**3. HO-1, BVRA, and Bilirubin as Inflammatory Mediators**

*3.1. Exercise-Induced HO-1 as a Mediator of Immune System Responses*

HO-1 produces a known potent antioxidant and enzyme responsible for the cleavage of heme, yielding biliverdin, iron (Fe2+), and carbon monoxide (CO) [41,42]. The HO-1 pathway also regulates some of the metabolic and inflammatory aspects of insulin

resistance. While there is a connection between inflammation and the development of insulin resistance, it is unclear which development precedes and which is causative [43]. HO-1's role in inflammation and insulin resistance appears equivocal in the literature.

The presence of HO-1 mimics the same efficacious properties as bilirubin [44]. Bilirubin downregulates the M1 macrophages associated with the release of pro-inflammatory cytokines [45–47]. Future work to understand how HO-1 may affect M1 cells may shed light on potential underlying mechanisms to explain how bilirubin downregulates M1 cells. HO-1 also protects the liver from ischemia-reperfusion injury by modulating the macrophage phenotype into the anti-inflammatory M2 state in mouse livers [48,49]. This serves as evidence for an HO-1 role as a hepatic cryoprotective agent. In this same study, low HO-1 mRNA levels in human liver transplants correlated with increased expression of M1 pro-inflammatory markers [48,49]. Liver-specific biliverdin reductase A (BVRA) knockout animals with reduced hepatic bilirubin had worsened fatty liver on a high-fat diet compared to littermate controls [50], which was confirmed in global BVRA knockout animals [51]. Bilirubin reduces lipid content and inflammatory markers in mouse models of obesity-induced NAFLD [46,47].

Work by Gobert et al. found HO-1 to prevent an inflammatory response and has implicated HO-1 as a virulence factor in *H. Pylori* and other bacteria in order to evade the immune system [52]. Other work has described using a heme-inducing compound to effectively reduce obesity, insulin sensitivity and increase serum adiponectin levels. Inhibition of the HO-1 system decreased adiponectin and increased pro-inflammatory cytokines, TNFα, IL-6, and IL-1 [53–55]. Adiponectin, a known anti-inflammatory hormone, is thought to be working indirectly through HO-1-activation [56]. The complete mechanism of this anti-inflammatory activity is not fully understood, but some theories with convincing evidence reveal new insight on HO-1 and the importance of its catabolic products.

The HO-1 pathway can decrease inflammation by producing biliverdin/bilirubin, which has protective anti-inflammatory effects, especially in vascular endothelial tissue [57]. Another anti-inflammatory action of HO-1 is through carbon monoxide production, which is a known cryoprotectant and anti-apoptotic factor in endothelial cells that have the potential to crosstalk with nitric oxide, a known vasodilator. Although this work serves as further evidence that HO-1 has important underlying anti-inflammatory and insulinsensitizing mechanisms that may augment bilirubin's therapeutic value, it is essential to note that the supporting evidence is not unequivocal. In contrast to these aforementioned findings, conflicting data suggest that HO-1 is implicated in driving inflammation and may even support insulin resistance in humans. Jais et al. demonstrated that HO-1 levels predict a strong positive prediction of metabolic disease in human subjects [58], while Ghio et al. reported HO-1 elevation due to cytokine stimulation in inflammatory disease [59]. Whether HO-1 is present in response to the inflammation or if it is the direct cause of inflammation is not completely clear.

Although HO-1's direct role on insulin resistance and inflammation has not been fully elucidated, the influence of exercise on the HO-1 pathway may shed light on these equivocal data. Niess et al. showed that HO-1 expression in leukocytes increased significantly after sustained endurance exercise in marathon runners. The authors interpreted this to be due to the excessive amount of free radical production, although the mechanism that causes this upregulation of HO-1 in exercise is not completely clear [60]. However, it can be postulated that since exercise promotes ROS generation, it would induce nuclear factor (erythroid-derived 2)-like-2 (Nrf2) expression, which is a key transcription factor in inducing HO-1 [61–63]. A potential explanation for the upregulation of HO-1 may be that acute exercise can also propagate a transient pro-inflammatory state to increase levels of HO-1 via increased cytokine activity. Others have suggested that ROS, themselves, can induce and upregulate HO-1. Kurata et al. found that the HO gene was induced by 12-O-tetradecanoylphorbol 13-acetate response element (TRE) in the presence of hydrogen peroxide, a ROS [64]. These oxidative free radical levels vary based on habitually trained versus untrained subjects. The trained individuals had a much more robust adaptive

