Reduced circulating oxidized LDL is associated with hypocholesterolemia and enhanced thiol status in Gilbert syndrome.

A protective association between bilirubin and atherosclerosis/ischemic heart disease clearly exists in vivo. However, the relationship between bilirubin and in vivo oxidative stress parameters in a clinical population remains poorly described. The aim of this study was to assess whether persons expressing Gilbert syndrome (GS; i.e., unconjugated hyperbilirubinemia) are protected from thiol oxidation and to determine if this, in addition to their improved lipoprotein profile, could explain reduced oxidized low-density lipoprotein (oxLDL) status in them. Forty-four matched GS and control subjects were recruited and blood was prepared for the analysis of lipid profile and multiple plasma antioxidants and measures of oxidative stress. GS subjects possessed elevated plasma reduced thiol (8.03±1.09 versus 6.75±1.39 nmol/mg protein; P<0.01) and glutathione concentrations (12.7±2.39 versus 9.44±2.45 μM; P<0.001). Oxidative stress status (reduced:oxidized glutathione; GSH:GSSG) was significantly improved in GS (0.49±0.16 versus 0.32±0.12; P<0.001). Protein carbonyl concentrations were negatively associated with bilirubin concentrations and were significantly lower in persons with >40 μM bilirubin versus controls (<17.1 μmol/L; P<0.05). Furthermore, absolute oxLDL concentrations were significantly lower in GS subjects (P<0.05). Forward stepwise regression analysis revealed that bilirubin was associated with increased GSH:GSSG ratio and reduced thiol concentrations, which, in addition to reduced circulating LDL, probably decreased oxLDL concentrations within the cohort. In addition, a marked reduction in total cholesterol concentrations in hyperbilirubinemic Gunn rats is presented (Gunn 0.57±0.09 versus control 1.69±0.40 mmol/L; P<0.001), arguing for a novel role for bilirubin in modulating lipid status in vivo. These findings implicate the physiological importance of bilirubin in protecting from atherosclerosis by reducing thiol and subsequent lipoprotein oxidation, in addition to reducing circulating LDL concentrations.

Introduction

Individuals with elevated circulating bilirubin concentrations (including Gilbert syndrome; GS) are potently protected from cardiovascular disease (CVD) [1,2] and cancer [3]. Persons with GS have a circulating total bilirubin concentration of >17.1 μM, due to additional TA repeats in the gene promoter for bilirubin uridine diphosphate glucuronosyltransferase (UGT) 1A1, decreasing expression [4]. Decreased UGT1A1 activity reduces bilirubin conjugation and increases the circulating unconjugated bilirubin concentration.

The mechanisms of in vivo protection in GS remain poorly described and are limited to investigations studying inhibition of lipid oxidation [5]. Reduced urinary excretion of bilirubin oxidation products [6] and circulating advanced glycation end products [7] in individuals with GS also suggest that bilirubin protects from macromolecule modification in vivo. Whether elevated unconjugated bilirubin concentrations are associated with reduced markers of lipid and protein oxidation, however, remains unknown. Evidence to support the possibility of reduced protein and lipid oxidation and inflammation in GS can be found in in vitro and animal studies [8,9]. Translation of these findings within clinical cohorts, such as GS, provides a unique opportunity to reveal the physiological importance of bilirubin, because circulating protein and lipid oxidation markers are related to future CVD mortality, which is reduced in GS [10].

Bilirubin, the pigment of jaundice, also causes neurotoxicity in vitro, in newborns [11] and in adults with the rare condition of Crigler–Najjar syndrome [12]. However, elevated circulating bilirubin concentrations have been associated with protection from neonatal hyperoxic exposure (and retinopathy of prematurity) in both rodent [13] and human [14] studies. Despite these findings, evidence to the contrary also exists and supports a popular debate in this area [15,16]. Together, these findings show that bilirubin is a molecule of diverse and currently unknown importance requiring further exploration [17]. This study investigated whether GS subjects possess improved reduced thiol and protein oxidation biomarkers versus controls and explored the possible involvement of bilirubin in modulating oxidative stress and lipid profile parameters using stepwise regression analysis.

