Anti-inflammatory role of fetuin-A in injury and infection.

Infection and injury are two seemingly unrelated processes that often converge on common innate inflammatory responses mediated by pathogen- or damage-associated molecular patterns (PAMPs or DAMPs). If dysregulated, an excessive inflammation manifested by the overproduction and release of proinflammatory mediators (e.g., TNF, IFN-γ, and HMGB1) may adversely lead to many pathogenic consequences. As a counter-regulatory mechanism, the liver strategically re-prioritizes the synthesis and systemic release of acute phase proteins (APP) including the fetuin-A (also termed alpha-2-HS-glycoprotein for the human homologue). Fetuin-A is divergently regulated by different proinflammatory mediators, and functions as a positive or negative APP in injury and infection. It not only facilitates anti-inflammatory actions of cationic polyamines (e.g., spermine), but also directly inhibits PAMP-induced HMGB1 release by innate immune cells. Peripheral administration of fetuin-A promotes a short-term reduction of cerebral ischemic injury, but confers a long-lasting protection against lethal endotoxemia. Furthermore, delayed administration of fetuin-A rescues mice from lethal sepsis even when the first dose is given 24 hours post the onset of disease. Collectively, these findings have reinforced an essential role for fetuin-A in counter-regulating injury- or infection-elicited inflammatory responses.


Fetuin-A as a negative or positive APP
Fetuin-A was first isolated by Pederson more than sixty years ago as a major plasma protein in the fetus (Pedersen 1944). During fetal development, it is expressed in most organs including the liver, kidney, gastrointestinal tract, skin and brain (Terkelsen et al. 1998;Kitchener et al. 1997;Dziegielewska et al. 2000;Kitchener et al. 1999). In adults however, fetuin-A is produced primarily by the liver, and its synthesis is divergently regulated during injury or infection, classifying it as a negative or positive APP. ng/ml, reduced fetuin-A expression levels by as much as 50-70% (Li et al. 2011a). In contrast, HMGB1 (1 μg/ml), a late proinflammatory mediator of lethal systemic inflammation Yang et al. 2004;Wang et al. 2008), elevated fetuin-A expression levels by 2-3 folds in HepG2 cells, suggesting that different cytokines divergently regulate hepatic fetuin-A expression.

Elevation of Fetuin-A Levels during ischemia
In patients with cerebral ischemic injury (stroke), plasma fetuin-A levels were paradoxically elevated (Weikert et al. 2008;Tuttolomondo et al. 2010). The elevation of circulating fetuin-A levels correlated with an increase not only in LDL-cholesterol levels (Tuttolomondo et al. 2010) but also in risk of cardiovascular disorders (Weikert et al. 2008). Similarly, serum fetuin-A levels were increased up to 10-fold in cattle following traumatic injury (Dziegielewska et al. 1992), suggesting fetuin-A as a positive APP during ischemic or traumatic injury. Notably, HMGB1 can be passively leaked from injured cells (Peltz et al. 2009), and functions as an early mediator of traumatic or ischemic injury Wu et al. 2007;Liu et al. 2007b;Tsung et al. 2005;Tsung et al. 2007;Watanabe et al. 2005). It is thus plausible that HMGB1 participates in the up-regulation of hepatic fetuin-A expression during injury. In an animal model of focal cerebral ischemia (i.e., permanent middle cerebral artery occlusion, MCAo), fetuin-A levels in the ischemic brain tissue were also elevated in a timedependent manner, starting between 2-6 h, peaking around 24-48 h, and returning towards base-line at 72 h post MCAo . This time-dependent increase in cerebral fetuin-A levels parallels with the transient elevation of the blood-brain barrier (BBB) permeability (Belayev et al. 1996), suggesting that circulating fetuin-A can gain entry across the BBB into the ischemic brain tissue. This possibility was supported by the observation that peripherally (intravenously) administered FITC-labeled fetuin-A was found in the ischemic brain region at 24 h after MCAo ).

Reduction of circulating Fetuin-A levels during infection
In animal models of endotoxemia and sepsis (induced by cecal ligation and puncture, CLP), circulating fetuin-A levels were decreased in a time-dependent fashion, starting between 2-6 h, reaching a nadir (with maximal reduction by 50-60%) around 24-48 h. Afterwards, fetuin-A levels started to increase, returning towards basal levels approximately 72 h post endotoxemia or sepsis, supporting fetuin-A as a negative APP in animal models of lethal endotoxemia and sepsis (Li et al. 2011a). Interestingly, disruption of expression of early proinflammatory cytokines (such as IFN-) impaired bacterial endotoxin-mediated down-regulation of fetuin-A expression (Li et al. 2011a). It thus appears that early proinflammatory cytokines (such as TNF and IFN-) function as negative regulators to reduce circulating fetuin-A levels during an early stage of endotoxemia or sepsis; whereas late-acting proinflammatory mediators (e.g., HMGB1) stimulate fetuin-A expression to restore its circulating levels at a late stage. In patients with other inflammatory diseases such as pancreatitis (Kusnierz-Cabala et al. 2010), chronic kidney diseases (Metry et al. 2008), and rheumatoid arthritis (Sato et al. 2007), serum fetuin A levels were also decreased by 20-30%. In these patients, circulating fetuin-A levels were not only inversely correlated with levels of inflammatory cytokines (such as IL-6) (Kusnierz-Cabala et al. 2010), but also associated with increased mortality rates (Metry et al. 2008). Collectively, these observations classify fetuin-A as a negative APP during infection or other inflammatory illness.