antioxidant defense system and thus a lower level of ROS production [65]. HO-1 levels at rest are significantly reduced in trained subjects compared to untrained subjects [60]. This suggests an adaptive regulatory feedback mechanism to which, at rest, basal ROS are downregulated in trained individuals and hence, a lower HO-1 level. The prevailing hypothesis surrounding this observation is that HO-1 is upregulated to offer protection from the free radicals that are produced with exercise (Figure 3) [59].

#### *3.2. The Emerging Role of Biliverdin Reductase in Immune Response*

BVRA plays a vital role in macrophage polarization and as a target for regulating responses to bacterial lipopolysaccharides and complement activation products. BVRA is expressed in macrophages where it is tyrosine phosphorylated. Phosphorylated BVRA then binds to phosphatidylinositol 3-kinase (PI3K) at the p85α subunit to activate downstream signaling to Akt [66,67]. Macrophage classification occurs according to activation state and function. M-1 macrophages are classically activated macrophages that express cytokines such as TNFα and interleukin-17A. M-2 macrophages are alternatively activated macrophages that express anti-inflammatory cytokines such as interleukin-10 (IL-10) and transforming growth factor-beta (TGFβ). Overexpression of BVRA in macrophages elevates expression of M-2 macrophage markers, while knockdown of BVRA increases M-1 macrophage markers [68]. Renal ischemia-reperfusion injury increases the levels of BVRA positive macrophages increasing the levels of IL-10, helping in the reparative process [68]. The recruitment of macrophages is an influential process in the inflammatory response. Release of chemokines that act on specific receptors such as the complement activation fragment 5a receptor one (C5aR1) recruits macrophages to sites of tissue injury. Loss of macrophage BVRA results in greater levels of C5aR1 increasing inflammation [69]. These studies demonstrate the critical role of BVRA in both macrophage chemotaxis and polarization. Augmentation of macrophage BVRA levels may be an effective treatment to bolster anti-inflammatory pathways in a number of inflammatory diseases. How they might affect metabolic adaptations to exercise is yet to be determined.

#### *3.3. The Effect of Exercise on Bilirubin and Its Actions*

Given that HO-1 expression is directly influenced by exercise training (Figure 4), it is logical to assume that exercise increases plasma bilirubin levels. Hinds et al. recently conducted a study where rats genetically selected for high capacity running (HCR) and low capacity running (LCR) were used to identify the metabolic pathways in the liver altering plasma bilirubin levels through exercise [70]. The investigators observed that HCR rats had significantly greater plasma bilirubin and hepatic BVRA expression while having a reduced expression of the glucuronyl hepatic enzyme UGT1A1. Significant increases in PPARα-target genes were also observed in HCR rats compared to the LCR. For the first time, these results suggest hepatic mechanisms involved in bilirubin synthesis and metabolism that may explain the positive effects of exercise on plasma bilirubin and metabolic health.

There are a limited number of articles published on this topic in humans [71,72]. In a controlled study that examined different levels of training intensity, researchers found that the high-intensity training group (defined as 12 kilocalories per kilogram per week (KKW) energy expenditure) presented a significant increase in total serum bilirubin in comparison to the sedentary control group. Those who trained at moderate intensity levels (defined as 4 and 8 KKW) experienced no significant differences in serum bilirubin levels [71]. Priest et al. observed an increase in bilirubin in male runners after a 13-mile run along with an increase in alkaline phosphatase. Bile acids and bilirubin have been shown to be elevated in these runners [73]. In both studies, bilirubin levels seem to be elevated in response to high-volume, exhaustive forms of exercise with high energy expenditure.

**Figure 4.** The heme oxygenase pathway signaling during exercise. Exercising (aerobic) raises plasma bilirubin levels by (1) suppression of the glucuronyl transferase enzyme UGT1A1 that conjugates bilirubin for removal from blood, and (2) activation of the heme oxygenase pathway (HO-1-BVRA-PPARα). The increased bilirubin combats reactive oxygen species (ROS) and ROS-induced inflammation and DNA damage. The bilirubin also activates the nuclear receptor transcription factor, PPARα, to reduce adiposity. Created with BioRender.com (accessed on 9 December 2021).