Methods

Subject recruitment

Subjects (18–63 years) were recruited from the general population via Internet forums, e-mail advertisements, and posters. Subject exclusion criteria included current or recent (2 weeks) bacterial or viral infection, smoking, antioxidant supplementation, prescribed medication (other than the contraceptive pill), excessive alcohol consumption (>8 standard drinks per week), elevated glucose or serum liver enzyme activities, or presence of a lipid metabolism disorder. The presence of a lipid metabolism disorder was defined using criteria developed by American [18,19] and European [20] working parties (i.e., lipid metabolism disorder is defined as cholesterol >7.8 mmol/L and triglycerides >2.3 mmol/L or HDL <0.9 mmol/L). Individuals were also excluded if they had a family history of cardiovasculardisease/hypertension, diabetes, or hematologic or liver disease. Blood pressure was measured using a commercial sphygmomanometer (Homedics BPA-110-AU, Australia) and body mass index (BMI) was calculated using standard digital scales and height assessment. Subjects were allocated to the GS group if they possessed a previous diagnosis from a medical practitioner and/or a circulating unconjugated bilirubin concentration in excess of 17.1 μM (>1 mg/dl) and normal serum liver enzyme activities. Circulating plasma heme concentrations were quantified to exclude individuals with overt hemolysis. In total 87 subjects were recruited and of these 44 (22 GS, 22 controls) were matched for gender, age, and BMI. There were equal numbers of female and male subjects in each group (11:11). The ethnicity of the recruited cohort included Caucasian (80%) and Indo/Asian (20%) individuals with similar distributions of ethnicity in each group. The study was approved by the Human Ethics Research Committee of Griffith University (MSC/02/10/HREC).

Blood sample collection

Fasting blood samples (30 ml total) were collected into serum, lithium heparin, and ethylenediaminetetraacetic acid vacutainers. Samples were kept in the dark, on ice, and were processed within 1 h.

Animal experiments

Breeding pairs of heterozygote (genotyped) Gunn rats were imported from the Rat Research and Resource Center (Columbia, MO, USA) and kept within an animal house facility at Griffith University (12-h light:dark cycle, constant temperature (22 °C) and humidity (60%)). Rats had continuous access to standard laboratory food pellets (Speciality Feeds, Glen Forrest, Australia) and fresh water. Homozygous Gunn rat offspring were assumed to possess jaundice at birth, were ear-tagged, and were housed together with female littermate (nonjaundiced) controls after weaning (all female, n=7 and 5, respectively). Concentrations of unconjugated bilirubin were measured in blood obtained from the tail tip, during brief isoflurane anesthesia (3% in 100 O2; 1–2 L/min), at 21 day of age to confirm the presence of jaundice. Serum unconjugated bilirubin concentrations were analyzed using high-performance liquid chromatography (HPLC) (see Bilirubin and heme analysis). Nonjaundiced controls were assumed to represent either heterozygous or wild-type Wistar rats that both are UGT1A1 competent and possess unconjugated bilirubin concentrations approximating 1 μM. At 13 months of age (similar relative age, in human years, to the clinical cohort) the animals were anesthetized using intraperitoneal injection with pentobarbital sodium (concentration 60 mg/ml; 100 μl per 100 g). A midline laparotomy was performed, including excision of the heart (for other purposes), and approximately 5 ml of whole blood was collected from the chest cavity using a syringe and prepared for analysis of various biochemical parameters (see Sample preparation). All analytical procedures (see Serum biochemistry) were identical to those used for clinical samples. All procedures were approved by the Griffith University Animal Ethics Research Committee before experimentation (MSC/04/09).

Sample preparation

Plasma/serum was centrifuged (Thermo Scientific 5810R, Australia) at 1500g for 15 min (4 °C). Aliquots were prepared immediately and stored at −80 °C. Trichloroacetic acid (TCA; 10%, 1:1 ratio) was added to heparinized plasma aliquots, which were then vortexed and centrifuged at 2200g for 5 min. Supernatant was frozen at −80 °C for ascorbate and reduced/oxidized glutathione analysis.

Bilirubin and heme analysis

The serum unconjugated bilirubin and heme concentrations were quantified using HPLC and a photodiode array detector (Waters, Australia) as previously described [5]. Slight variation to this method included the use of a C18 reverse-phase HPLC guard and analytical column (4.6×150 mm, 3 μm; Phenomenex, Australia) that was perfused at 1 ml min−1. Extracted samples (100 μl) were injected with a run time of 14 min in triplicate. Unconjugated bilirubin (450 nm) and heme (400 nm) eluted at 12 and 8.5 min, respectively. Unconjugated bilirubin (0–100 μM) and hemin (0–5 μM) served as external standards (Frontier Scientific, Logan, UT, USA).

