Alloalbuminemia in Sweden: structural study and phenotypic distribution of nine albumin variants

Mouse Type-I Interferon-Mannosylated Albumin Fusion Protein for the Treatment of Chronic Hepatitis

Although a lot of effort has been put into creating drugs and combination therapies against chronic hepatitis, no effective treatment has been established. Type-I interferon is a promising therapeutic for chronic hepatitis due to its excellent anti-inflammatory effects through interferon receptors on hepatic macrophages. To develop a type-I IFN equipped with the ability to target hepatic macrophages through the macrophage mannose receptor, the present study designed a mouse type-I interferon-mannosylated albumin fusion protein using site-specific mutagenesis and albumin fusion technology. This fusion protein exhibited the induction of anti-inflammatory molecules, such as IL-10, IL-1Ra, and PD-1, in RAW264.7 cells, or hepatoprotective effects on carbon tetrachloride-induced chronic hepatitis mice. As expected, such biological and hepatoprotective actions were significantly superior to those of human fusion proteins. Furthermore, the repeated administration of mouse fusion protein to carbon tetrachloride-induced chronic hepatitis mice clearly suppressed the area of liver fibrosis and hepatic hydroxyproline contents, not only with a reduction in the levels of inflammatory cytokine (TNF-α) and fibrosis-related genes (TGF-β, Fibronectin, Snail, and Collagen 1α2), but also with a shift in the hepatic macrophage phenotype from inflammatory to anti-inflammatory. Therefore, type-I interferon-mannosylated albumin fusion protein has the potential as a new therapeutic agent for chronic hepatitis.

Human serum albumin variants in China: a molecular epidemiological investigation and literature review

Background

Bisalbuminemia is a hereditary and/or acquired abnormality characterized by a double albumin (ALB) band on serum protein electrophoresis. However, there have been no epidemiological investigations of ALB variants in Chinese populations.

Methods

This retrospective study examined 71,963 unrelated subjects from five provinces in southern China. ALB variants were screened by cellulose acetate electrophoresis at pH 8.6 and ALB mutations were confirmed by polymerase chain reaction-DNA sequencing.

Results

The average incidence of inherited bisalbuminemia in the southern Chinese population was 0.0264% (19/71,963). Thirteen cases showed slow and six showed fast genetic variants on cellulose acetate electrophoresis. Four kinds of ALB variants were identified: proalbumin Lille (p.Arg23His), ALB Castel di Sangro (p.Lys560Glu), ALB Fukuoka-1 (p.Asp587Asn), and a novel ALB Wuxi (p.Lys562Glu). The gene frequency of ALB variants in the Wuxi region (0.126%, 13/10,297) was significantly higher than in other regions in southern China, and 90.9% (10/11) of cases of proalbumin Lille were also found in the Wuxi region.

Conclusions

This study provides the first report of the detailed prevalence and molecular characterization of ALB variants in southern China. Compared with other areas of China, Wuxi had a different pattern of ALB variants and a high prevalence of proalbumin Lille.

Immune Dysfunction and Albumin-Related Immunity in Liver Cirrhosis

Liver cirrhosis yearly causes 1.2 million deaths worldwide, ranking as the 10th leading cause of death in the most developed countries. High susceptibility to infections along with a significant risk for infection-related mortality justifies the description of liver cirrhosis as the world's most common immunodeficiency syndrome. Liver cirrhosis is an end-stage organic disease hallmarked by a multifaceted immune dysfunction due to deterioration of antimicrobial recognition and elimination mechanisms in macrophages along with an impaired antigen presentation ability in circulating monocytes. Bacterial translocation supports—and is supported by—uncontrolled activation of immune cell responses and/or loss of toll-like receptor (TLR) tolerance, which can turn exaggerated inflammatory responses to systemic inflammation. Lipopolysaccharide (LPS) or endotoxin boosts systemic inflammatory activity through activation of TLR-2- and TLR-4-dependent pathways and facilitate a massive production of cytokines. This, in turn, results into elevated secretion of reactive oxygen species (ROS), which further enhances intestinal hyperpermeability and thus sustains a vicious circle of events widely known as “leaky gut.” Albumin can be of particular benefit in cirrhotic patients with spontaneous bacterial peritonitis and/or hepatorenal syndrome type of acute kidney injury (HRS-AKI) due to anti-inflammatory and antioxidative stress as well as volume-expanding properties and endothelial-stabilizing attributes. However, presence of autoantibodies against albumin in patients with liver cirrhosis has been described. Although previous research suggested that these antibodies should be regarded as naturally occurring antibodies (NOA), the origin of the antialbumin immune response is obscure. High occurrence of NAO/albumin complexes in patients with liver disease might reflect a limited clearance capacity due to bypassing portal circulation. Moreover, high burden of oxidized albumin is associated with less favorable outcome in patients with liver cirrhosis. To date, there is no data available as to whether oxidized forms of albumin result in neoepitopes recognized by the immune system. Nevertheless, it is reasonable to hypothesize that these alterations may have the potential to induce antialbumin immune responses and thus favor systemic inflammation.

