fatty acid, presumably to concentrate it in proximity to enzymes for icosanoid synthesis.

A combined Golgi and autoradiographic study of 3H-glycine-accumulating cone bipolar cells in the cat retina

Morphological subclasses of cone bipolar cells in the cat retina were studied in Golgi preparations and through the combination of Golgi impregnation with quantitative electron-microscopic autoradiography of 3H-glycine accumulation. Camera lucida drawings, computer-assisted rotations, and radial sections provided the morphological details used in assigning individual cells to morphologically defined subclasses. Selected Golgi-impregnated cells from retinas preloaded with 3H-glycine were sectioned for autoradiographic analysis. Grain densities were quantitated and compared on a normalized scale with the labeling of amacrine cells with 3H-glycine. Eleven types of cone bipolar cells were identified, 6 with axon terminals ramifying in sublamina a and 5 with axons terminating in sublamina b. Of these cells, types CBb2 and CBb5 showed the highest affinity for glycine, with labeling at 41 and 38% with respect to the most heavily labeled amacrine cells. Types CBb1, CBb3, CBa2, and CBa6 were labeled at approximately 20% of maximal labeling. The accumulation of glycine by these cells suggests that they may be involved in providing inhibitory input to the inner retina.

Turnover of heparan sulfate proteoglycan in human colon carcinoma cells. A quantitative biochemical and autoradiographic study.

The metabolism of heparan sulfate proteoglycan, a major product of human colon carcinoma cells, was investigated in a series of pulse-chase experiments using a combination of quantitative biochemistry and electron microscope autoradiography. This was possible primarily because these cells incorporate [35S]sulfate exclusively into heparan sulfate proteoglycan, thus allowing the possibility of correlating the two sets of information. The results showed a progressive movement of the newly synthesized proteoglycan from the Golgi to the cell surface, where it became closely associated with the plasma membrane and was labeled ultrastructurally by both ruthenium red and radiosulfate. Subsequently, about 55% was released into the medium (t1/2 approximately 2.5 h) where it resided as intact macromolecule and was neither endocytosed nor degraded further. The remaining 45% was internalized and converted into smaller species through a series of degradative steps. Initially (Step 1) there was proteolytic cleavage of the protein core and partial endoglycosidic cleavage of the heparan sulfate chains (t1/2 approximately 6 h), with generation of larger glycosaminoglycan-peptide intermediates with chains of Mr approximately 10,000, about one-third their original size. These components were subsequently converted (Step 2) to yet smaller, limiting fragments of Mr approximately 5,000, which were finally depolymerized (Step 3) with quantitative release of free sulfate. The intracellular degradation of the proteoglycan, particularly Steps 2 and 3, was markedly inhibited by choloroquine, implicating the involvement of acidic compartments in the catabolism of these macromolecules. This was corroborated by the autoradiographic studies which showed the close association of 35S-labeled products with secondary lysosomes. However, the initial degradation of the proteoglycan might have occurred in a prelysosomal compartment since Step 1 was not totally blocked by chloroquine. The combined results indicate that the intracellular degradation of heparan sulfate follows structural as well as functional compartmentalization and provide a model that may be shared by other cell systems.

Origin and function of tomato bushy stunt virus-induced inclusion bodies: II. Quantitative electron microscope autoradiography and immunogold cytochemistry

The possible role of two types of inclusion bodies induced by tomato bushy stunt virus in Gomphrena globosa cells was investigated by electron microscope autoradiography after administration of tritiated uridine and by immunogold labeling on ultrathin sections. Our observations indicate that both multivesicular bodies (MVBs) and associated chloroplast invaginations are involved in virus replication. Since MVBs are always present in infected cells, even in those not containing chloroplasts, they appear to be the main site of virus RNA synthesis, while chloroplast invaginations are likely to provide, when present, an additional vesicular site for the same purpose. Neither of these structures contains viral protein. By contrast, the dense granules are not involved in virus RNA synthesis and consist mainly of virus coat protein, whose production is likely to occur in the cytoplasm.