Serum biochemistry

Serum biochemistry including liver enzymes (alanine aminotransferase, aspartate aminotransferase, γ-glutamyltransferase), glucose, uric acid (Thermo Scientific), antioxidant status including Trolox equivalent antioxidant capacity (TEAC) [21], and ferric reducing ability of plasma (FRAP) [22] was assessed using a Cobas Mira automated analyzer (Roche, Switzerland). Lipid status including assessment of total cholesterol, triglycerides, and high-density lipoprotein were analyzed using commercially available kits on a COBAS Integra 400 blood chemistry analyzer (Roche Diagnostics, Australia). Cholesterol analyses were conducted using appropriate lipid standards (Calibrator for Automated Systems Lipids) and quality controls (Preci Control Clin Chem Multi; Roche Diagnostics). The concentration of low-density lipoprotein (LDL) cholesterol was calculated using the Friedewald equation [23]. All analyses were conducted in duplicate.

Ascorbate analysis

Ascorbate concentrations were measured spectrophotometrically using a 96-well plate reader (Multiskan FC, Thermo Scientific) according to the method of Wei et al. [24] Supernatant from TCA-precipitated samples (100 μl) was used to determine ascorbate concentrations. All samples were analyzed in triplicate and ascorbic acid (from 0 to 20 mg/100 ml 5% TCA) was used as an external standard.

Quantification of reduced thiol/protein concentration

Reduced thiol concentrations were quantified in heparinized plasma using a 96-well plate reader and 5,5-dithiobis(2-nitrobenzoic acid) (DTNB) reagent. The concentration of TNB was quantified at 412 nm according to the method of Hawkins et al. [25] (in triplicate). GSH (0–0.5 mM) served as an external standard. Thiol concentrations were expressed initially in micromoles per liter and then converted to nmol/mg of protein. Protein concentrations were determined according to the bicinchoninic acid protein assay kit (Thermo Scientific).

Determination of protein carbonyls

Detection of protein carbonyls with 2,4-dinitrophenylhydrazine was modified from the original method of Levine et al. [26], performed using an ELISA kit (Sapphire Biosciences, Australia) and expressed in nmol/mg of protein.

Measurement of glutathione

Oxidized (GSSG) and reduced (GSH) glutathione concentrations were measured using N-ethylmaleimide and o-phthalaldehyde (in triplicate) via a modified method of Hissin and Hilf [27] using a Fluoroskan Ascent 96-well plate reader (Thermo Scientific). The concentrations of GSH and GSSG in the samples were determined using external standards ranging from 5 to 50 and from 1.56 to 100 μM, respectively.

Oxidized low-density lipoprotein assay

A monoclonal antibody (mAb-4E6)-based ELISA kit (Dynamika MH, Australia) was used for the quantification of oxidized low-density lipoprotein (oxLDL) in plasma. The assay was performed in duplicate and expressed in U/L.

Statistical analysis

SigmaPlot software (version 11.0) was used to analyze all data. Two-tailed, unpaired t tests (Student's t test or Mann–Whitney rank sum test) tested for differences in biochemical/clinical variables between GS and control groups. Group data are reported as means±standard deviation or median (25–75% interquartile range). Pearson correlation tested the relationship between unconjugated bilirubin and other dependent variables. Analysis of the relationship between unconjugated bilirubin and protein carbonyls was tested by dividing the group into quartiles each containing approximately the same number of subjects. These groups included persons with <10 (n=11), 10.1–17.1 (n=11), 17.2–40 (n=15), and >40 μM (n=7) unconjugated bilirubin. The chosen cutoffs are representative of the concentrations defining increased (<10 μM) and decreased (>10 μM) risk of CVD [1] and the approximate unconjugated bilirubin concentration associated with visible jaundice (<40 μM versus >40 μM) [28]. All antioxidant and blood lipid/glucose parameters were entered as independent variables in a forward stepwise regression analysis and their effects on oxidative stress-dependent variables were analyzed. The level of significance was set at P<0.05.