Biological functions of Fetuin-A
Despite its abundance, the functions of fetuin-A remain poorly understood. A wide range of biological functions have been proposed for fetuin-A based on its structural similarities to other proteins or physical interactions with biogenic molecules.

Inhibitor of insulin or TGF-β Signalling
Fetuin-A shares sequence similarity to type II TGF-receptors (Demetriou et al. 1996) and insulin receptor tyrosine kinases (Mathews et al. 1997;Haasemann et al. 1991), and has thus been proposed as an inhibitor of the TGF-or insulin signaling pathways. After binding to to TGF-1, fetuin-A prevents TGF-1 from binding to its receptors, thereby antagonizing TGF-1-mediated antiproliferative effects (Demetriou et al. 1996). Similarly, fetuin-A can also bind to the insulin receptor, and consequently inactivate (rather than activate, as in the case for insulin) the receptor tyrosine kinase (Goustin & Abou-Samra 2010). This may partly explain why higher fetuin-A levels were associated with insulin resistance in some patients with type 2 diabetes (Ix et al. 2008).

Inhibition of pathological calcification
As a glycoprotein, fetuin-A carries two N-linked and three O-linked oligosaccharide chains that terminate with sialic acid residues, and can bind cationic Ca 2+ ions. Accordingly, fetuin-A has been proposed as an endogenous inhibitor of pathological mineralization or calcification in soft tissues (Jahnen-Dechent et al. 2001;Schinke et al. 1996;Szweras et al. 2002;Schafer et al. 2003;Ketteler et al. 2003). Specifically, fetuin-A forms protein-mineral colloids with calcium and phosphate Wu et al. 2009), thereby preventing uncontrolled mineralization that may otherwise occur under pathological conditions (Rochette et al. 2009).

Inhibition of inflammation
While investigating the mechanism underlying a cationic molecule spermine-mediated antiinflammatory actions, we serendipitously discovered that macrophages lost their responsiveness to spermine when cultured under low serum conditions . That is, despite the addition of cytokine-suppressing concentrations of spermine, the bacterial lipopolysaccharide (LPS)-induced production of TNF by these serum-starved macrophages was uninhibited. Subsequently, we discovered that these serum-starved macrophages became deprived of fetuin-A that was required for spermine to inhibit TNF production . The involvement of fetuin-A in spermine-mediated immunosuppression was confirmed by adding highly purified fetuin-A or fetuin-specific antibodies, which respectively restored or impaired spermine-mediated TNF inhibition . It is plausible that fetuin-A functions as an opsonin for cationic spermine, and its availability to immune cells may be critical in regulating the innate immune response . Indeed, levels of fetuin-A in macrophage cultures could be altered by LPS stimulation or fetuin-A supplementation ( Figure 1A). Intriguingly, the exogenously administered fetuin-A was predominantly localized in cytoplasmic punctate structures ( Figure 1B), which co-localized with vesicles containing an autophagy marker (LC3) -possibly autophagosomes or amphisomes -in LPS-stimulated macrophages.
When given at higher concentrations (e.g., 3.5 mg/ml), crude fetuin-A (> 98%, Sigma-Aldrich) abrogated endotoxin-induced release of IL-1 and nitric oxide . Upon purification by gel filtration and ion-exchange chromatography, the highly purified intact fetuin-A could effectively inhibit IFN--or LPS-induced release of HMGB1 (Li et al. 2011a), a newly identified late mediator of lethal endotoxemia and sepsis Wang et al. 2009). However, even at the concentrations (e.g., 100 μg/ml) that abrogated LPS-induced HMGB1 release, fetuin-A only partly inhibited LPS-induced TNF secretion, suggesting fetuin-A as an effective inhibitor of HMGB1 release.