A subgroup analysis from Swift et al. revealed another interesting trend that showed those who were insulin resistant in the high-intensity exercise group had a significant increase in bilirubin compared to the insulin-sensitive group [71]. A more recent study confirmed these observations by examining why moderate-to-vigorous physical activity (MVPA) resulted in a significant increase in serum bilirubin in insulin-resistant subjects but not in insulin-sensitive subjects. The authors hypothesized that the observed increase in bilirubin in the insulin-resistant subjects could be due to a lower basal level of bilirubin, resulting in a more remarkable absolute change in bilirubin in response to MVPA [74]. The underlying rationale for this pattern of bilirubin change in response to different exercise volumes should be further explored to improve our understanding of the connection between insulin resistance and changing bilirubin levels.

Several studies in athletes have also reported a strong correlation between elevated bilirubin and the degree of exercise intensity and an associated increase in erythrocyte hemolysis [71,75–77]. Witek et al. reported normal bilirubin reference ranges for 339 male and female Polish athletes [72]. While approximately 45% of the samples had bilirubin levels in the range of 7–14 μM, 12% of the athletes had 21–28 μM. Nineteen percent of the total bilirubin values exceeded the established normal limit of 21 μM. These elevated concentrations appeared to be related to changes caused by regular exercise and were not directly related to increased hemolysis. The authors suggested that other exercise-induced mechanisms seem to affect bilirubin concentrations, such as altered liver function and upregulation of bilirubin production (to serve as an antioxidant) in response to increased oxidative stress (ROS). A study of young Polish athletes (aged 18–40 years) reported that bilirubin levels increased in response to both a ketogenic diet and short-term, high-intensity exercise (CrossFit) [78]. Study subjects increased their bilirubin concentrations in both diet groups in response to exercise (Customary diet: 10 ± 5 to 19 ± 8; Ketogenic diet: 14 ± 0 to 20 ± 8 μM; *p* < 0.05). These studies bring to question if bilirubin levels are being controlled by exercise to correct metabolic imbalances, mitigate oxidative stress, and reduce inflammation.

#### *3.4. The Hormonal Function of Bilirubin in Exercise and the Impact of PPAR Signaling*

The PPARs are a subfamily of ligand-activated nuclear receptor transcription factors with three distinct isoforms: α, β/δ, γ [79]. These isoforms are found in different tissues, each with a predominant isoform. PPARα is expressed in hepatocytes, enterocytes, and vascular endothelium and works to improve mitochondrial efficacy in FA oxidation in these tissue types. PPARβ/δ are expressed more ubiquitously in the body but predominate in skeletal muscle and macrophages and are important in fatty acid oxidation and macrophage immunosuppression through the reduction in NF-κB inflammatory cytokines [80,81]. PPARγ is found mainly in white and brown fat adipocytes and enhances genes involved in the metabolism of glucose and adipocyte differentiation [82–86]. PPAR's are activated in the presence of their specific corresponding natural or synthetic pharmacological ligands. All PPAR isoforms will activate in the presence of unsaturated fatty acid (PPAR pan agonist), which acts as a ligand to the PPAR isoforms [87]. It should be noted that all of the PPAR isoforms are considered to drive anti-inflammatory pathways. A hepatocyte-specific and adipocyte-specific knockout of PPARα in mice fed a high-fat diet showed greater fat content in each of the KOs, which both also exhibited significantly higher inflammation compared to control littermates [88,89]. Similarly, studies showing that overexpression of inflammatory meditator glucocorticoid receptor beta (GRβ) in the liver of C57/bl6 mice induced hepatic lipid accumulation in 5 days on a normal chow diet by suppression of hepatic PPARα [90].