Results

General biochemistry and lipid status

Serum glucose, liver enzyme concentrations, and the lipid profiles of recruited subjects can be found in Table 1. GS subjects possessed a reduced total cholesterol and LDL concentration in addition to an increased HDL:LDL ratio (P<0.05 all variables) compared to control subjects. Lipid variables (total cholesterol, triglycerides, LDL, HDL:LDL ratio) were significantly correlated or tended to be correlated with bilirubin concentrations when the cohort was investigated collectively (P=0.03, 0.193, 0.01, 0.02, respectively; Supplementary Figs. S1–S4). However, the significance of these relationships was lost when analyses within control and GS groups were conducted (data not shown). Lipid status in Gunn rat and littermate controls generally supported the lipid biochemistry data in patients showing that hyperbilirubinemic rodents possessed significantly reduced total cholesterol and HDL concentrations (P<0.001; Table 2). Hyperbilirubinemia and hypocholesterolemia in Gunn rats were also associated with significantly reduced body mass (P=0.002; Table 2).

Table 1

Clinical characteristics of the recruited GS and control subjects (n=22 per group).

Variable	GS	Control	P value
Age (years)	32.6±11.2	32.6±10.9	0.860
BMI (kg/m2)	22.3±2.78	23.7±4.07	0.208
Glucose (mmol/L)	4.69±0.38	4.85±0.56	0.300
Total cholesterol (mmol/L)	4.48±0.82	5.31±1.29	0.014⁎
Triglycerides (mmol/L)	0.83±0.31	1.06±0.49	0.068⁎⁎
HDL (mmol/L)	1.71±0.44	1.59±0.43	0.345
LDL (mmol/L)	2.61±0.67	3.52±1.25	0.004⁎
HDL:LDL ratio	0.71±0.25	0.52±0.24	0.016⁎
Alanine aminotransferase (IU/L)	17.8±8.78	19.2±10.5	0.618
Aspartate aminotransferase (IU/L)	17.5±5.99	16.8±6.09	0.708
γ-Glutamyltransferase (IU/L)	21.7±24.6	17.3±10.6	0.440
Plasma heme (μmol/L)a	0.28 (0.21–0.38)	0.24 (0.16–0.32)	0.193

Data presented as median (25–75% interquartile range).

Significant difference (P<0.05).

Trend approaching significance (P<0.1).

Table 2

General characteristics, unconjugated bilirubin, and lipid status in female, middle-aged (13 months old), littermate-matched Gunn (n=7) and Wistar (n=5) rats.

Variable	Gunn	Wistar	P value
Mass (g)	208±29	272±27	0.002⁎
Unconjugated bilirubin (μmol/L)	41.4±8.39	1.06±0.37	<0.001⁎
Total cholesterol (mmol/L)	0.57±0.09	1.69±0.40	<0.001⁎
Triglycerides (mmol/L)	1.49±0.47	1.65±0.43	0.574
HDL (mmol/L)	0.12±0.05	1.06±0.21	<0.001⁎
LDL (mmol/L)	0.15±0.08	0.30±0.26	0.194
HDL:LDL ratio	0.87±0.45	6.75±5.39	0.015⁎

Significant difference (P<0.05).

Antioxidant status

Unconjugated bilirubin concentrations varied from 5.1 to 85.3 μM (0.3–5.0 mg/dl) in the whole cohort. Ascorbate, protein, and uric acid concentrations were not significantly different between the groups; however, unconjugated bilirubin, reduced thiol, and GSH concentrations were significantly elevated in GS (P<0.05, Table 3). Global measures of antioxidant status, TEAC, and FRAP, which were nonsignificantly greater in GS (Table 3), significantly correlated with bilirubin concentrations (P<0.05; Supplementary Figs. S5 and S6). Of the elevated antioxidants in GS, reduced thiols and GSH were positively correlated with circulating bilirubin concentrations (r=0.412; P<0.01; r=0.444; P<0.01, respectively; Fig. 1).

Table 3

Serum/plasma antioxidant status in GS and controls (n=22 per group).