Carrageenan-induced paw edema
In an animal model of carrageenan-induced inflammation, intraperitoneal administration of fetuin-A (5 to 500 mg/kg) dose-dependently attenuated the development of paw edema (Ombrellino et al. 2001). The sialic acid moieties of fetuin-A might be required for its antiinflammatory activities. When these sialic acid residues were removed by neuraminidase, the resultant asialofetuin-A failed to potentiate the anti-inflammatory activities of spermine ) and failed to attenuate carrageenan-induced TNF production in vivo

Cerebral ischemic injury
Cerebral ischemia is frequently caused by an obstruction of a cerebral artery. Despite advances in acute and prophylactic therapies, stroke represents the leading cause of longterm disability (500,000-700,000 cases per year), and the third most common cause of death (with a mortality rate of 20-25%) in the United States.

Pathogenesis of cerebral ischemic injury
Cerebral ischemic injury consists of two stages: primary tissue damage in the ischemic core and secondary tissue injury in the surrounding penumbra. The primary injury in the ischemic core is primarily mediated by tissue ion (Ca 2+ and Na + ) overload (Taylor & Meldrum 1995) and excitotoxicity (Lee et al. 1999); whereas the secondary injury in the surrounding penumbra is partly mediated by proinflammatory cytokines (Figure 2, Feuerstein et al. 1998).

Primary early injury in the core
Within seconds to minutes after cerebral ischemia, decreased ATP production leads to failure of the Na + /K + -ATPase pump, disruption of membrane potentials, influx of sodium and calcium, and subsequent release of excitatory amino acids (such as glutamate, Figure 2). Engagement of glutamate with the ionotropic N-methyl-D-aspartate receptor (NMDA) leads to Ca 2+ influx and activation of damaging proteases (e.g., phospholipase A 2 , nitric oxide synthase, endonucleases, and calpain) that compromise the functional and structural integrity of neuronal cells within 20-60 minutes (Figure 2). Early-stage therapeutics that block ion (Na + and Ca 2+ ) channels (Taylor & Meldrum 1995) and glutamate receptors (Meldrum 1990) fail in clinical trials, partly because of the impracticalities of administering such drugs at a time when those mechanisms are already activated. These failures have prompted the search for downstream targets that also mediate ischemic injury.

Secondary late injury in the penumbral zone
Outside of the ischemic core where cells are destined to die lies a penumbral zone where brain cell death continues slowly for hours and even days after the onset of ischemia ( Figure  2). This progressive expansion of cell death in the penumbra (i.e., secondary injury) is mediated by ischemia-elicited inflammatory responses. Within a few hours, microglia and neurons become activated to produce TNF and other cytokines (Kato et al. 1996;). Subsequently, polymorphonuclear cells infiltrate into the ischemic brain tissue within 12-48 hours (Akopov et al. 1996), followed by an influx of monocytes and macrophages over a period of one to several days. Together, these centrally-and peripherally-derived cells orchestrate a potentially injurious inflammatory response by overproducing various proinflammatory cytokines (Figure 2). Many pro-inflammatory cytokines (e.g., TNF and IL-1) contribute to cerebral ischemic injury (Buttini et al. 1996;Zaremba & Losy 2001), because inhibition of their production Bertorelli et al. 1998) or activity (Barone et al. 1997;Yang et al. 1999) confers protective effects. In addition, an ubiquitous nuclear protein, HMGB1, can be passively released from the ischemic core, and spilled into the surrounding periphery (Qiu et al. 2008). In the penumbra, it amplifies a potentially injurious inflammatory response by inducing various cytokines, chemokines, tissue factor and adhesion molecules (Andersson et al. 2000;Lv et al. 2009;Fiuza et al. 2003;Treutiger et al. 2003) (Figure 2). Indeed, HMGB1-specific neutralizing antibodies and antagonists (e.g., the A box) have been proven protective (Liu et al. 2007a;Muhammad et al. 2008), supporting a pathogenic role for HMGB1 in ischemic injury.
During cerebral ischemia, brain spermine levels are decreased , owing largely to an accompanying increase in the enzymatic activity of brain polyamine oxidase (Ivanova et al. 1998). The loss of spermine consequently tilts the balance towards neurotoxicity through activating the NMDA receptor, and increasing susceptibility to oxidative stress as well as excessive inflammatory response. Fig. 3. Divergent roles of spermine in cerebral ischemic injury.

