Wavelength-dependence of the relative rate constants for the main geometric and structural photoisomerization of bilirubin IX alpha bound to human serum albumin. Demonstration of green light at 510 nm as the most effective wavelength in photochemical changes from (ZZ)-bilirubin IX alpha to (EZ)-cyclobilirubin IX alpha via (EZ)-bilirubin.

The kinetics for the quantitatively important reaction: (Formula: see text) that is, the photochemical interconversion between bilirubin and its geometric and structural photoisomers bound to human serum albumin in aqueous solution when various wavelengths of monochromatic light were used, were assayed by h.p.l.c. In order to clarify the wavelength-dependence of the relative rate constants in the individual steps, a light-source with a half-bandwidth of 10 nm was used at increments of 20 nm, in the range from 410 nm to 550 nm. We describe for the first time studies on the wavelength-dependence of rate constants in geometric and structural photoisomerization reactions in vitro of (ZZ)-bilirubin or (EZ)-bilirubin bound to human serum albumin, especially the relative rate constants of cyclization of (EZ)-bilirubin into (EZ)-cyclobilirubin. Because studies in vitro have demonstrated that the wavelengths from 350 to 450 nm are mutagenic, the results obtained indicated that the safest and ideal light-source for phototherapy is green light of 510 nm, which keeps (ZE)-bilirubin concentrations as low as possible, as shown by a maximal value of k2 at 510 nm and a relatively low value of k1 at 510 nm. This light-source still ensures the substantial absorption of (ZZ)-bilirubin, which is the precursor of (EZ)-bilirubin, the intermediate in (EZ)-cyclobilirubin formation and, furthermore, as shown by the maximal value of k5 and a considerable value of k4 at 510 nm, promotes the cyclization of (EZ)-bilirubin derived from (ZZ)-bilirubin even though k3 at 510 nm also shows a peak value.     

Kinetic study of the photochemical changes of (ZZ)-bilirubin IX alpha bound to human serum albumin. Demonstration of (EZ)-bilirubin IX alpha as an intermediate in photochemical changes from (ZZ)-bilirubin IX alpha to (EZ)-cyclobilirubin IX alpha.

The present study was performed to elucidate why the photochemical reaction of (ZZ)-bilirubin bound to human serum albumin is singularly selective, and only one of the two (EZ)- and (ZE)-bilirubins, the (ZE)-isomer, is produced. In a kinetic study of the photochemical reaction in vitro, the sum of the relative rate constants of photochemical transformation of (EZ)-bilirubin into both (EZ)-cyclobilirubin and (ZZ)-bilirubin, with a significant preference for the former, was proved to be considerably larger than that of the transformation of (ZZ)-bilirubin into (EZ)-bilirubin. Therefore only one of the geometrical isomers, namely (ZE)-bilirubin, is apparently formed. It was concluded that (EZ)-bilirubin photochemically undergoes (EZ)-cyclization, i.e. structural photoisomerization, while bound to its high-affinity site on human serum albumin, and is an intermediate in the transformation of (ZZ)-bilirubin into (EZ)-cyclobilirubin.     

Biliary and urinary excretion rates and serum concentration changes of four bilirubin photoproducts in Gunn rats during total darkness and low or high illumination.

On cycled exposure of Gunn rats to total darkness and low and high illumination, biliary excretion rates of (EZ)- and (ZE)-bilirubin and (EZ)-cyclobilirubin increased up to approx. 10-fold from the mean basal values of 1.2 and 0.2 microgram/h to the mean maximum values of 25.2 and 4.2 micrograms/h respectively, and at the same time those of (EE)-bilirubin and (EE)-cyclobilirubin also increased, but at very much lower rates than those of the first-mentioned two. During the low illumination only (EZ)- and (ZE)-bilirubin and (EZ)-cyclobilirubin appeared in the urine; during the high illumination (EE)-bilirubin and (EE)-cyclobilirubin also appeared, showing a similar excretion pattern to that observed in the bile, but the total urinary excretion rates were lower than the total biliary excretion rates. The serum bilirubin concentrations fell gradually to lower values, accompanied by an increment in (EZ)- and (ZE)-bilirubin, but (EZ)-cyclobilirubin was not detected. It is concluded that during phototherapy the predominant pathway for the removal of bilirubin from the body in the Gunn rat is by biliary excretion of the geometric photoisomers (EZ)- and (ZE)-bilirubin, derived from Z----E isomerization, and the structural photoisomer (EZ)-cyclobilirubin, formed from intramolecular endo-vinyl cyclization.     