Variable	GS	Control	P value
Unconjugated bilirubin (μmol/L)	35.1±18.2	10.3±2.89	<0.001⁎
Ferric reducing ability of plasma (mmol Fe2+/L)	1.12±0.23	1.02±0.19	0.129
Trolox equivalent antioxidant capacity (mmol Trolox eq/L)	1.36±0.13	1.30±0.10	0.073
Uric acid (μmol/L)	266±72.1	272±75.6	0.793
Ascorbate (μg/ml)	6.02±1.87	6.22±1.71	0.710
Reduced thiols (μmol/L)	550±64.2	493±83.3	0.015⁎
Reduced thiols (nmol/mg protein)	8.03±1.09	6.75±1.39	0.002⁎
Reduced glutathione (μmol/L)	12.7±2.39	9.44±2.45	<0.001⁎
Protein (mg/ml)	69.1±7.90	74.6±12.0	0.083

Significant difference (P<0.05).

Fig. 1

Correlation between (A) reduced thiols (r=0.412; P<0.01) and (B) GSH (r=0.444; P<0.01) and unconjugated bilirubin concentration in GS (●) and control (○) subjects (n=22 per group).

Oxidative stress status

The GSH:GSSG ratio was significantly greater in the GS group (Table 4; P<0.001) and was positively correlated with bilirubin concentration (r=0.441, P<0.01; Fig. 2). The protein carbonyl concentration was negatively associated with bilirubin (r=−0.505; P<0.001) and was lower in persons with >40 μM bilirubin (n=7), versus non-GS subjects (Fig. 3). OxLDL concentrations decreased with increasing bilirubin concentration (Fig. 4A). Elevated bilirubin was also associated with reduced total and LDL cholesterol concentrations (Supplementary Figs. S1 and S2). LDL cholesterol concentrations effectively explained a large proportion of variance in oxLDL concentrations, which was significantly decreased in GS (Fig. 4B, Tables 4 and 5). Interestingly, when oxLDL concentrations were expressed relative to the LDL concentration, the oxidative stress marker was significantly increased in GS (Table 4).

Table 4

Oxidative stress status in GS and control subjects (n=22 per group).

Variable	GS	Control	P value
Protein carbonyl (nmol/mg protein)	0.18±0.04	0.19±0.03	0.242
GSSG (μmol/L)	27.1±9.44	31.7±7.48	0.081
GSH:GSSG ratio	0.49±0.16	0.32±0.12	<0.001⁎
oxLDL (U/L)	16.2±2.24	18.6±3.57	0.011⁎
oxLDL (U/mmol LDL)	6.47±1.27	5.63±1.32	0.037⁎

Significant difference (P<0.05).

Fig. 2

Correlation between GSH:GSSG ratio and unconjugated bilirubin concentration in GS (●) and control (○) subjects (n=22 per group; r=0.441; P<0.01).

Fig. 3

(A) Correlation between protein carbonyl and unconjugated bilirubin concentration (r=−0.505; P<0.001) in GS (●) and control (○) subjects (n=22 per group). (B) Box plot showing the protein carbonyl concentration in unconjugated bilirubin quartiles (<10 μM unconjugated bilirubin, n=11; 10.1–17.1 μM, n=11; 17.2–40 μM, n=15; and >40 μM, n=7). *P<0.05 versus all other groups.

Fig. 4

(A) Correlation between oxLDL and bilirubin concentration (r=−0.277; P=0.069). (B) OxLDL and LDL concentration (r=0.879; P<0.001) in GS (●) and control (○) subjects (n=22 per group).

Forward stepwise regression analysis revealed that bilirubin was the only independent variable, of those tested, that influenced reduced thiol concentration, GSH:GSSG ratio, and protein carbonyl formation. Together, LDL, glucose, uric acid, and reduced thiols explained more than 85% of the variance in oxLDL concentrations (Table 5).

Table 5

Regression modeling of independent variables.

Dependent variable	R2 value
	LDL	Glucose	Uric acid	Reduced thiols	Bilirubin	Total
GSH:GSSG	–	–	–	–	0.194	0.194
Reduced thiols	–	–	–	–	0.197	0.197
Protein carbonyls	–	–	–	–	0.255	0.255
OxLDL	0.772	0.037	0.035	0.015	–	0.859

R2 values represent the percentage (×100) of variance that the independent variables could explain in the dependent variable within the model.

Discussion

This report documents reduced baseline oxidative stress in GS, which was specifically related to protection from thiol oxidation. Reduced thiol concentration and GSH:GSSG ratio increased and protein carbonyl concentration decreased in association with increased circulating bilirubin concentration. By employing forward stepwise regression analysis, multiple additional (and cumulative) associations between independent variables (LDL, glucose, uric acid, reduced thiols, and bilirubin) and biomarkers of cardiovascular risk (LDL, oxLDL, protein carbonyl) have been identified in this cohort. These data suggest bilirubin's antioxidant effects are only one of many possible mechanisms leading to reduced mortality in this population.