Determination of true ratios of different N-glycan structures in electrospray ionization mass spectrometry

An ideal method for the analysis of N-glycans would both identify the isomeric structure and deliver a true picture of the relative, if not absolute, amounts of the various structures in one sample. Porous graphitic carbon chromatography coupled with electrospray ionization mass spectrometry (ESI-MS) detection has emerged as a method with a particularly high potential of resolving isomeric oligosaccharides, but little attention has so far been paid to quantitation of the results obtained. In this work, we isolated a range of structures from Man5 to complex type N-glycans with zero to four sialic acids and blended them into an equimolar "glyco tune mix". When subjected to liquid chromatography-ESI-MS in positive and negative modes, the glyco tune mix clearly demonstrated the futility of quantitation of N-glycans of different overall composition, different number of sialic acids, and strongly differing size without compensation for their very different molar responses. Relative quantitation of human plasma N-glycans was performed with correction factors deduced from this external glyco tune mix. Addition of just one isotope-coded internal standard with enzymatically added 13C-galactose led to absolute quantification in the same experiment. Graphical Abstract Discrepancy between desirable (grey bars) and real (green bars) relative ion abundance of equimolar amounts of glycans in positive mode ESI-MS.

Human serum albumin isoforms: genetic and molecular aspects and functional consequences

BACKGROUND

At present, 67 different genetic variants of human serum albumin and proalbumin have been molecularly characterized at the protein and/or gene level.

SCOPE OF REVIEW

This review summarizes present knowledge about genetic and molecular aspects, functional consequences and potential uses of the variants.

MAJOR CONCLUSIONS

The frequency of bisalbuminemia in the general population is probably about 1:1000, but it can be much higher in isolated populations. Mutations are often due to hypermutable CpG dinucleotides, and in addition to single-amino acid substitutions, glycosylated variants and C-terminally modified alloalbumins have been found. Some mutants show altered stability in vivo and/or in vitro. High-affinity binding of Ni(++) and Cu(++) is blocked, or almost so, by amino acid changes at the N-terminus. In contrast, substitution of Leu90 and Arg242 leads to strong binding of triiodothyronine and l-thyroxine, respectively, resulting in two clinically important syndromes. Variants often have modified plasma half-lives and organ uptakes when studied in mice.

GENERAL SIGNIFICANCE

Because alloalbumins do not seem to be associated with disease, they can be used as markers of migration and provide a model for study of neutral molecular evolution. They can also give valuable molecular information about albumins binding sites, antioxidant and enzymatic properties, as well as stability. Mutants with increased affinity for endogenous or exogenous ligands could be therapeutically relevant as antidotes, both for in vivo and extracorporeal treatment. Variants with modified biodistribution could be used for drug targeting. In most cases, the desired function can be further elaborated by producing site-directed, recombinant mutants. This article is part of a Special Issue entitled Serum Albumin.

