Robust cross-links in molluscan adhesive gels: testing for contributions from hydrophobic and electrostatic interactions.

The cross-linking interactions that provide cohesive strength to molluscan adhesive gels were investigated. Metal-based interactions have been shown to play an important role in the glue of the slug Arion subfuscus (Draparnaud), but other types of interactions may also contribute to the glue's strength and their role has not been investigated. This study shows that treatments that normally disrupt hydrophobic or electrostatic interactions have little to no effect on the slug glue. High salt concentrations and non-ionic detergent do not affect the solubility of the proteins in the glue or the ability of the glue proteins to stiffen gels. In contrast, metal chelation markedly disrupts the gel. Experiments with gel filtration chromatography identify a 40 kDa protein that is a central component of the cross-links in the glue. This 40 kDa protein forms robust macromolecular aggregations that are stable even in the presence of high concentrations of salt, non-ionic detergent, urea or metal chelators. Metal chelation during glue secretion, however, may block some of these cross-links. Such robust, non-specific interactions in an aqueous environment are highly unusual for hydrogels and reflect an intriguing cross-linking mechanism.

The role of metals in molluscan adhesive gels.

Several gastropod molluscs produce glues that are interesting because they are dilute gels and yet they produce strong adhesion. Specific glue proteins have been identified that play a central role in this adhesion, possibly by crosslinking other polymers in the gel. This study investigates the role of metals in the action of these glue proteins. Atomic absorption spectrometry showed that glue from the slug Arion subfuscus contains substantial quantities of zinc (46+/-7 p.p.m. and 189+/-80 p.p.m. in two different sets of experiments) and also iron, copper and manganese (2-7 p.p.m.). Iron-specific staining demonstrates that iron is bound specifically to the 15 kDa glue protein. Several approaches were used to show that these metals have important functional effects. Adding iron or copper to dissolved glue causes the proteins to precipitate rapidly, although zinc has no effect. Removing iron and related transition metals with a chelator during secretion of the glue causes a sixfold increase in the solubility of the glue. Once the glue has set, however, removing these metals has no effect. Finally, the gel-stiffening activity of the glue proteins was measured in the presence and absence of the chelator. The chelator eliminated the gel-stiffening effect of the proteins, suggesting that transition metals were necessary for the proteins to act on the gel. Thus, the glue contains transition metals and these metals play an essential role in glue function.

Functionally important residues in the peptidyl-prolyl isomerase Pin1 revealed by unigenic evolution.

Pin1 is a phosphorylation-dependent member of the parvulin family of peptidyl-prolyl isomerases exhibiting functional conservation between yeast and man. To perform an unbiased analysis of the regions of Pin1 essential for its functions, we generated libraries of randomly mutated forms of the human Pin1 cDNA and identified functional Pin1 alleles by their ability to complement the Pin1 homolog Ess1 in Saccharomyces cerevisiae. We isolated an extensive collection of functional mutant Pin1 clones harboring a total of 356 amino acid substitutions. Surprisingly, many residues previously thought to be critical in Pin1 were found to be altered in this collection of functional mutants. In fact, only 17 residues were completely conserved in these mutants and in Pin1 sequences from other eukaryotic organisms, with only two of these conserved residues located within the WW domain of Pin1. Examination of invariant residues provided new insights regarding a phosphate-binding loop that distinguishes a phosphorylation-dependent peptidyl-prolyl isomerase such as Pin1 from other parvulins. In addition, these studies led to an investigation of residues involved in catalysis including C113 that was previously implicated as the catalytic nucleophile. We demonstrate that substitution of C113 with D does not compromise Pin1 function in vivo nor does this substitution abolish catalytic activity in purified recombinant Pin1. These findings are consistent with the prospect that the function of residue 113 may not be that of a nucleophile, thus raising questions about the model of nucleophilic catalysis. Accordingly, an alternative catalytic mechanism for Pin1 is postulated.