Effect of the binding of bilirubin to either the first class or the second class of binding sites of the human serum albumin molecule on its photochemical reaction.

The kinetics of the photochemical changes of bilirubin were studied at a constant concentration of bilirubin bound either to the first class or to the second class of binding sites of the human serum albumin molecule. The more the bilirubin binds to the first class of binding sites in the human serum albumin molecule, the more readily geometric photoequilibrium to give (ZE)-bilirubin takes place. The more the bilirubin binds to the second class of binding sites or allosterically transformed binding sites induced by added SDS, the more readily structural photoisomerization, i.e. the formation of (EZ)-cyclobilirubin, takes place. When the serum bilirubin concentration is at low, safe, values bilirubin binds exclusively to the first class of binding sites and serves as an antioxidant [Onishi, Yamakawa & Ogawa (1971) Perinatology 1, 373-379]; at these concentrations human serum albumin protects bilirubin from irreversible photodegradation by only allowing readily reversible geometric photoisomerization. As the serum bilirubin concentration increases to high, and potentially dangerous, values, bilirubin binds to the second class of binding sites, and under these conditions human serum albumin seems to promote the photocyclization of bilirubin. During irradiation human serum albumin seems to act by retaining low, useful, concentrations of bilirubin while facilitating irreversible photoisomerization of excess bilirubin.     

Metabolism of bilirubin and its photoisomers in newborn infants during phototherapy

Bilirubin and its photoisomers in the biological fluids of a hyperbilirubinaemic newborn infant before and during phototherapy were analyzed by a recently improved HPLC method. In the serum, the percentages of (EZ)- and (ZE)-bilirubin in the total bilirubin concentration before phototherapy were approximately 10% and on average increased over 1.5-fold at 2 h after initiation of phototherapy. The percentage of the (EZ)-cyclobilirubin in the serum bilirubin was under 1%. In the bile, the mean concentration of (ZZ)-bilirubin, derived mainly from (ZE)-bilirubin, nearly tripled during phototherapy. The (EZ)-cyclobilirubin concentration in the bile was very low before phototherapy, increased nearly ten-fold at 3 h after initiation of phototherapy, and was 5- to 6-fold as high as that of (ZZ)-bilirubin. In the urine, upon exposure to light, the urinary concentration of (EZ)-cyclobilirubin is apparently equivalent to half of the biliary concentration of (ZZ)-bilirubin and one-fifth of that of (EZ)-cyclobilirubin. It was concluded that during phototherapy of neonatal hyperbilirubinaemia the structural photoisomer [(EZ)-cyclobilirubin] predominates considerably over the geometric photoisomer [(ZE)-bilirubin].     

The effect of bilirubin photoisomers on unbound-bilirubin concentrations estimated by the peroxidase method.

Unbound bilirubin is oxidized to nearly colourless substances in the presence of H2O2 or ethyl hydroperoxide and horseradish peroxidase. To predict the risk of kernicterus (degenerated yellow pigmentation of nerve cells), this principle has been widely utilized for estimating the concentration of unbound bilirubin in hyperbilirubinaemic serum. However, the serum contains polar geometric photoisomers of bilirubin. Therefore, to clarify the effect of bilirubin photoisomer concentrations on unbound-bilirubin concentration, the concentration of bilirubin and its photoisomer and of unbound bilirubin in samples obtained from experiments in vivo and in vitro were simultaneously and individually estimated by h.p.l.c. and the peroxidase method. During photoirradiation, both in vivo and in vitro, the serum polar (ZE)-bilirubin IX alpha concentration increased remarkably, but unbound-bilirubin values were not affected at all. However, during experiments in vitro, unbound bilirubin concentrations increased only when concentrations of the rather polar (EZ)- and (EE)-cyclobilirubin IX alpha increased considerably in a human serum albumin-bilirubin solution irradiated with blue light. Thus it is concluded that unbound-bilirubin concentrations, and consequently the initial rate of the peroxidase reaction, is not accelerated by the increase in either (ZE)-bilirubin or (EZ)-cyclobilirubin concentration within the clinically observed range.     