Human serum albumin: from bench to bedside

Human serum albumin (HSA), the most abundant protein in plasma, is a monomeric multi-domain macromolecule, representing the main determinant of plasma oncotic pressure and the main modulator of fluid distribution between body compartments. HSA displays an extraordinary ligand binding capacity, providing a depot and carrier for many endogenous and exogenous compounds. Indeed, HSA represents the main carrier for fatty acids, affects pharmacokinetics of many drugs, provides the metabolic modification of some ligands, renders potential toxins harmless, accounts for most of the anti-oxidant capacity of human plasma, and displays (pseudo-)enzymatic properties. HSA is a valuable biomarker of many diseases, including cancer, rheumatoid arthritis, ischemia, post-menopausal obesity, severe acute graft-versus-host disease, and diseases that need monitoring of the glycemic control. Moreover, HSA is widely used clinically to treat several diseases, including hypovolemia, shock, burns, surgical blood loss, trauma, hemorrhage, cardiopulmonary bypass, acute respiratory distress syndrome, hemodialysis, acute liver failure, chronic liver disease, nutrition support, resuscitation, and hypoalbuminemia. Recently, biotechnological applications of HSA, including implantable biomaterials, surgical adhesives and sealants, biochromatography, ligand trapping, and fusion proteins, have been reported. Here, genetic, biochemical, biomedical, and biotechnological aspects of HSA are reviewed.

Mutations and polymorphisms of the gene of the major human blood protein, serum albumin

We have tabulated the 77 currently known mutations of the familiar human blood protein, serum albumin (ALB). A total of 65 mutations result in bisalbuminemia. Physiological and structural effects of these mutations are included where observed. Most of the changes are benign. The majority of them were detected upon clinical electrophoretic studies, as a result of a point mutation of a charged amino acid residue. Three were discovered by their strong binding of thyroxine or triiodothyronine. A total of 12 of the tabulated mutations result in analbuminemia, defined as a serum albumin concentration of <1 g/L. These were generally detected upon finding a low albumin concentration in patients with mild edema, and involve either splicing errors negating translation or premature stop codons producing truncated albumin molecules. A total of nine mutations, five of those with analbuminemia and four resulting in variants modified near the C-terminal end, cause frameshifts. Allotypes from three of the point mutations become N-glycosylated and one C-terminal frameshift mutation shows O-glycosylation.

Analbuminemia in a Swedish male is caused by the Kayseri mutation (c228_229delAT)

BACKGROUND

Analbuminemia is a rare autosomal recessive disorder manifested by the absence, or severe reduction, of circulating serum albumin. Here we report the first case of hereditary analbuminemia in the ethnic Swedish population, and we define the molecular defect that causes the analbuminemic trait.

METHODS

Total DNA, extracted from peripheral blood samples from the analbuminemic proband and his parents, was PCR-amplified using oligonucleotide primers designed to amplify the 14 exons, the exon-intron splice junctions, and the 5' and 3' untranslated regions of the albumin gene. The products were screened for mutations by single-strand conformation polymorphism and heteroduplex analyses. The latter allowed the identification of the abnormal fragment, which was then sequenced.

RESULTS

The analbuminemic trait of the proband was caused by a homozygous AT deletion at nucleotides c. 228-229, the 91st and 92nd bases of exon 3. This defect, previously identified as Kayseri mutation [M. Galliano, M. Campagnoli, A. Rossi, et al. Molecular diagnosis of analbuminemia: a novel mutation identified in two Amerindian and two Turkish families. Clin Chem 2002;48: 844-849.], produces a frameshift leading to a premature stop, two codons downstream.

CONCLUSIONS

The Kayseri mutation appears to be the most common cause of analbuminemia in humans, and is found in individuals belonging to geographically distant, and apparently unrelated ethnic groups.

Detection of hereditary bisalbuminemia in a Greek family by capillary zone electrophoresis

It is presented herein a case of a family, four members of which suffer from hereditary bisalbuminemia. The abnormality was initially detected in a 29-year old male, by serum protein electrophoresis (SPE), during the investigation for possible multiple sclerosis. SPE also revealed the presence of a double albumin band in sera of the patient's sister, father and grandmother, almost confirming the inherited (genetic) form of bisalbuminemia. Possible causes related with the acquired form of bisalbuminemia were excluded for all examined individuals. SPE was performed by both automatic capillary zone electrophoresis and agaroze gel electrophoresis. All tested samples were immunofixated with special antisera, in order to exclude the presence of monoclonal fractions. Total albumin, total proteins and immunoglobulins varied in normal ranges. The relative mobility of the albumin variant was determined by a simple mixing experiment, which gave evidence of the fast-type form of inherited bisalbuminemia. This is the first report of hereditary bisalbuminemia in Greece.