Structure of a conjugating enzyme-ubiquitin thiolester intermediate reveals a novel role for the ubiquitin tail.

BACKGROUND: Ubiquitin-conjugating enzymes (E2s) are central enzymes involved in ubiquitin-mediated protein degradation. During this process, ubiquitin (Ub) and the E2 protein form an unstable E2-Ub thiolester intermediate prior to the transfer of ubiquitin to an E3-ligase protein and the labeling of a substrate for degradation. A series of complex interactions occur among the target substrate, ubiquitin, E2, and E3 in order to efficiently facilitate the transfer of the ubiquitin molecule. However, due to the inherent instability of the E2-Ub thiolester, the structural details of this complex intermediate are not known. RESULTS: A three-dimensional model of the E2-Ub thiolester intermediate has been determined for the catalytic domain of the E2 protein Ubc1 (Ubc1(Delta450)) and ubiquitin from S. cerevisiae. The interface of the E2-Ub intermediate was determined by kinetically monitoring thiolester formation by 1H-(15)N HSQC spectra by using combinations of 15N-labeled and unlabeled Ubc1(Delta450) and Ub proteins. By using the surface interface as a guide and the X-ray structures of Ub and the 1.9 A structure of Ubc1(Delta450) determined here, docking simulations followed by energy minimization were used to produce the first model of a E2-Ub thiolester intermediate. CONCLUSIONS: Complementary surfaces were found on the E2 and Ub proteins whereby the C terminus of Ub wraps around the E2 protein terminating in the thiolester between C88 (Ubc1(Delta450)) and G76 (Ub). The model supports in vivo and in vitro experiments of E2 derivatives carrying surface residue substitutions. Furthermore, the model provides insights into the arrangement of Ub, E2, and E3 within a ternary targeting complex.

The role of androgen receptors in the clinical course of nevus sebaceus of Jadassohn.

Nevus sebaceus of Jadassohn (NSJ) is a benign, congenital hamartoma that often presents at birth, appears to regress in childhood, and grows during puberty, suggesting possible hormonal control. We studied 18 cases of NSJ from children and adults for immunohistochemical evidence of androgen receptor expression. The lesions were evaluated for location and pattern of immunostaining, and these findings were compared between age groups, sexes, and to androgen receptor expression in normal skin. Androgen receptor positivity was seen in the sebaceous glands, in eccrine glands with and without apocrine change, and rarely in keratinocytes in the sebaceous nevi. There were no significant differences in staining location or pattern between the age groups or sexes. Normal skin showed similar staining in the sebaceous glands but did not show staining of the eccrine glands or keratinocytes. Androgen receptors are present in all epithelial components of NSJ, but there is no change in androgen receptor expression during puberty.

Identification of the ubiquitin interfacial residues in a ubiquitin-E2 covalent complex.

One of the key intermediates formed during the protein ubiquitination cycle is a covalent complex between ubiquitin (Ub) and the conjugation enzyme, UBC1. In order to probe the interface between these two proteins we have formed the covalent complex in situ (in the NMR tube) using Ub, the catalytic domain of UBC1, UBC1 delta450, an activation enzyme, E1, and Mg2+-ATP. The size of the Ub-UBC1 delta450 complex (25 kDa) and its relatively short lifetime (approximately 4 h) makes assignment of the backbone resonances in the covalent species difficult. In order to monitor the formation and identify the interface in the complex we have used fast 1H-15N HSQC spectra to monitor the decay of 1H-15N correlations as a function of time until the complex formed reached about 90%. The residual peak intensities were used to probe the surface of interaction between Ub and UBC1 delta450 and provided a clear surface of interaction on Ub.

Effect of fast duration on disposition of an intraduodenal glucose load in the conscious dog.