Multiple binding of bilirubin to human serum albumin and cobinding with laurate

Numerical analysis of multiple binding of two ligands to one carrier has been accomplished, using the principle of several sets of acceptable binding constants, with bilirubin-laurate-albumin as an example. Binding of bilirubin to defatted human serum albumin was investigated by a spectroscopic method, based upon a difference of light absorption spectrum for free and bound bilirubin. The observations were supplemented with previous data from an independent technique, measurement of oxidation rates of free bilirubin with hydrogen peroxide and peroxidase. A continuous isotherm was obtained, showing binding of at least 4 mol bilirubin per mole albumin with the following stoichiometric binding constants, 1.11 X 10(8), 1.7 X 10(7), 8 X 10(5), and 4 X 10(4) M-1 at pH 8.2, ionic strength 0.15 M, 25 degrees C. The binding is anticooperative at all steps. A saturation level was not reached. Cobinding of bilirubin and laurate was studied, with up to 2 mol of each ligand per mole albumin, using the peroxidase method for determination of free equilibrium concentrations of bilirubin, and a dialysis rate technique for free laurate. The findings could be described in terms of a stoichiometric model. Heterotropic cooperativity was present among the first bilirubin and the first and second laurate molecules. More than two molecules of either ligand can be bound at the same time.     

A dynamic model for bilirubin binding to human serum albumin

Site-directed mutagenesis of human serum albumin was used to study the role of various amino acid residues in bilirubin binding. A comparison of thermodynamic, proteolytic, and x-ray crystallographic data from previous studies allowed a small number of amino acid residues in subdomain 2A to be selected as targets for substitution. The following recombinant human serum albumin species were synthesized in the yeast species Pichia pastoris: K195M, K199M, F211V, W214L, R218M, R222M, H242V, R257M, and wild type human serum albumin. The affinity of bilirubin was measured by two independent methods and found to be similar for all human serum albumin species. Examination of the absorption and circular dichroism spectra of bilirubin bound to its high affinity site revealed dramatic differences between the conformations of bilirubin bound to the above human serum albumin species. The absorption and circular dichroism spectra of bilirubin bound to the above human serum albumin species in aqueous solutions saturated with chloroform were also examined. The effect of certain amino acid substitutions on the conformation of bound bilirubin was altered by the addition of chloroform. In total, the present study suggests a dynamic, unusually flexible high affinity binding site for bilirubin on human serum albumin.     

The binding of bilirubin to albumin. A study using spin-labelled bilirubin

Binding between human serum albumin and a spin-labelled derivative of bilirubin was investigated by circular dichroism, fluorescence quenching, electron spin resonance and visible spectroscopy. The orders of magnitude of the binding constants obtained by flurorescence quenching and electron spin resonance spectroscopies were 10(7) and 10(3) 1 . mol-1, respectively. These data suggest that most spin-labelled bilirubin interacts with human serum albumin at the side not holding the spin-labelled side-arm. CD measurements showed the presence of at least two sites, associated with opposite Cotton effects. It is worthy of note that the Cotton sign of the first site is inverted with respect to the corresponding one of bilirubin. CD measurements on mixed systems (spin-labelled bilirubin/human serum albumin/bilirubin) were also performed. The decomposition of the ternary curves shows that the rotatory power of bilirubin bound to human serum albumin is higher in the ternary system than in the binary (bilirubin/human serum albumin). The corresponding CD measurements for the binding between spin-labelled bilirubin and bovine serum albumin are also reported and discussed.     

The kinetics of interactions of bilirubin with lipid bilayers and with serum albumin.