Pituitary Binding and Internalization of Radioiodinated Gonadotropin-Releasing Hormone Agonist and Antagonist Ligandsin Vitroandin Vivo

In rat pituitary gonadotrophs, the rates of binding and endocytosis of two GnRH superagonist analogs, [D-Ala6,Pro9-NEt]GnRH and [D-Lys6,Pro9-NEt]GnRH, were compared with those of the potent antagonist analog [N-acetyl-D-pCl-Phe1,2,D-Trp3,D-Lys6,D-Ala10]GnRH by quantitative electron microscopic autoradiography. In dispersed pituitary cells, the two agonist analogs showed similar binding kinetics and comparable degrees of sequestration, as measured by their resistance to dissociation by low pH buffer. However, quantification of silver grain localization suggested that cellular internalization of the [D-Ala6]GnRH agonist increased more rapidly than that of the [D-Lys6]GnRH analog. These discrepancies, and the finding that a larger amount of the specifically bound 125I-[D-Ala6]GnRH agonist was removed during glutaraldehyde fixation, indicated that the proportional internalization of this analog was over estimated by quantitative autoradiography owing to loss of cell surface-bound radioligand. We, therefore, employed radioiodinated D-Lys6-substituted analogs to analyze the receptor binding and cellular uptake of GnRH agonist and antagonist derivatives in vivo. After iv injection, a high proportion of the 125I-[D-Lys6]GnRH agonist was translocated into pituitary gonadotrophs within 60 min, whereas the D-Lys6 antagonist was predominantly associated with the plasma membrane during that time. Four hours after injection of the antagonist, an appreciable proportion of silver grains was associated with intracellular organelles, and this trend increased progressively at later time points. The relatively prolonged cellular processing of the GnRH antagonist is consistent with in vivo binding kinetics, and its slower internalization may reflect the basal rate of GnRH receptor turnover in the cell membrane. In addition, the marked difference between the rates of internalization of the bound [D-Lys6]GnRH agonist and antagonist ligands supports the proposal that receptor activation is responsible for the rapid endocytosis of agonist ligands by the GnRH-stimulated pituitary gonadotroph.

Internalization and recycling of 125I-photoreactive insulin-receptor complexes in hepatocytes in primary culture

When the insulin receptor is tagged with a 125I-photoreactive insulin analogue that can be covalently coupled to it by UV irradiation, the fate of this labeled receptor can be followed both morphologically and biochemically. In the present study we have applied this tool to trace the pathway followed by 125I-photoreactive insulin-receptor complex in hepatocytes in primary culture. As determined by quantitative electron microscopic autoradiography, the internalized labeled material first associates with clear vesicles, second is found in multivesicular bodies, third associates with dense bodies and fourth returns to the cell surface via clear vesicles. This recycling process is inhibited by lysosomotropic agents, i.e. NH 4Cl or chloroquine. These data confirm, in another cell system, our previous observations carried out in freshly isolated rat hepatocytes and demonstrate the feasibility and complementarity of both freshly isolated hepatocytes and hepatocytes in primary culture to study internalization and recycling of the insulin receptor.

Role of intracellular calcium and protein kinase C in the endocytosis of transferrin and insulin by HL60 cells.

The role of the cytosolic free calcium concentration ([Ca2+]i) and of protein kinase C on the internalization of transferrin and insulin in the human promyelocytic cell line HL60 was investigated. [Ca2+]i was selectively monitored and manipulated by the use of the fluorescent Ca2+ indicator and buffer quin2, while receptor-ligand internalization was studied directly by quantitative electron microscope autoradiography. Decreasing the [Ca2+]i up to 10-fold below resting level had no effect on the internalization of transferrin or insulin. Similarly, a 10-fold elevation of the [Ca2+]i using the calcium ionophore ionomycin caused little or no change in the endocytosis of the two ligands. In contrast, activation of protein kinase C by phorbol myristate acetate markedly stimulated the internalization of both occupied and unoccupied transferrin receptors, even in cells with very low [Ca2+]i. The insulin receptor was found to behave differently in response to phorbol myristate acetate, however, in that only the occupied receptors were stimulated to internalize. We conclude that the [Ca2+]i plays only a minor role in regulating receptor-mediated endocytosis, whereas protein kinase C can selectively modulate receptor internalization depending on receptor type and occupancy.

Intracellular potassium depletion in IM-9 lymphocytes suppresses the slowly dissociating component of human growth hormone binding and the down-regulation of its receptors but does not affect insulin receptors.