UNLABELLED: The effects of prior fast duration (18 h, n = 8; 42 h, n = 8) on the glycemic and tissue-specific responses to an intraduodenal glucose load were studied in chronically catheterized conscious dogs. [3-3H]glucose was infused throughout the study. After basal measurements, glucose spiked with [U-14C]glucose was infused for 150 min intraduodenally. Arterial insulin and glucagon were similar in the two groups. Arterial glucose (mg/dl) rose approximately 70% more during glucose infusion after 42 h than after an 18-h fast. The net hepatic glucose balance (mg. kg-1. min-1) was similar in the two groups (basal: 1.8 +/- 0.2 and 2.0 +/- 0.3; glucose infusion: -2.2 +/- 0.5 and -2.2 +/- 0.7). The intrahepatic fate of glucose was 79% glycogen, 13% oxidized, and 8% lactate release after a 42-h fast; it was 23% glycogen, 21% oxidized, and 56% lactate release after an 18-h fast. Net hindlimb glucose uptake was similar between groups. The appearance of intraduodenal glucose during glucose infusion (mg/kg) was 900 +/- 50 and 1,120 +/- 40 after 18- and 42-h fasts (P < 0.01). CONCLUSION: glucose administration after prolonged fasting induces higher circulating glucose than a shorter fast (increased appearance of intraduodenal glucose); liver and hindlimb glucose uptakes and the hormonal response, however, are unchanged; finally, an intrahepatic redistribution of carbons favors glycogen deposition.

Enhanced muscle glucose uptake facilitates nitrogen efflux from exercised muscle.

The hypothesis that glucose ingestion in the postexercise state enhances the synthesis of glutamine and alanine in the skeletal muscle was tested. Glucose was infused intraduodenally for 150 min (44.5 micromol . kg-1 . min-1) beginning 30 min after a 150-min period of exercise (n = 7) or an equivalent duration sedentary period (n = 10) in 18-h-fasted dogs. Prior exercise caused a twofold greater increase in limb glucose uptake during the intraduodenal glucose infusion compared with uptake in sedentary dogs. Arterial glutamine levels fell gradually with the glucose load in both groups. Net hindlimb glutamine efflux increased in response to intraduodenal glucose in exercised but not sedentary dogs (P < 0. 05-0.01). Arterial alanine levels, depleted by 50% with exercise, rose with intraduodenal glucose in exercised but not sedentary dogs (P < 0.05-0.01). Net hindlimb alanine efflux also rose in exercised dogs in response to intraduodenal glucose (P < 0.05-0.01), whereas it was not different from baseline in sedentary controls for the first 90 min of glucose infusion. Beyond this point, it, too, rose significantly. We conclude that oral glucose may facilitate recovery of muscle from prolonged exercise by enhancing the removal of nitrogen in the form of glutamine and alanine.

A brain of her own: a neural correlate of song assessment in a female songbird.

The song control region in the avian forebrain is a series of discrete, interconnected nuclei mediating song learning and production. It has been studied in males or in species where both sexes sing. Little is known about the neural correlates of song perception in nonsinging females, often the intended recipients of song. We studied cowbirds (Molothrus ater), a species in which only males sing but in which females discriminate between males on the basis of song. We focused on nucleus lMAN because it has been implicated in early song acquisition, a stage relevant to both sexes to choose among competing acoustic models. We found that volume of lMAN was monomorphic in cowbirds. Moreover, the volume and neuronal number of female lMAN were positively correlated with selectivity of copulatory responding. The results provide strong evidence of nonsinging female's use of "song" control nuclei for song perception without the possibility of song production.

Effect of prior exercise on the partitioning of an intestinal glucose load between splanchnic bed and skeletal muscle.