Rate constants for the hydration of bilirubin bound to unilamellar bilayers of dioleoylphosphatidylcholine and albumin were measured by stopped-flow methods. Rate constants for association of bilirubin with these vesicles and albumin were calculated from measured rate constants for dissociation and the equilibrium binding constants of bilirubin and lipids or albumin. Rate constants for hydration (dissociation) for bilirubin bound to dioleoylphosphatidylcholine and albumin were 71 s-1 and 1.8 s-1 respectively. Rate constants for association were 4.0 10(7) s-1 and 1.1 10(9) M-1 s-1, respectively. Both rates for interactions of bilirubin with bilayers were essentially independent of temperature in the range 0-40 degrees C, indicating that barriers to entry and exit of bilirubin from bilayers were entropic. Rates of transbilayer movement of bilirubin in dioleoylphosphatidylcholine were too fast to resolve by measuring rates of hydration of bilirubin. Rate constants for hydration of bilirubin bound to bilayers with less avidity for bilirubin as compared with dioleoylphosphatidylcholine also were too fast to measure with stopped-flow methods. In addition to providing details of the energetic basis for interactions between bilirubin and membranes, the data allow for calculating the maximal rates at which bilirubin could transfer spontaneously from sites on albumin in blood to the interior of cells. The data show, in this regard, that this rate is 10-50 fold faster than measured rates of uptake of bilirubin by intact liver.     

The binding of bilirubin to albumin a study using spin-labelled bilirubin

Binding between human serum albumin and a spin-labelled derivative of bilirubin was investigated by circular dichroism, fluorescence quenching, electron spin resonance and visible spectroscopy. The orders of magnitude of the binding constants obtained by flurorescence quenching and electron spin resonance spectroscopies were 10⁷ and 10³ I·-mol⁻¹, respectively. These data suggest that most spin-labelled bilirubin interacts with human serum albumin at the side not holding the spin-labelled side-arm. CD measurements showed the presence of at least two sites, associated with opposite Cotton effects. It is worthy of note that the Cotton sign of the first site is inverted with respect to the corresponding one of bilirubin. CD measurements on mixed systems (spin-labelled bilirubin/human serum albumin/bilirubin) were also performed. The decomposition of the ternary curves shows that the rotatory power of bilirubin bound to human serum albumin is higher in the ternary system than in the binary (bilirubin/human serum albumin). The corresponding CD measurements for the binding between spin-labelled bilirubin and bovine serum albumin are also reported and discussed.     

Species-dependent stereoselective drug binding to albumin: a circular dichroism study

Drug binding to albumins from different mammalian species was investigated to disclose evidence of species-dependent stereoselectivity in drug-binding processes and affinities. This aspect is important for evaluating the reliability of extrapolating distribution data among species. The circular dichroism (CD) signal induced by drug binding to the albumins [human serum albumin (HSA), bovine serum albumin (BSA), rat serum albumin (RSA), and dog serum albumin (DSA)] were measured and analyzed. The binding of selected drugs and metabolites to HSA significantly differed from the binding to the other albumins in terms of affinity and conformation of the bound ligands. In particular, phenylbutazone, a marker of site one on HSA, showed a higher affinity for binding to BSA with respect to RSA, HSA, and DSA, respectively. In the case of diazepam, a marker of site two on HSA, the affinity decreased in order from HSA to DSA, RSA, and BSA. The induced CD spectra were similar in terms of energy and band signs, suggesting almost the same conformation for the bound drug to the different albumins. Stereoselectivity was high for the binding of ketoprofen to HSA and RSA. A different sign was observed for the CD spectra induced by the drug to the two albumins because of the prevalence of a different conformation of the bound drug. Interestingly, the same induced CD spectra were obtained using either the racemic form or the (S)-enantiomer. Finally, significant differences were observed in the affinity of bilirubin, being highest for BSA, then decreasing for RSA, HSA, and DSA. A more complex conformational equilibrium was observed for bound bilirubin.     

Crystallographic analysis of human serum albumin complexed with 4Z,15E-bilirubin-IXalpha

Bilirubin, an insoluble yellow-orange pigment derived from heme catabolism, accumulates to toxic levels in individuals with impaired or immature liver function. The resulting jaundice may be managed with phototherapy to isomerize the biosynthetic 4Z,15Z-bilirubin-IXα to more soluble and excretable isomers, such as 4Z,15E-bilirubin. Bilirubin and its configurational isomers are transported to the liver by human serum albumin (HSA) but their precise binding location(s) on the protein have yet to be determined. To investigate the molecular details of their interaction, we co-crystallised bilirubin with HSA. Strikingly, the crystal structure—determined to 2.42 Å resolution—revealed the 4Z,15E-bilirubin-IXα isomer bound to an L-shaped pocket in sub-domain IB. We also determined the co-crystal structure of HSA complexed with fusidic acid, an antibiotic that competitively displaces bilirubin from the protein, and showed that it binds to the same pocket. These results provide the first crystal structure of a natural bilirubin pigment bound to serum albumin, challenge some of the present conceptions about HSA-bilirubin interactions, and provide a sound structural framework for finally resolving the long-standing question of where 4Z,15Z-bilirubin-IXα binds to the protein.     