We have investigated whether the slowly dissociating component of insulin and human growth hormone (hGH) binding and the homologous down-regulation of their receptors in IM-9 cultured human lymphocytes are due to distinct conformations of the receptor or, rather, to a redistribution within the cell. To do so, we used intracellular K+ depletion, which has been shown to inhibit reversibly coated-pit formation and ligand internalization in some cell lines. IM-9 cells incubated in K+-free buffer after a hypotonic shock rapidly lost their K+, which was stabilized at +/- 50% of control by incubation in K+-free binding assay buffer. In K+-depleted cells, the hGH dissociation kinetics became monoexponential and, in contrast with control cells, compatible with the equilibrium constant derived from saturation and association data using a simple model. The loss of hGH receptors during competition studies was abolished. The down-regulation by unlabeled hGH was decreased by 80%. In contrast, insulin receptor kinetics remained unchanged (non-first-order) in the K+-depleted cells; the negative cooperativity and the down-regulation (60%) were identical to those of control cells. Quantitative electron microscopic autoradiography showed a decrease of +/- 50% in the fraction of 125I-labeled hGH internalized. The number of visible coated pits in the membrane was reduced by 80%. Thus, in IM-9 cells, association with coated pits and endocytosis appear to play a major role in the kinetics of hGH binding and in the down-regulation of its receptors, but not in insulin-receptor binding kinetics and down-regulation.

A concentration of fucosylated glycoconjugates at the base of cone outer segments: Quantitative electron microscope autoradiography

Rod photoreceptors show a discrete band of labeled molecules at the outer segment base shortly after the administration of radioactive protein precursors (Young, 1967). The band signifies the insertion of radiolabeled protein, primarily opsin, into rod disc membrane (Basinger, Bok and Hall, 1976). In contrast, the autoradiographic labeling pattern in cones is characteristically diffuse (Young, 1971), although cones as well as rods are now thought to replace their disc membranes continually (Anderson, Fisher, and Steinberg, 1978). Using grain count analysis in conjunction with electron microscope autoradiography, we demonstrate for the first time that the initial cone outer segment (COS) labeling pattern in the retinas of cone-dominant rodents is not random or diffuse after intraocular injection of [ 3H]-fucose — a specific glycoprotein precursor. Thirty minutes after [ 3H]-fucose injection, the Golgi complex is heavily labeled in both rods and cones. At 1·5 hr, labeling first appears in the periciliary region and in the basal portion of the COS. We detected a basal concentration of fucosylated molecules up to 12 hr after injection; thereafter, we found no differential distribution of labeled product molecules in COS. In rods, we observed no band of labeled protein at any of the time points sampled. The low level of labeling in rods is uniformly distributed between the basal and apical portion of the outer segments. Although the identity of the fucosylated material remains unknown, the early labeling of the Golgi complex — followed by labeling over the periciliary region and the basal portion of the outer segment — suggests that newly synthesized membrane protein is transported from sites of synthesis in the inner segment and inserted into disc membrane at the COS base. However, in contrast to rods, the protein is apparently capable of diffusing longitudinally throughout the interconnected membrane network.

Intracellular pathway followed by the insulin receptor covalently coupled to 125I-photoreactive insulin during internalization and recycling.

After it interacts with a specific receptor on the cell surface, insulin is internalized in its target cell by an adsorptive endocytotic process and eventually degraded in lysosomes. It was also recently shown that the initial surface interaction between the hormone and its receptor is followed by an internalization of the receptor, which later is recycled back to the cell surface. In the present study the insulin receptor was tagged with a 125I-photoreactive insulin analogue that can be covalently coupled to the insulin receptor by ultraviolet irradiation. Using this tool we could trace by quantitative electron microscope autoradiography the intracellular pathway followed by this labeled receptor. The quantitative analysis of the intracellular distribution of the labeled material as a function of incubation time at 37 degrees C supports the following sequence of events: association first with clear vesicles, second with multivesicular bodies, third with dense bodies, and fourth, a return to the cell surface via clear vesicles. This insulin receptor recycling process is inhibited by monensin but unaffected by cycloheximide.

The hypothalamo-hypophysial system of the frog Rana temporaria

The median eminence (ME) of the adult frog, Rana temporaria, was studied by means of electron microscopy including quantitative electron-microscopic autoradiography. In frogs captured in May and June numerous peptidergic neurosecretory fibres extending via the internal zone to the pars nervosa display large swellings containing few granules, mitochondria, neurotubules and cisternae of the smooth endoplasmic reticulum. In addition, few secretory globules up to 1.5 μm in diameter occur in these varicosities. In animals collected during the autumn period many of these neurosecretory swellings filled with neurosecretory granules and polymorphic inclusions resemble Herring bodies. Three types of granule-containing neurosecretory fibres were observed in the external zone (EZ) of the ME of adult R. temporaria. Peptidergic A1- and A2-type fibres are characterized by granules 150–220 nm and 100–160 nm in diameter, respectively. Monoaminergic fibres of type B with granules approximately 100 nm in diameter represent ∼ 50% of all neurosecretory elements in the EZ of the frog ME; ∼12% of the total number of granule-bearing axons in the EZ actively taking up radiolabelled 5-hydroxytryptophan are thought to be serotoninergic terminals. Neurosecretory terminals of all types and glial vascular endfeet establish direct contacts with the perivascular space of the primary portal capillaries. Some neurosecretory terminals are separated from the lumen of the third ventricle by a thin cytoplasmic lamella of tanycytes. The possible physiological significance of this structural pattern is discussed.