Exercise leads to marked increases in muscle insulin sensitivity and glucose effectiveness. Oral glucose tolerance immediately after exercise is generally not improved. The hypothesis tested by these experiments is that after exercise the increased muscle glucose uptake during an intestinal glucose load is counterbalanced by an increase in the efficiency with which glucose enters the circulation and that this occurs due to an increase in intestinal glucose absorption or decrease in hepatic glucose disposal. For this purpose, sampling (artery and portal, hepatic, and femoral veins) and infusion (vena cava, duodenum) catheters and Doppler flow probes (portal vein, hepatic artery, external iliac artery) were implanted 17 d before study. Overnightfasted dogs were studied after 150 min of moderate treadmill exercise or an equal duration rest period. Glucose ([14C]glucose labeled) was infused in the duodenum at 8 mg/kg x min for 150 min beginning 30 min after exercise or rest periods. Values, depending on the specific variable, are the mean +/- SE for six to eight dogs. Measurements are from the last 60 min of the intraduodenal glucose infusion. In response to intraduodenal glucose, arterial plasma glucose rose more in exercised (103 +/- 4 to 154 +/- 6 mg/dl) compared with rested (104 +/- 2 to 139 +/- 3 mg/dl) dogs. The greater increase in glucose occurred even though net limb glucose uptake was elevated after exercise (35 +/- 5 vs. 20 +/- 2 mg/min) as net splanchnic glucose output (5.1 +/- 0.8 vs. 2.1 +/- 0.6 mg/kg x min) and systemic appearance of intraduodenal glucose (8.1 +/- 0.6 vs. 6.3 +/- 0.7 mg/kg x min) were also increased due to a higher net gut glucose output (6.1 +/- 0.7 vs. 3.6 +/- 0.9 mg/kg x min). Adaptations at the muscle led to increased net glycogen deposition after exercise [1.4 +/- 0.3 vs. 0.5 +/- 0.1 mg/(gram of tissue x 150 min)], while no such increase in glycogen storage was seen in liver [3.9 +/- 1.0 vs. 4.1 +/- 1.1 mg/(gram of tissue x 150 min) in exercised and sedentary animals, respectively]. These experiments show that the increase in the ability of previously working muscle to store glycogen is not solely a result of changes at the muscle itself, but is also a result of changes in the splanchnic bed that increase the efficiency with which oral glucose is made available in the systemic circulation.

Structure/functional properties of the yeast dual regulator protein NGG1 that are required for glucose repression.

NGG1p/ADA3p is a yeast dual function regulator required for the complete glucose repression of GAL4p-activated genes (Brandl, C. J., Furlanetto, A. M., Martens, J. A., and Hamilton, K. S. (1993) EMBO J. 12, 5255-5265). Evidence for a direct role for NGG1p in regulating activator function is supported by the finding that NGG1p is also required for transcriptional activation by GAL4p-VPl6 and LexA-GCN4p (Pina, B., Berger, S. L., Marcus, G. A., Silverman, N., Agapite, J., and Guarente, L. (1993) Mol. Cell. Biol. 13, 5981-5989). By analyzing deletion derivatives of the 702-amino acid protein, we identified a region essential for glucose repression within residues 274-373. Essential sequences were further localized to a segment rich in Phe residues that is predicted to be an amphipathic alpha helix. As well as finding mutations within this region that reduced glucose repression, we identified mutations that made NGG1p a better repressor. In addition, NGG1p probably represses GAL4p activity as part of a complex containing ADA2p because single and double disruptions of ngg1 and ada2 had comparable effects on glucose repression. We also localized a transcriptional activation domain within the amino-terminal amino acids of NGG1p that is proximal or overlapping the region required for glucose repression. Activation by GAL4p-NGG1p(1-373) requires ADA2p; however, activation by GAL4p-NGG1p(1-308), is ADA2p-independent. This suggests that a site required for ADA2p interaction lies between amino acids 308 and 373 and that ADA2p has a regulatory role in activation by GAL4p-NGG1p(1-373).

Behaviour of complex oligosaccharides at a bilayer membrane surface: probed by 2H-NMR.