Genetic variants showing apparent hot-spots in the human serum albumin gene

The molecular defects of three different slow-migrating genetic variants of human serum albumin, albumins Kamloops (formerly RIH), Stirling and Amsterdam, previously characterized only by electrophoretic and dye-binding studies, are now reported. Two of them are proalbumin variants: sequential analysis of the purified whole proteins has established the mutation responsible for albumin Kamloops as -1Arg-->Gln, and for albumin Stirling as -2Arg-->His. A Glu-->Lys substitution in position 570 of the mature albumin molecule was determined in albumin Amsterdam by sequential analysis of two abnormal tryptic fragments. The three alloalbumins are caused by single-base changes all of which seem to represent hot-spots in the albumin gene. The possible functional consequences of the presence of a circulating alloalbumin are discussed.

Structural characterization, stability and fatty acid-binding properties of two French genetic variants of human serum albumin

Four bisalbuminemic, unrelated persons were found in Bretagne, France, and their variant and normal albumins were isolated by DEAE ion exchange chromatography, reduced, carboxymethylated and treated with CNBr. Comparative two-dimensional electrophoresis of the CNBr digests showed that three of the variants were modified in fragment CB4, whereas the fourth had an abnormal fragment CB1. These fragments were isolated, digested with trypsin and mapped by reverse-phase HPLC. Sequencing of altered tryptic peptides showed that the three variants modified in CB4 were caused by the same, previously unreported, amino acid substitution: Asp314-->Val (albumin Brest). The fourth, however, was a proalbumin variant with the change Arg-2-->Cys (albumin Ildut). Both amino acid substitutions can be explained by point mutations in the structural gene: GAT-->GTT (albumin Brest) and CGT-->TGT (albumin Ildut). The proalbumin Ildut is very unstable and already in vivo it is to a large extent cleaved posttranslationally to Arg-Albumin and normal albumin. Furthermore, we observed that during a lengthy isolation procedure the remaining proalbumin was changed to Arg-Albumin or proalbumin lacking Arg-6. In addition, part of normal albumin had lost Asp1. Gas chromatographic investigations using isolated proteins indicated that albumin Brest has improved in vivo fatty acid-binding properties, whereas the structural modification(s) of albumin Ildut does not affect fatty acid binding.

Effect of genetic variation on the fatty acid-binding properties of human serum albumin and proalbumin

In the circulation, non-esterified fatty acids are transported by albumin which also facilitates their removal from donor cells and uptake into receptor cells. We have studied whether genetic variations in the albumin molecule can affect its in vivo fatty acid-binding properties. The fatty acids bound to 25 structurally different variants and to their wildtype counterparts, isolated from heterozygous carriers, were determined gas chromatographically. The variants were proalbumins, albumins with single amino acid substitutions and glycosylated or truncated albumins. In eight cases the total amount bound to the variants was diminished (0.4-0.8-fold), and in seven cases the load was increased to 1.3 or more of normal. Twenty-one fatty acids were quantitated, and for 19 alloalbumins significant deviations from normal were found. Usually, changes in total and individual fatty acid binding were of the same type, but several exceptions to this rule was found. The glycosylated albumin Casebrook showed the largest changes, the total load and the amount of bound palmitate was 8.6 and 14 times, respectively, the normal. The most pronounced changes and the majority of cases of increased binding were caused by molecular changes in domain III. Mutations in domain I, II and the propeptide resulted in smaller effects, if any, and these were often reductions in binding.

Structural characterization of four genetic variants of human serum albumin associated with alloalbuminemia in Italy

A long-term electrophoretic survey on plasma proteins, which was carried out in several clinical laboratories in Italy, identified 28 different genetic variants of human serum albumin and four cases of analbuminemia. We have previously characterized 16 point mutations, 3 C-terminal mutants, and the genetic defects in two analbuminemic subjects. Here, we report the molecular defects of four alloalbumins that have been characterized by protein structural analysis. Of these, three represent new single-point mutations: albumins Tregasio, Val122-->Glu, Bergamo, Asp314-->Gly, and Maddaloni, Val533-->Met. The fourth, albumin Besana Brianza, has the same Asp494-->Asn mutation that introduces a glycosylation site which has been previously reported in a variant from New Zealand, albumin Casebrook. However, in contrast to albumin Casebrook, albumin Besana Brianza is only partially glycosylated and the oligosaccharide is heterogeneous, consisting of a biantennary complex type N-glycan with either two or one sialic acid residue(s) on the antennae. Both albumin Maddaloni and Besana Brianza represent mutations at hypermutable CpG dinucleotide sites; albumin Maddaloni is a mutant that does not involve a charged amino acid.