We have shown that bilirubin (unconjugated form) binds directly to the PPARα nuclear receptor to induce transcription of genes (Figure 5) [91–94], which control adiposity and glucose sensitivity. Interestingly, competitive binding studies and transcriptional activity assays demonstrated that bilirubin's binding to the PPARs is specific to only PPARα, and it has no actions or binding to PPARγ or PPARβ/δ [91,92]. In looking more specifically at ligands for PPARα, a synthetic ligand such as fenofibrate (fibrates) is widely used in the treatment of hypertriglyceridemia in order to reduce serum triglyceride levels. Through the binding and subsequent activation of the PPARα nuclear receptor, fenofibrate reduces plasma triglycerides and VLDL/LDL concentrations [95]. An increased expression of PPARα offers significant induction of β-oxidation [46,47,63,92,94] and myocardial ATP production, which are markers for myocardial viability [96,97]. It can also reduce the oxidative stress that occurs after a high-fat meal [95]. As mentioned above, unconjugated bilirubin has been demonstrated to act as a novel endocrine ligand that activates the transcriptional activity of PPARα by direct interaction, which changes coregulator proteins bound to the nuclear receptor to control gene activity [92]. PPARα activation by bilirubin in obese mice with glucose intolerance leads to a decrease in fasting blood glucose, as well increase in lean body mass and an increased presence of FGF21 (fibroblast growth factor 21) [42]. FGF21 can act as a metabolic regulator by rapid reduction in blood glucose and insulin levels in obese models [42,98] (readers are referred to another review discussing modulation of metabolism by FGF21 for more information [99]). The impact that bilirubin has on exercise via FGF21 is unknown. More studies are needed to elucidate the protective properties of bilirubin that occur via it driving the PPARα-FGF21 pathway that reduces adiposity and improves insulin sensitivity.

**Figure 5.** Selectivity of bilirubin for the PPAR isoforms and signaling mechanisms. The PPAR isoforms are bound by corepressors proteins until they are bound to the ligand, which induces a change from co-repressors to co-activators. Unconjugated bilirubin enters the cells and activates PPARα and not the PPARγ or PPARβ/δ isoforms. Bilirubin binding to PPARα induces a complex with RXR causing an exchange of corepressor proteins for co-activators. The bilirubin-induced PPARα-RXR complex controls specific genes for metabolic control of adiposity (*UCP1*, *CPT1A*, *FGF21*, *ADRB3*, and others), which might be based upon specific co-activators (PGC1α, NCOA1, NCOA2, MED1, etc.) bound in the complex.

Exercise plays a role in the activation of the PPAR systems. Exercise increases the levels of AMP-activated kinase (AMPK), ERK1/2-MAPK, and PKC, which are kinases in the skeletal muscle involved in increasing the expression of downstream transcription factors. These kinases are found to increase the transactivation of PPARα and thus an increase in FA oxidation and glucose production, which can be used as fuel during exercise [100]. PPARα, in particular, has strong actions on improving the efficacy of FA oxidation in the liver and adipose tissues [88,89]. PPARα mRNA upregulates in bouts of exercise and in times of starvation in order to metabolize fat and use it for an effective energy source [101,102]. Acute exercise also provides increases in liver and serum FGF21, which provides systemic insulin sensitization [103]. PPARα expression is necessary for optimized endurance exercise. PPARα knockout models had significantly less tolerance to endurance exercise than the control. This lack of tolerance is due to a rapid depletion of hepatic glycogen [104]. We have shown that reducing PPARα activity in the liver leads to lower hepatic glycogen content [88,90], and activation by bilirubin increases it [47,70]. Similarly, hepatocyte-specific BVRA knockout animals on a high-fat diet had reduced bilirubin-PPARα activity and lower glycogen levels [50]. Endurance athletes were found to have a specific polymorphism that produces an increased binding capacity of PPARα in skeletal muscle and more type I slow-twitch fibers [105]. This suggests that PPARα may have critical roles in exercise and is necessary to perform enhanced endurance activity [106]. Similar to PPARα, PPARγ and PPARβ/δ mRNA is also elevated as a result of an aerobic exercise training program [107,108]. PPARβ/δ are the least studied of the isoforms. There is evidence to support PPARβ/δ's ability to rectify metabolic disorders and enhance βoxidation in muscle [109]. Many of its effects mimic the functionality of PPARα; however, the PPARβ/δ is more ubiquitously expressed than PPARα [110].