Deuterium wideline NMR was used in an attempt to directly assess oligosaccharide arrangement and motional characteristics of complex glycosphingolipids dispersed as minor components in phospholipid membranes. A convenient, general synthetic approach was developed which involved replacement of the acetate group of amido sugars with deuteroacetate (-COCD3). This provided excellent signal-to-noise when applied to the terminal GalNAc residue of globoside, and the terminal NANA residue of GM1. Simultaneously, globoside and GM1 fatty acids were replaced with stearic acid deuterated at C-2- a probe location sensitive to glycolipid hydrophobic backbone orientation and rigid body motion. Deuterated GM1 and globoside were studied by 2H-NMR in bilayers of 1-palmitoyl-2-oleoyl phosphatidylcholine, in the presence and absence of physiological quantities of cholesterol. The monoglycosyl glycosphingolipid, glucosyl ceramide, which is the common skeleton of many complex glycosphingolipids including those studied here, was also deuterated at fatty acid C-2 for comparative study in the same matrices. Correlation with spectra of the complex glycolipids demonstrated that, for a given temperature and membrane composition, ceramide backbone conformation was very similar amongst the species studied. Spectral features of GM1 deuterated on terminal NANA and assembled at a membrane surface, were found to be highly consistent with the oligosaccharide conformation determined in studies of GM1 in solution. In contrast, globoside deuterated in the terminal GalNAc residue gave spectra very different from those predicted on the basis of the conformation considered to exist in solution. It seems likely that this result reflects a combination of greater oligosaccharide chain flexibility relative to GM1, and the presence of the membrane environment. Interestingly, although there was highly significant spatial geometry associated with the complex oligosaccharide chains, and although temperature and the presence of cholesterol exert measurable effects on the membrane-inserted portion, these factors had very little impact on the measured spectral parameters associated with the NANA residue of GM1 or the terminal GalNAc residue of globoside. This seems to indicate lack of sensitivity of the complex oligosaccharide chains to conformation and internal motions of the hydrophobic chain segments in these fluid and semi-fluid membranes; and has important implications for mechanisms of crypticity.

Acyl chain length effects related to glycosphingolipid crypticity in phospholipid membranes: probed by 2H-NMR.

Wideline 2H-NMR was used to consider the relationships amongst glycosphingolipid and phospholipid fatty acid chain length and glycosphingolipid receptor function, in a system classically associated with crypticity. Galactosyl ceramide (GalCer), having 18- or 24-carbon fatty acid, was deuterium labelled at the conformationally-restricted fatty acid alpha-carbon (C-2). 2H-NMR spectra of N-[2,2-2H2]stearoyl and N-[2,2-2H2]lignoceroyl GalCer (GalCer with 18-vs. 24-carbon selectively deuterated fatty acid) were then compared over a range of temperatures in phosphatidylcholine/cholesterol membranes in which the host phospholipid had dimyristoyl, dipalmitoyl, or distearoyl fatty acid composition. Findings were evaluated in the light of known sensitivity of antibody interaction with GalCer to temperature and to both glycolipid fatty acid chain length and host matrix fatty acid chain length. Under the conditions of experimentation, spectra were not obtainable for glycolipids having rigid body motions that were slow on the NMR timescale (10(-4)-10(-5) s)-i.e.. motions typical of non-fluid (gel phase) membranes. The systems, DPPC/cholesterol and DSPC/cholesterol, in which the original observation was made of increased antibody binding to GalCer with long fatty acid, proved to be characterised by receptor motions that were in this slow timescale for both 18:0 and 24:0 GalCer at 22-24 degrees C. Under conditions for which spectra could be obtained, those for GalCer with [2,2-2H2]lignoceroyl (24-carbon alpha-deuterated) fatty acid were qualitatively similar to those of its 18-carbon analogue in all (fluid) membranes examined. However, spectral splittings differed quantitatively between deuterated 18:0 and 24:0 GalCer at a given temperature, dependent upon host matrix. These differences were most marked at lower temperatures and in the longer chain (more ordered) matrices, DPPC/cholesterol and DSPC/cholesterol. This suggests that maximum effects of glycolipid chain length on glycolipid receptor function may be expected to occur in spatially and motionally constrained lipid environments. There was little effect of temperature on spectral splittings seen for a given sample containing deuterated 18:0 GalCer. The small differences seen could be adequately accounted for by relatively minor alterations in glycolipid order and backbone conformation. In contrast, 24:0 GalCer in DPPC/cholesterol and DSPC/cholesterol displayed significant variation in its spectral splittings as the temperature was reduced; and these proved to be the source of the quantitative differences between 18:0 and 24:0 GalCer referred to above.(ABSTRACT TRUNCATED AT 400 WORDS)

Characterization of NGG1, a novel yeast gene required for glucose repression of GAL4p-regulated transcription.