The effect of ligand presaturation on the interaction of serum albumins with an immobilized Cibacron Blue 3G-A studied by affinity gel electrophoresis.

The interaction of the immobilized triazine dye Cibacron Blue 3G-A with rat, rabbit, sheep, goat, bovine and human serum albumins was studied by affinity gel electrophoresis. Dissociation constants were estimated in each instance and showed human serum albumin to have a significantly higher affinity for the dye than did albumin from any other species. Pretreatment of the defatted proteins with bilirubin (3 mol of bilirubin/mol of protein) did not increase the dissociation constants of the serum albumins, whereas pretreatment with palmitate (7 mol of palmitate/mol of protein) increased the dissociation constant in all cases: 3-fold for human serum albumin, 15-fold for other serum albumins. Increasing the bilirubin/albumin ratio (to 7:1) did not affect the dissociation constant of the albumins studied. Decreasing the palmitate/albumin ratio decreased the dissociation constant for human serum albumin, but did not affect those of bovine and rat albumins. Altering the chain length of the presaturating fatty acid dramatically changed the dissociation constant of both human and bovine serum albumins. Butyrate, hexanoate, octanoate and decanoate did not significantly influence the dissociation constants of bovine and human serum albumins for Cibacron Blue, whereas laurate, myristate and palmitate greatly increased the dissociation constant. These data are discussed in relationship to the behaviour of albumins during dye--agarose column chromatography. In Addendum the effect of nucleotide presaturation on the interaction between Bacillus stearothermophilus 6-phosphogluconate dehydrogenase and the immobilized triazine dyes Cibacron Blue 3G-A and Procion Red HE-3B was examined, and the implications for dye--ligand chromatography are discussed.     

Kinetics and mechanism of bilirubin binding to human serum albumin

The kinetics of bilirubin binding to human serum albumin at pH 7.40, 4 degrees C, was studied by monitoring changes in bilirubin absorbance. The time course of the absorbance change at 380 nm was complex: at least three kinetic events were detected including the bimolecular association (k1 = 3.8 +/- 2.0 X 10(7) M-1 S-1) and two relaxation steps (52 = 40.2 +/- 9.4 s-1 and k3 = 3.8 +/- 0.5 s-1). The presence of the two slow relaxations was confirmed under pseudo-first order conditions with excess albumin. Curve-fitting procedures allowed the assignment of absorption coefficients to the intermediate species. When the bilirubin-albumin binding kinetics was observed at 420 nm, only the two relaxations were seen; apparently the second order association step was isosbestic at this wavelength. The rate of albumin-bound bilirubin dissociation was measured by mixing the pre-equilibrated human albumin-bilirubin complex with bovine albumin. The rate constant for bilirubin dissociation measured at 485 nm was k-3 = 0.01 s-1 at 4 degrees C. A minimum value of the equilibrium constant for bilirubin binding to human albumin determined from the ratio k1/k-3 is therefore approximately 4 X 10(9) M-1.     

A novel binding site for bile acids on human serum albumin

Binding sites of bile acids on human serum albumin were studied using various probes: dansylsarcosine (site I probe), 7-anilinocoumarin-4-acetic acid (ACAA, site II probe), 5-dimethylaminonaphthelene-1-sulfonamide (DNSA, site III probe), cis-parinaric acid (probe for fatty acid binding site) and bilirubin. Bile acids competitively inhibited the binding of dansylsarcosine to human serum album whereas bile acids enhanced the binding of ACAA, DNSA, cis-parinaric acid and bilirubin. Considering the concentrations of bile acids required to inhibit the binding of dansylsarcosine to human serum albumin, the secondary binding site of bile acids may correspond to site I. Dissociation constants (K_d) of the primary binding sites of lithocholic and chenodeoxycholic acid to human serum albumin were approximately 0.2 and 4 μM, respectively, which was measured by equilibrium dialysis at 37° C. All the bile acids and their sulfates and glucuronides inhibited the binding of chenodeoxycholic acid to human serum albumin. Lithocholic and chenodeoxycholic acid and their sulfates and glucuronides exhibited more inhibition than cholic acid and its conjugates. In conclusion, bile acids may bind to a novel binding site on human serum albumin.     