Bilirubin antioxidant effects

In this study, persons with GS tended to possess elevated measures of global antioxidant capacity (Table 3), in agreement with previous findings [5,10]. The difference in FRAP (nonthiol antioxidant capacity) could be explained by the difference in bilirubin concentration alone (Table 3; ∼100 μM Fe2+ equivalents, each bilirubin molecule reduces four ferric tripyridyltriazine radicals). The difference in TEAC (Table 3; ∼60 μM Trolox equivalents; radical oxidizes all major antioxidants in plasma) was greater than could be explained by the difference in bilirubin concentration alone. Bilirubin can reduce 1.5 azinobis-(3-ethylbenzothiazoline-6-sulfonic acid) radicals and, therefore, the difference in bilirubin concentration between the groups accounted for ∼37.5 μM Trolox equivalents. Therefore, it was hypothesized that preserved thiol groups explained the additional TEAC antioxidant capacity in GS. Elevated reduced thiol concentrations were subsequently confirmed in GS (Table 3). These findings are supported by elevated circulating sulfhydryl groups in a small, unmatched Hungarian population of GS and control subjects [29]. Plasma GSH levels were then explored and were found to be significantly elevated in GS (Table 3) and were also positively correlated with circulating bilirubin concentrations (Fig. 1). These data suggested dose-dependent protection from GSH oxidation in GS and/or, intriguingly, that glutathione synthesis may be enhanced by bilirubin [30].

Bilirubin is associated with reduced macromolecule oxidation and modulates lipid profile

Excessive radical production relative to antioxidant bioavailability is fundamentally related to the likelihood of macromolecule modification and underpins past and present attempts to prevent/treat diseases using antioxidant therapy [31]. To date, only Dennery et al. [13] have shown that endogenously elevated bilirubin protects from oxidative stress in rats exposed to hyperoxia. These data are supported by animal studies showing that elevated endogenous or exogenously applied bilirubin protects from bile acid- [32] and hydroxyl radical- [33] induced oxidative stress, ex vivo.

Little evidence of bilirubin protection from in vivo protein oxidation exists. Bilirubin protects albumin from in vitro oxidation by superoxide, hydrogen peroxide, hypochlorite, and peroxynitrite [34,35]. Reaction of superoxide with nitric oxide (NO) forms peroxynitrite that can oxidize thiol groups and proteins [36,37]. Bilirubin scavenges both superoxide [34] and nitrite and inhibits the release of NO [37,38] and, therefore, the formation of peroxynitrite. Bilirubin also directly scavenges peroxynitrite and inhibits the chain reaction of oxidation through the protein backbone [35], which leads to inevitable terminal protein carbonyl formation. Protection of proteins from oxidation is important for the maintenance of protein/enzyme function and for protection from a variety of diseases, including CVD, via apolipoprotein B oxidation [39]. A meta-analysis has concluded that a bilirubin concentration of below 10 μM delineates a cutoff for increased risk of CVD in men [1]. Protein carbonyl concentration is related to CVD [40]; however, we observed no difference in carbonyl concentrations between the lowest three bilirubin quartiles (0–10, 10.1–17.1, 17.2–40 μM). Despite this, dose-dependent protection from protein carbonyl formation exists, supported by lower carbonyl concentrations in persons with >40 μM bilirubin (Fig. 3). These data suggest a molar ratio of ∼1:10 bilirubin:albumin is required to significantly protect from protein carbonyl formation in vivo.

In vitro evidence supports a role for bilirubin in protecting lipids from various radical species [8]. A recent study by Tapan et al. [41] demonstrated significantly reduced oxLDL in GS compared to controls. Indeed, we showed that oxLDL levels were correlated with bilirubin (Fig. 4A); however, LDL was also correlated with oxLDL concentrations (Fig. 4B). Clearly LDL concentration inherently affects the likelihood of LDL oxidation; however, so would bilirubin and other antioxidants. Therefore, we employed regression modeling to quantitate the effects of multiple independent variables on dependent variables, including oxLDL. The results showed that decreased LDL concentrations (in the entire cohort) largely (77.2%) influenced oxLDL concentrations. However, glucose, uric acid, and free thiol concentrations further and significantly explained the variance in oxLDL concentrations (total prediction 85.9%). Regression modeling also showed bilirubin was the only variable measured that influenced the concentration of reduced thiols and, therefore, may assist in protecting LDL from oxidation, by preserving thiol status. It should be noted, however, that this effect is probably minor.