The GAL1-10 genes of Saccharomyces cerevisiae are regulated by the interaction of cis- and trans-acting factors which facilitate activated transcription in galactose but not in glucose medium. By selecting mutations that allow expression of a defective gal1-10-his3 hybrid promoter, we have identified a novel gene, NGG1, which is required for glucose repression of the GAL10-related his3-G25 promoter. ngg1 was identified as a recessive null mutation that in the presence of a gal80 background resulted in a 300-fold relief of glucose repression for the his3-G25 promoter. This compared with a 20-fold and negligible relief of repression in gal80 and ngg1 strains, respectively. Deletion analysis of the his3-G25 promoter showed a correlation between the number of GAL4p binding sites and the relative level of NGG1p activity. Relief of glucose repression by NGG1 was dependent on the presence of GAL4, but was independent of the GAL4 promoter. In addition, NGG1p activity was seen for a promoter construct containing independent GAL4p binding sites. These results suggest that NGG1p acts to inhibit GAL4p function in glucose medium. We have cloned NGG1 by complementation and found that it contains an open reading frame of 2106 bp which could encode a protein with a molecular weight of 79,230.

Glycosphingolipid backbone conformation and behavior in cholesterol-containing phospholipid bilayers.

2H NMR spectroscopy was used to consider correspondence between existing single-crystal X-ray data for glycosphingolipids and their ceramide backbone conformation in fluid phospholipid membranes. A monoglycosylated sphingolipid, glucosylceramide (GlcCer), which represents the core structure of many important glycosphingolipids, was derived by partial synthesis through replacement of all native fatty acids with the 18-carbon species, stearic acid, deuterated at C2. N-[2,2-2H2]stearoyl-GlcCer was used to probe glycosphingolipid orientation and motion at low concentration in "fluid" phospholipid bilayers composed of dimyristoylphosphatidylcholine (DMPC), with and without physiological amounts of cholesterol. Spectral analysis, aided by stereoselective monodeuteration of the GlcCer fatty acid at C2, demonstrated that glycosphingolipid average acyl chain backbone conformation in fluid phospholipid membranes, with or without cholesterol, is likely closely related to that predicted from single crystal X-ray studies [Pascher, I. (1976) Biochim. Biophys. Acta 455, 433-451; Pascher, I., & Sundell, S. (1977) Chem. Phys. Lipids 20, 175-191]. To test the generality of this observation, specific comparisons were made involving galactosylceramide (GalCer) and globoside. GalCer provided a glycolipid differing only in monosaccharide stereochemistry (galactose vs glucose). Globoside permitted isolation of the effect of headgroup size, since it is derived from GlcCer via extension of the carbohydrate portion by the oligosaccharide, GalNAc beta 1-->3Gal alpha 1-->4Gal attached in beta 1-->4 linkage to the Glc residue.(ABSTRACT TRUNCATED AT 250 WORDS)

Phase behaviour of amphotericin B multilamellar vesicles.