PPARγ is upregulated after sustained exercise programs and showed beneficial effects in skeletal muscle [111]. This skeletal muscle had signs of mitochondrial biogenesis and thus, improved aerobic respiration. The mitochondrial biogenesis is also seen in adipose tissue and is phenotypically evident by the increased conversion of white fat into brown fat in the presence of a highly induced PPARγ [112]. This exercise-induced PPARγ can provide antidiabetic effects through upregulation of monocyte PPARγ-control genes [111]. PPARγ is also in charge of controlling adipocyte differentiation [113,114]. In a PPARγ knockout model, severe lipoatrophy is observed, along with insulin resistance [115]. The PPARγ knockout mice have significantly decreased body mass; however, the liver showed a 1.5-fold increase in weight and increased lipid deposition in hepatic tissue. The increased lipid deposition in the liver is due to disrupted adipogenesis in white adipose tissue (WAT), causing increased plasma triglycerides that can deposit in the liver [116]. These PPAR systems have been correlated with decreased levels of atherosclerosis, insulin resistance, and inflammation in conjunction with metabolic syndrome and hypertriglyceridemia [114].

Another novel metabolic role designated to bilirubin is its natural ability to act as an insulin sensitizer [44,117]. PPARγ is elevated following bilirubin administration in mice with improved insulin sensitivity. This isomer of PPAR is implicated as a potent factor in adipocyte differentiation and adiponectin secretion [117]. Bilirubin administration has also improved obesity and hyperglycemia in rodent models. Bilirubin-treated obese mice increased phosphorylation of Akt (Thr309), an insulin-signaling molecule, in skeletal muscle and hepatocytes, indicating preservation of insulin sensitivity [44]. Bilirubin-treated mice also presented with greater adiponectin levels [117]. It should be noted that while bilirubin induced PPARγ expression in diabetic mice, it is not a ligand for this receptor as was previously demonstrated [91,92]. Because bilirubin levels rise with exercise more effectively in insulin-resistant subjects, there is therapeutic potential for bilirubin to control cholesterol metabolism and glucose tolerance in insulin-resistant patients. Therefore, exercise in pre-diabetic patients may offer metabolic benefits by raising HO-1, upregulating adiponectin and bilirubin levels, enhancing insulin signaling, activating PPARα pathways, and thus, decreasing insulin resistance (Figure 6).

**Figure 6.** Model of how HO-1-bilirubin improves insulin sensitivity. Upregulation of HO-1-bilirubin is a multifactorial influencer of different metabolic processes such as induction phosphorylated Akt (Thr309), adiponectin production, and activation of PPARα-FGF21 pathways.

#### **4. The Signaling Mechanisms of Heme Oxygenase and Bilirubin in Metabolism**

#### *4.1. Generation and Catabolism of Bilirubin*

Bilirubin is a tetrapyrrole compound formed from the catabolism of heme to biliverdin that is converted to bilirubin by biliverdin reductase (BVR) [17,41,42,118–122]. Tetrapyrroles are seen as an orange-yellow pigment, which may indicate underlying disease processes if extremely elevated (>150 μM) in the skin (jaundice) or the urine [41]. When erythrocytes (red blood cells) are lysed, the hemoglobin is broken down into heme and protoporphyrin. The heme is oxidatively cleaved by the enzyme heme oxygenase (HO), yielding biliverdin, iron, and carbon monoxide (CO) [41]. This biliverdin can be converted to bilirubin through the cytosolic enzyme biliverdin reductase [123–125]. The conversion to bilirubin has been empirically shown to produce potent antioxidant effects that can regulate cellular redox reactions, decrease ROS, and decrease the activity of NADPH oxidase [3]. Bilirubin circulates bound to water-soluble albumin, where it is transferred to the hepatocyte as unconjugated bilirubin. Then, bilirubin is conjugated by the UDP glucuronosyltransferase 1A1 (UGT1A1) enzyme [121,126]. Once conjugated, bilirubin is then metabolized by colonic bacterial proteases and is either reabsorbed into the hepatobiliary system as urobilinogen or excreted in the feces and urine as stercobilin or urobilin, respectively [122]. The bilirubin pathway (illustrated and described in more detail elsewhere [17,41,42,119–122,126]) is increased with exercise [70], and a better understanding of the pathway regulation may identify areas that alter the bilirubin half-life that might lead to pathological consequences.