The significant increase in oxLDL, when expressed relative to LDL content in GS (Table 4), was an interesting and unexpected finding of this study. Although the significance of this finding to atherogenesis is questionable (because absolute oxLDL concentrations were lower in GS), this finding suggests that circulating LDL was more susceptible to oxidation in GS. This observation is in disagreement with recent reports showing elevated circulating bilirubin improves resistance of plasma lipids and LDL to oxidation, when quantified using lag-phase duration [5,42]. Interestingly however, when total oxidation after 6 h was quantified relative to lipid content, we previously showed a trend toward elevated lipid oxidation in GS samples (P=0.166) [5]. A theoretical basis for such a conclusion was proposed by Stocker and Ames [43], who hypothesized that bilirubin ditaurate could also act as a peroxidase, liberating lipid peroxide radicals in the presence of copper. Therefore, we speculate that additional bilirubin in the plasma of GS probably delays the initiation of copper-induced oxidation; however, it might result in additional lipid peroxidation in the longer term as documented here.

Relationships between circulating bilirubin and lipid status have been presented in previous studies, although these findings have gone essentially unnoticed or have been published only recently. For example, reduced circulating total cholesterol [41], LDL [44], triglyceride concentrations [6], and BMI [45] and elevated HDL:LDL ratio [5] have been documented in GS. Although these findings are by no means consistent between all published studies [46,10] (because of differences in subject matching and cohort characteristics), they generally agree with findings in this cross-sectional, age-, gender-, and BMI-matched study (see Table 1, Supplementary Figs. S1–S4). These data are further supported by clinical data showing lower total cholesterol [47,48] and BMI [49,50] in persons with elevated circulating bilirubin, in addition to significant negative correlations between lipid parameters and circulating bilirubin species [49,51].

We provide additional data supporting a possible role for unconjugated bilirubin (or UGT1A1 activity) in reducing circulating total and HDL cholesterol concentrations in the hyperbilirubinemic Gunn rat (Table 2). The greater reduction in HDL (versus LDL) is probably reflected by a greater contribution of HDL to total cholesterol in rats and suggests underlying perturbation of cholesterol metabolism/excretion/absorption. This speculation is supported by decreased total cholesterol and, importantly, LDL in GS (Table 1), which forms the greatest component of total cholesterol in humans. Interestingly, hypocholesterolemia in the Gunn rat was accompanied by significantly decreased body mass (Table 2). Therefore, it would seem reasonable to speculate that the perturbation of lipid metabolism associated with elevated bilirubin (or reduced UGT1A1 activity) is species independent and responsible for reduced body mass in Gunn rats and BMI in human studies [49,50]. These findings will encourage investigations exploring how bilirubin modulates lipid metabolism, which could further explain protection from ischemic heart disease in GS [10].

Bilirubin may also impart protection from atherosclerosis by attenuating endothelial dysfunction [52] in response to oxidative stress and inflammation. Interestingly, increased expression of heme-oxygenase (HO-1) would degrade more heme and generate more bilirubin in rabbit atherosclerotic lesions [53]. Circulating heme concentrations were not significantly elevated in the GS group of this study and suggest that HO-1 is not responsible for the cardiovascular protection in GS. However, bilirubin also accumulates in foam cells [53], which suggests elevated bilirubin could protect plasma [5] and LDL [42] from oxidation in atherosclerotic plaques, where copper [54] accumulates. Together multiple beneficial effects of bilirubin may exist and explain the remarkably decreased risk of CVD in GS, in addition to novel conclusions presented here.

In conclusion, these data show that elevated circulating bilirubin is associated with improved circulating thiol status and hypocholesterolemia, which could protect from atherosclerosis in GS via the “response to retention” and “oxidation modification” hypotheses [55]. The possible mechanisms responsible for these effects represent a very novel and important avenue for future research. The search for safe partial inhibitors of UGT1A1 that cause mildly elevated circulating unconjugated bilirubin concentrations could support current efforts to induce a GS phenotype [56,57] for translation into the clinic [9].