A genetic variant of albumin (albumin Asola; Tyr140-->Cys) with no free -SH group but with an additional disulfide bridge

A slow migrating variant of human serum albumin, present in lower amount than the normal protein, has been detected by routine clinical electrophoresis at pH 8.6 in two members of a family living in Asola (Lombardia, Italy). Ion-exchange chromatography of serum samples failed to separate the normal protein from the variant. Analysis of the albumin peak by SDS/PAGE revealed that the variant had a lower apparent molecular mass than its normal counterpart. However, the abnormal band was not detectable when the separation was performed under reducing conditions or when both albumins were carboxymethylated. Isoelectric-focusing analysis of CNBr fragments localized the mutation to fragment CNBr 3 (residues 124-298). This fragment was isolated on a preparative scale and subjected to tryptic digestion. Sequence determination of the abnormal tryptic peptide revealed that the variant arises from a Tyr140--> Cys substitution. This result was confirmed by DNA sequence analysis, which showed a single transition of TAT-->TGT at nucleotide position 5074. Despite the presence of an additional cysteine residue, several lines of evidence indicated that albumin Asola has no free -SH group; therefore, we propose the formation of a new S-S bond between Cys140 and Cys34, the only free sulphydryl group present in the normal protein. The relatively low level of the variant in serum and its abnormal mobility on cellulose acetate electrophoresis and SDS/PAGE are probably caused by a gross conformational change of the molecule induced by the new S-S bridge.

Analbuminemia: three cases resulting from different point mutations in the albumin gene

Analbuminemia is a very rare recessive disorder in which subjects have little or no circulating albumin, although albumin is normally the most abundant plasma protein and has many functions. Analbuminemia is caused by a variety of mutations in the albumin gene and is exhibited only by subjects homozygous for the defect. Previously the mutation had been identified at the molecular level in only two human cases; in one case it resulted from an exon-splicing defect, and in the other case it was caused by a nucleotide insertion that caused a frameshift and premature stop codon. In this investigation we identified the mutations in three unrelated subjects from different countries. In each instance a single-nucleotide mutation produced a stop codon, but the mutations occurred at three different sites: (i) in an Italian male a C-->T transition at nt 2368 in the genomic sequence of albumin, (ii) a C-->T transition at nt 4446 for an American female, and (iii) a G-->A transition at nt 7708 in a Canadian male. The size of the albumin fragment that might have been produced for the three cases varied from 31- to 213-amino acid residues, but no evidence for a circulating albumin fragment was obtained. The paradox is that analbuminemia is extremely rare (frequency < 1 x 10(6)); yet the virtual absence of albumin is tolerable despite its multiple functions.

Genetic variants of human serum albumin in Italy: point mutants and a carboxyl-terminal variant

Of the > 50 different genetic variants of human serum albumin (alloalbumins) that have been characterized by amino acid or DNA sequence analysis, almost half have been identified in Italy through a long-term electrophoretic survey of serum. Previously we have reported structural studies of 11 Italian alloalbumins with point mutations, 2 different carboxyl-terminal variants, and 1 case of analbuminemia in an Italian family. This article describes confirmation by DNA sequencing of mutations previously inferred from protein sequencing of 4 of the above alloalbumins; it also reports the mutations identified by protein and DNA sequence analysis of 4 other Italian alloalbumins not previously recorded: albumin Larino, His3-->Tyr; Tradate-2 (protein sequencing only), Lys225-->Gln; Caserta, Lys276-->Asn; and Bazzano, a carboxyl-terminal variant. The first 3 have point mutations that produce a single amino acid substitution, but a nucleotide deletion causes a frameshift and an altered and truncated carboxyl-terminal sequence in albumin Bazzano. In these 4 instances the expression of the alloalbumin is variable, ranging from 10% to 70% of the total albumiN, in contrast to the usual 50% each for the normal and mutant albumin. The distribution of point mutations in the albumin gene is nonrandom; most of the 47 reported point substitutions involve charged amino acid residues on the surface of the molecule that are not concerned with ligand-binding sites.