#### *4.2. Biliverdin Reductase and Metabolism*

While there are limited studies showing that BVRA is regulated by exercise [70], there have been supporting studies showing a role for the enzyme in metabolism [50,51,124,125,127]. Adipocyte-specific deletion of BVRA results in adipocyte hypertrophy and increased inflammation while decreasing mitochondrial number and markers of adipocyte browning such as PPARα and β3 adrenergic receptor (*Adrb3*) [127]. The loss of adipocyte BVRA also decreases insulin signaling in white adipose tissue contributing to increased fasting hyperglycemia in knockout mice [127]. These results agree with the finding from obese human patients who exhibit lower levels of BVRA, increased levels of inflammation, and increased adipocyte size [128]. CRISPR knockout of BVRA in hepatocytes and kidney proximal tubules cells induces oxidative stress and lipid accumulation [124,125]. Similarly, mice with a global knockout of BVRA have increased oxidative stress [41]. Deficiencies in BVRA also correlate with brain insulin resistance in Alzheimer's disease patients [129,130]. Administration of BVRA peptides improved intranasal insulin treatment in a mouse model of Alzheimer's disease, suggesting a potential therapeutic role for targeting BVRA for treatment [131].

BVRA also plays an essential role in the development of metabolic diseases associated with obesity, like NAFLD. Hepatocyte-specific BVRA knockout mice develop more severe dietary-induced NAFLD as compared to wild-type littermates [50]. The loss of hepatocyte BVRA increases activation of glycogen synthase kinase 3beta (GSK3β) via decreased levels of serine 9 (Ser9) phosphorylation which in turn increases serine 73 (Ser73) phosphorylation of PPARα, increasing protein turnover and decreasing its transcriptional activity [50]. Interestingly, reduced adipocyte levels in BVRA in obese human patients resulted in significantly more hepatic steatosis and NAFLD [128]. These results suggest that BVRA can have both direct and indirect effects to contribute to hepatic steatosis and the development of NAFLD. More studies are needed to determine factors that regulate BVRA expression and how these are affected by exercise.

#### *4.3. Bilirubin and Metabolic Dysfunction*

Bilirubin was once believed to act only as a toxic bile substance and end-product. However, more recent studies have uncovered potential metabolic benefits of greater yet subclinical bilirubin concentrations. These include its role in ROS scavenging, anti-inflammatory properties, and reduction in adipocyte size from increased fat oxidation [94,132,133]. While

marked extreme hyperbilirubinemia (>150 μM) can be a sign of a more ominous clinical diagnosis, raised basal concentrations are also associated with protecting metabolic function (25–50 μM as discussed in [41]). The metabolic syndrome [134] is associated with increased insulin resistance and oxidative stress, which can also lend significant inflammatory and cardiovascular risk factors. Increased serum bilirubin concentration acts as a protective factor against the development of MetS. Subjects with increased basal bilirubin levels have a lower odds ratio to develop MetS [134]. It is thought this observation is due to the antioxidant, anti-inflammatory, and hormonal properties of bilirubin (discussed in more detail above). Conversely, in subjects diagnosed with metabolic syndrome, serum bilirubin is typically reduced (<10 μM, discussed further in [41]) [135]. Thus, the clinical assessment of serum bilirubin may have some future utility as a screening or prediction tool for those with high risk for metabolic dysfunction. In support of this, coronary artery disease severity was recently predicted with an odds ratio of 0.155 (95% confidence interval), revealing an inverse relationship between bilirubin and CAD severity [136]. NAFLD was also predicted in patients with an odds ratio of 0.88 (95% confidence interval), showing a strong inverse relationship between serum bilirubin and NAFLD [137]. A study of obese children showed that those with NAFLD had the lowest serum bilirubin [133]. Low bilirubin has also been associated with a greater risk of cerebral deep white matter lesions in healthy subjects [138], suggesting that low levels may impair cognitive function or lead to stroke [139–141]. These studies might suggest that increasing bilirubin levels could be therapeutic for improving metabolic dysfunction and reducing stroke risk. Factors that induce heme oxygenase production of bilirubin, such as nutraceuticals, may have several benefits [63]. These studies highlight a potential protective effect of bilirubin against metabolic disease and should be examined further to elucidate more of its positive benefits.