Because side effect profiles and key physical properties of liposomal amphotericin B reflect the molecular nature of the hydrated preparations, effort has been directed toward understanding this nature. We describe here an examination by differential scanning calorimetry in the region of the main transition of the phase behaviour of amphotericin B multilamellar liposomes used investigationally for patient treatment. Liposomes were composed of 7:3 dimyristoylphosphatidylcholine/dimyristoylphosphatidylglycerol (7:3 DMPC/DMPG) containing up to 33 mol% drug. Preparations in which pure DMPC or pure 1-oleoyl-2-stearoylphosphatidylcholine (OSPC) was substituted for 7:3 DMPC/DMPG were subjected to the same measurements for comparison. The DSC-derived partial phase diagrams were similar to those previously recorded using EPR spectroscopy for unsonicated liposomes of 7:3 DMPC/DMPG containing amphotericin B, and for mixtures with different pure saturated and unsaturated phosphatidylcholines (Grant, C.W.M., et al. (1989) Biochim. Biophys. Acta 984, 11-20). Fluidization onset temperatures for liposome host matrices were relatively unaffected by drug compared to the temperatures of completion. This effect was particularly marked for the unsaturated phospholipid matrix. Partial phase diagrams were interpreted as demonstrating that amphotericin B has a tendency to separate into a rigid phase within the membrane. This is consistent with molecular modelling considerations which suggest that amphotericin B may exist as oligomers in a phospholipid matrix. Drug-induced alterations of DSC melting profiles for the phospholipid bilayers studied were less extensive than those reported for partially sonicated preparations of 7:3 DMPC/DMPG (Janoff, A.S., et al. (1989) Proc. Natl. Acad. Sci. USA 85, 6122-6126). Melting profiles obtained did not change upon further sample incubation, suggesting that the hydrated preparation represented a thermodynamically stable form.

Physical biochemistry of a liposomal amphotericin B mixture used for patient treatment.

There seems little doubt now that intravenous liposomal amphotericin B can be a useful treatment modality for the management of immunocompromised patients with suspected or proven disseminated fungal infections. Interestingly, the very significant reduction in toxicity reported when amphotericin B is part of a bilayer membrane is closely tied to the physical characteristics of the liposomes involved, although these are poorly understood at the molecular level. We record here an examination by spectroscopy and freeze-etch electron microscopy of unsonicated amphotericin B multilamellar vesicles prepared along the lines that we and others have followed for samples used in clinical trials and preclinical in vivo or in vitro studies. Our study has focussed on liposomes of 7:3 dimyristoylphosphatidylcholine/dimyristoylphosphatidylglycerol (DMPC/DMPG) bearing 0-25 mol% amphotericin B, since this lipid mixture has been the choice for the first clinical trials. Phase transition behaviour of these liposomes was examined by electron paramagnetic resonance (EPR) spectroscopy of a nitroxide spin label partitioning into the bilayers. The same experiments were then performed on similarly prepared liposomes of the disaturated species, dipalmitoylphosphatidylcholine (DPPC), and the diunsaturated species, dielaidoylphosphatidylcholine (DEPC). Partial phase diagrams were constructed for each of the lipid/drug mixtures. Melting curves and derived phase diagrams showed evidence that amphotericin B is relatively immiscible with the solid phase of bilayer membranes. The phase diagram for DEPC/amphotericin B was very similar to that of DPPC/amphotericin B, and both exhibited less extensive temperature ranges of phase separation than did the 7:3 DMPC/DMPG mixture with amphotericin B. Between 25 and 37 degrees C the measured fluidity of the 7:3 DMPC/DMPG liposomes was similar to that of the (unsaturated fatty acid) DEPC liposomes, and considerably higher than that seen for (saturated fatty acid) DPPC liposomes. Preparations of 7:3 DMPC/DMPG, DPPC, and DEPC containing 0-25 mol% amphotericin B were examined by freeze-etch electron microscopy at 35 and 22 degrees C (to cover the temperature range of the mammalian body core and periphery). The same liposome features were present in all three liposome types studied. The appearance of individual liposomes at x 100,000 magnification reflected their molecular characteristics, which were found to be significantly heterogeneous within each batch. The lipid/drug structures were bilayer in nature, although liposomes showing considerable disruption were common, particularly at the highest drug concentrations.(ABSTRACT TRUNCATED AT 400 WORDS)