Spectroscopic properties of bilirubin-human serum albumin complexes: a stoichiometric analysis

Light absorption spectra, fluorescence of bound bilirubin, fluorescence of albumin as quenched by bilirubin, and circular dichroism spectra have been studied in mixtures of bilirubin and defatted human serum albumin in variable proportions at 25 degrees C and at pH 7.4, 8.2, and 9.0. Corresponding spectral data have been calculated for the stoichiometric bilirubin-albumin complexes, 1:1, 2:1, and 3:1. Light absorption spectra as well as the bound bilirubin fluorescence indicate that all three bound bilirubin dianions are internalized. These data were obtained by curve fitting to least sum of squared deviations. In addition to the best fit we obtained 30 acceptable curves, located within an F contour, thus producing a rough estimate of the variation of the resulting spectral data.     

Ionization of tyrosine residues in human serum albumin and in its complexes with bilirubin and laurate.

Spectrophotometric titration of human serum albumin indicates that ionization of the 18 tyrosine residues takes place between pH 9 and 12.7. A Hill plot indicates that protons dissociate co-operatively from tyrosine residues, in pure albumin between pH 11.0 and 11.4 with a Hill coefficient 1.7, and in the bilirubin-albumin complex between pH 11.2 and 11.7 with a Hill coefficient 1.6. With a stopped-flow technique it is shown that about seven of the tyrosines ionize fast, with rate constants well above 10(2) s-1, when pH is suddenly changed from near neutral to pH 11.76. Further residues ionize slowly, with rate constants around 10(2) s-1 or less. The N-form of albumin (pH 6) contains one more fast ionizing tyrosine than the B-form of albumin (pH 10). Binding of bilirubin or laurate to the albumin molecule (molar ratio 1:1) transforms one to three of the fast ionizing tyrosines to slowly ionizing.     

Human serum albumin. Spectroscopic studies of binding and proximity relationships for fatty acids and bilirubin.

Binding and proximity relationships of hydrophobic ligands on human serum albumin have been studied using absorption, fluorescence, circular dichroism, and electron paramagnetic resonance spectroscopy. The ligands studied were bilirubin, two conjugated linear polyene fatty acids, cis-parinaric acid and cis-eleostearic acid, and three nitroxide derivatives of stearic acid with doxyl groups at positions 5, 10, and 12, respectively. Binding of polyene fatty acids was monitored by absorption peak shifts, induced circular dichroism, enhancement of fluorescence, and energy transfer between albumin's single tryptophanyl residue and the polyene chromophore. Induced circular dichroism studies indicate excitonic ligand-ligand interaction between bound fatty acids. Fluorescence enhancement of cis-parinaric acid was analyzed using a stepwise multiple equilibrium model, and six binding constants in the range 10(8) to 10(6) M-1 were obtained, in agreement with previous measurements for other fatty acids. The temperature dependence of the equilibrium constants indicates that the binding enthalpy is nearly zero. Fluorescence energy transfer was similarly used to quantitate bilirubin binding to albumin. Energy transfer, nitroxide quenching of fluorescence, and electron paramagnetic resonance spectroscopy were used to elucidate binding geometries which support and extend proposed structural models for albumin. It is suggested that the first two fatty acids bind side-by-side in an antiparallel fashion in domain III of human serum albumin.     

Existence of a bioactive lipid, cyclic phosphatidic acid, bound to human serum albumin

A novel bioactive lipid, cyclic phosphatidic acid (cPA), was identified in lipids bound to human serum albumin. A cPA fraction was extracted and purified from human serum albumin by use of a combination of preparative TLC and HPLC. Electrospray ionization mass spectrometry of the purified fraction showed molecular ions corresponding to cPA, which was composed of some different fatty acid species. The most abundant component was identified as palmitoyl-cPA by tandem mass spectrometry using collision-induced dissociation. These data have established that cPA is a naturally occurring lipid bound to human serum albumin.