Intracellular Processing of Human Vs. Rat Immunoglobulin A in the Rat Liver

It is well established that in the rat, rat dimeric IgA is transported from blood to bile across rat liver parenchymal cells via a series of minute smooth membrane-limited vesicles. This pathway is unique from that taken by a number of other ligands, which are internalized for degradation, in that there appears to be little involvement of coated vesicles, multivesicular bodies and lysosomes. The transmembrane receptor for IgA, secretory component, is not recycled but is secreted in part with the ligand into bile and must be produced continuously within the liver cell. Several recent studies have suggested that the receptor for asialoglycoproteins, as well as the structures involved in its processing, may play an important role in IgA processing. It was noted, however, that in all of these studies human polymeric IgA1 was used in the rat model. Using purified rat and human IgA preparations, we have demonstrated by light and quantitative electron microscopic autoradiography, as well as by certain biochemical procedures, that the two ligands are processed quite differently from one another in the rat. Human IgA disappears from the plasma at a slower rate and is much less efficiently transported into the bile. In addition, up to as much as 30% of the human IgA is diverted to the lysosomal pathway. This diversion of human polymeric IgA may be related to either the association of a serine-linked oligosaccharide at the hinge region of the human polymeric IgA1 or that large polymers, often found in human IgA preparations may initiate secretory component receptor aggregation, which in turn, interferes with the normal physiological processing of the IgA molecule.

Mammotroph autoregulation: intracellular fate of internalized prolactin

As part of a study concerning the mechanism of mammotroph autoregulation, the intracellular fate of [ 125I]iodo rat PRL, internalized by mammotrophs, was determined by quantitative electron microscope autoradiography. Simple grain density (GD) analysis of mammotroph autoradiograms showed that 5 cellular compartments (plasma membrane, Golgi apparatus, secretory granules, nucleus and mitochondria) were significantly labeled and that the labeling pattern in these compartments changed with time. Inclusion of a 25-fold excess of unlabeled PRL inhibited the binding of radiolabeled hormone to mammotrophs by approximately 85%. At no time were any grains found associated with lysosomes and there was no evidence of degradation of the radiolabeled PRL as judged by SDS-polyacrylamide gel electrophoresis or cold perchloric acid precipitation of cell extracts or media samples. Delivery of internalized PRL to a number of specific intracellular organelles suggests that these organelles may serve in the autoregulation of PRL secretion.

Redistribution of 125I-Insulin on the Surface of Rat Hepatocytes as a Function of Dissociation Time

In the present experiments, we have correlated the distribution of 125I-insulin on the surface of rat hepatocytes with the dissociation of 125I-insulin from the cell. When 125I-insulin interacts with isolated rat hepatocytes at 15 degrees C, an increasing proportion of the bound ligand becomes nondissociable under the influence of acid pH (6.0), trypsin (0.5 mg/ml), or an excess of unlabeled insulin (10(-6) M). Under these conditions, only a small percentage of the labeled material is internalized as determined by quantitative electron microscope (EM) autoradiography. This progressive nondissociability of the ligand parallels its movement from microvilli to coated pits and its progressive concentration in these later surface specializations. These data suggest that receptors in different domains of the plasma membrane may have different dissociation rates for the ligand.

Plasma membrane biogenesis in the avian salt gland: A biochemical and quantitative electron microscopic autoradiographic study