A nucleotide insertion and frameshift cause analbuminemia in an Italian family

In analbuminemia, a very rare inherited syndrome, subjects produce little or no albumin (1/100th to 1/1000th normal), presumably because of a mutation in the albumin gene; yet, they have only moderate edema and few related symptoms owing to a compensatory increase in other plasma proteins. Because of the virtual absence of albumin the defect must be identified at the DNA level. In this study the mutation causing analbuminemia in an Italian family was investigated by analysis of DNA from a mother and her daughter. The mother was homozygous for the trait and had a serum albumin value of < 0.01 g/dl (about 1/500th normal); the daughter was heterozygous for the trait and had a nearly normal albumin value. Molecular cloning and sequence analysis of DNA from both mother and daughter showed that the mutation is caused by a nucleotide insertion in exon 8; this produces a frameshift leading to a premature stop, seven codons downstream. The methods of heteroduplex hybridization and single-strand conformation polymorphism were used to compare the DNA of the mother and daughter to the DNA of two unrelated analbuminemic individuals (one Italian and one American). This showed that all three analbuminemic individuals had different mutations; these also differed from the mutation in the only human case previously studied at the DNA level, which was a splicing defect affecting the ligation of the exon 6-exon 7 sequences. Thus, analbuminemia may result from a variety of mutations and is genetically heterogeneous.

The structural characterization and bilirubin-binding properties of albumin Herborn, a [Lys240-->Glu] albumin mutant

We report the molecular defect of albumin Herborn, a new genetic variant of human serum albumin which has been found in Germany. Isoelectric focusing analysis of CNBr fragments from the purified variant allowed us to localize the mutation in fragment CNBr 3 (residues 124-298). This fragment was isolated on a preparative scale and subjected to tryptic and V8 protease digestion. Sequence determination of the abnormal tryptic and V8 peptides revealed that the variant arises from the substitution Lys240-->Glu. The -2 charge change of albumin Herborn, which is probably due to a A-->G transition in the first position of the corresponding codon in the structural gene, has no significant effect on its electrophoretic mobility under non-denaturating conditions. Therefore we have assumed that residue 240, which has been implicated in the bilirubin primary binding site (Jacobsen, C. (1978) Biochem. J. 171, 453-459), is buried. The binding of bilirubin and biliverdin by albumin Herborn was quantified using the fluorescence quenching method. The apparent equilibrium association constants (Ka +/- SD) and the number of high-affinity binding sites (n) of the defatted variant for bilirubin and biliverdin were Ka = 1.03 +/- 0.18 x 10(8) M-1, n = 1.07; and Ka = 7.48 +/- 1.10 x 10(6) M-1, n = 1.01, respectively. The Ka values are about 93.3% and 99.1% of the values found for the normal protein under the same conditions. These results strongly suggest that Lys240 of human serum albumin is not the basic residue involved in ion pairing with one of the carboxylate groups of bilirubin at its high-affinity site.

cDNA and protein sequence of polymorphic macaque albumins that differ in bilirubin binding

The rhesus monkey, Macaca mulatta, exhibits a geographically restricted polymorphism of serum albumins Mac A and Mac B that is recognized by electrophoresis and is associated with a difference in bilirubin-binding parameters. To identify the basis of the polymorphism, the cDNA and protein sequences of serum albumin from M. mulatta were determined. Screening of a lambda gt11 rhesus liver cDNA library yielded a 1988-bp cDNA sequence that encodes the complete amino acid sequence of mature albumin, the entire propeptide, and part of the prepropeptide. Isoelectric focusing and amino-terminal protein sequencing of CNBr fragments of albumin from A/A and B/B homozygotes were performed, and the structural difference was localized to a CNBr fragment (MCB3) spanning residues 124-264. Sequence analysis of lysyl endopeptidase peptides of MCB3 established that Mac A albumin has a glutamine residue at position 188 while the Mac B albumin has a glutamic residue at the same position. PCR amplification, subcloning, and DNA sequence analysis of clones from A/A and B/B homozygotes confirmed the protein sequence data and the codon difference of CAA versus GAA, respectively. Comparison of macaque and human serum albumin shows a 93.5% identity at the amino acid level. In human serum albumin, Glu188 is located close to the IIA binding pocket for ligands, probably including bilirubin. Derivatives of coumarin compete more efficiently with bilirubin for binding sites on the Mac A albumin than on the Mac B albumin. In regions where coumarin-containing plants are important food resources, Mac B albumin may confer a selective advantage because bilirubin is less readily displaced from it.