The avian salt gland provides an ideal system for the study of plasma membrane (PM) biogenesis. Feeding ducklings 1% sodium chloride (salt stress) induces the secretory cells of the gland to synthesize large amounts of PM, which forms an extensive basolateral PM domain after 7-9 days of treatment. In the present study, the initial biosynthetic events following salt stress were investigated. In vivo studies using 3H-uridine indicated that increased rates of RNA synthesis could be detected by 2 hr after the beginning of salt stress and continued through at least 12 hr. Under in vitro conditions, increased rates of protein and glycoprotein synthesis (as monitored by 3H-leucine and 3H-fucose incorporation, respectively) were also detected after 2 hr and continued through 7-9 days. Increased levels of Na,K-ATPase, a specific secretory cell PM marker, were detected after 8 hr of treatment as monitored by specific activity and 3H-ouabain binding. Sodium dodecyl sulfate-polyacrylamide slab gel electrophoresis coupled with fluorography indicated that both 3H-leucine and 3H-fucose were incorporated into partially purified preparations of Na,K-ATPase isolated after 12 hr. Light microscopic autoradiographic analysis of pulse-chase experiments indicated that in secretory cells of 12-hr salt-stressed glands, 3H-leucine- and 3H-fucose-labelled products reached the cell periphery by 1-2 hr after the initial pulse. The incorporation of both tritiated precursors was predominantly associated with the secretory cells. Quantitative electron microscopic autoradiography indicated that 3H-leucine is initially taken up by elements of the rough endoplasmic reticulum (RER) and cytoplasm (5 min postpulse), subsequently transported to and concentrated within components of the Golgi apparatus (10 min of chase), and ultimately incorporated into all domains of the plasma membrane of secretory cells by 1-2 hr of chase. The data is consistent with a flow of newly synthesized membrane components from RER to Golgi to plasma membrane and is analogous to the pattern previously found for the synthesis and processing of PM proteins in a wide variety of cell types.

Quantitative assessment of GABA-uptake sites in the neural lobe by electron-microscopic autoradiography.

Distribution of (3H)GABA in the rat neural lobe was investigated 5 min after intracarotid administration using quantitative electron-microscopic autoradiography. Specificity of (3H)GABA-uptake was tested by pretreatment of control animals with nipecotic acid. It was concluded that, apart from a small fraction in the perivascular spaces, radioactivity was present exclusively in pituicytes. The results confirm and quantify earlier in-vitro observations; they are compared with recent immunocytochemical findings that reveal the presence of glutamate-decarboxylase-containing axons in the neural lobe. It is concluded that there may be GABAergic terminals that lack an uptake mechanism for exogenous transmitter. Nevertheless, (3H)GABA autoradiography is useful in demonstrating other functional components of GABAergic systems, i.e., glial cells.

Localization of lead in rat peripheral nerve by electron microscopy.

Lead intoxication in rats reliably produces segmental demyelination. Following a single intravenous injection of radioactive lead, localization of tracer was observed sequentially by quantitative electron microscopical autoradiography. The animals injected had been on a lead-containing diet for 70 days; as a result, the blood-nerve barrier was broken down and demyelination was proceeding. Six hours after a single dose, the lead was localized to the endoneurial space of the peroneal nerve, and 72 hours later, to the myelin membrane. Lead may exert a direct effect on the membrane and alter its stability both by altering the lipid content of the membrane and by directly interfering with the lamellar structure.

Uptake and subcellular distribution of [3H]arachidonic acid in murine fibrosarcoma cells measured by electron microscope autoradiography.

We have used quantitative electron microscope autoradiography to study uptake and distribution of arachidonate in HSDM1C1 murine fibrosarcoma cells and in EPU-1B, a mutant HSDM1C1 line defective in high affinity arachidonate uptake. Cells were labeled with [3H]arachidonate for 15 min, 40 min, 2 h, or 24 h. Label was found almost exclusively in cellular phospholipids; 92-96% of incorporated radioactivity was retained in cells during fixation and tissue processing. All incorporated radioactivity was found to be associated with cellular membranes. Endoplasmic reticulum (ER) contained the bulk of [3H]arachidonate at all time points in both cell types, while mitochondria, which contain a large portion of cellular membrane, were labeled slowly and to substantially lower specific activity. Plasma membrane (PM) also labeled slowly, achieving a specific activity only one-sixth that of ER at 15 min in HSDM1C1 cells (6% of total label) and one-third of ER in EPU-1B (10% of total label). Nuclear membrane (NM) exhibited the highest specific activity of labeling at 15 min in HSDM1C1 cells (twice that of ER) but was not preferentially labeled in the mutant. Over 24 h, PM label intensity increased to that of ER in both cell lines. However, NM activity diminished in HSDM1C1 cells by 24 h to a small fraction of that in ER. In response to agonists, HSDM1C1 cells release labeled arachidonate for eicosanoid synthesis most readily when they have been labeled for short times. Our results therefore suggest that NM and ER, sites of cyclooxygenase in murine fibroblasts, are probably sources for release of [3H]arachidonate, whereas PM and mitochondria are unlikely to be major sources of eicosanoid precursors.