Peripheral administration of fetuin-A reduced cerebral ischemic injury
As mentioned earlier, when given peripherally, exogenous fetuin-A gains entry across the BBB into the ischemic brain tissue (Figure 4). The time-course of fetuin-A extravasation in the ischemic brain tissue parallels with the time-dependent alteration of the BBB permeability (Belayev et al. 1996), which was transiently elevated (5 -25 h post MCAo) followed by a return towards baseline at 72 h post MCAo (Belayev et al. 1996). It is possible that the temporal breakdown of the BBB is required for circulating fetuin-A to transiently gain entry into the brain. Consistently, peripheral administration of fetuin-A (50 mg/kg) promoted a dose-dependent protection against cerebral ischemic injury during an early stage of cerebral ischemia (i.e., 24 h post MCAo) . However, the fetuin-Amediated protection was not long-lasting, and gradually diminished at a later stage (e.g., 7 days post MCAo). It is possible that the restore of BBB function at a late stage (3 days after MCAo) limits subsequent fetuin-A extravasation, thereby diminishing fetuin-A-mediated long-lasting protective effects. Given the aforementioned pathogenic roles of Ca 2+ and spermine in cerebral ischemia (in section 4. 2.1 and 4.2.2), as well as the capacity of fetuin-A in binding Ca 2+ and spermine (in secton 3.2 and 3.3) (Suzuki et al. 1994;Wang et al. 1997), it is plausible that fetuin-A confers protection by caging these toxic cationic molecules (Lee et al. 1999;Ivanova et al. 1998), thereby depriving them from damaging enzymes (such as Ca 2+ -dependent proteases and polyamine oxidase). Furthermore, the fetuin-A-mediated protection is associated with a reduction of ischemia-elicited HMGB1 leakage from the ischemic core, and an inhibition of

Experimental sepsis
Sepsis is the most common cause of death in intensive care units, claiming approximately 225,000 victims annually in the U.S. alone. The high mortality of sepsis is in part mediated by bacterial endotoxin , which activates macrophages and monocytes to sequentially release early (e.g., TNF and IL-1) (Dinarello 1996) and late (e.g., HMGB1) proinflammatory cytokines.

Pathogenesis of sepsis
The pathogenesis of sepsis is partly attributable to dysregulated systemic inflammatory responses that are initiated by early proinflammatory cytokines and sustained by late-acting proinflammatory mediators. For instance, excessive accumulation of early proinflammatory cytokines, including TNF (Tracey et al. 1987), interleukin (IL)-1 (Dinarello & Thompson 1991), interferon (IFN)- (Heinzel 1990), individually or in combination, contribute to the pathogenesis of lethal systemic inflammation. Because these early cytokines are difficult to target in clinical settings, we searched for other late proinflammatory mediators that may offer a wider therapeutic window. As aforementioned, HMGB1 is released from activated innate immune cells in response to microbial products (such as endotoxin or CpG-DNA) Ivanov et al. 2007), or host cytokines (e.g., TNF or IFN-) Rendon-Mitchell et al. 2003), and functions as a late mediator of endotoxemia and sepsis Yang et al. 2004;Wang et al. 2008;Wang et al. 2009). In murine models of endotoxemia and sepsis, HMGB1 is first detectable in the circulation eight hours after the onset of diseases, subsequently increasing to plateau levels from 16 to 32 hours Yang et al. 2004) (Figure 4). This late appearance of circulating HMGB1 parallels with the onset of animal lethality from endotoxemia or sepsis, and distinguishes itself from TNF and other early proinflammatory cytokines . Therefore, agents capable of selectively attenuating systemic HMGB1 accumulation at a late stage may hold potential in the treatment of lethal sepsis.

Dual roles of spermine in experimental sepsis
In light of the anti-inflammatory activities of spermine in vitro Zhu et al. 2009), we evaluated the effects of spermine on animal survival in animal models of sepsis. Intraperitoneal administration of spermine (1.0 -10 mg/kg, twice daily, for three days) did not protect mice against lethal endotoxemia, but confers a dose-dependent protection against lethal sepsis. This protection was associated with a significant attenuation of systemic accumulation of HMGB1 and other cytokines (e.g., IL-6, KC, MCP-1, MIP-2, TIMP-1, sTNFRI and sTNFRII) (Zhu et al. 2009). At a higher dose (100 mg/kg), however, spermine decreased animal survival rate from 58% to 38% at 48 h post CLP, and further decreasing it to 0% at 72 h post CLP. It is possible that spermine is enzymatically converted by polyamine oxidases into cytotoxic metabolites (e.g., 3-aminopropanal), thereby exerting these potentially toxic effects when given at higher doses.

Protective role of Fetuin-A in sepsis
To understand the role of fetuin-A in systemic inflammatory diseases, we determined the influence of fetuin-A disruption on endotoxemic and septic lethality. Although fetuin-Adeficient C57BL/6J mice were not more susceptible to cerebral ischemic insult than sex-and body-matched (male, 27-29 g) wild-type C57BL/6J mice , they were more susceptible to lethal endotoxemic or septic insult (Li et al. 2011a). It suggests that endogenous fetuin-A occupies an integral role in host defense against lethal systemic inflammation.
The protective role of fetuin-A was further supported by the observations that supplementation with exogenous fetuin-A (20-100 mg/kg) provided a dose-dependent protection against lethal endotoxemia (Li et al. 2011a). In an animal model of sepsis, delayed administration of fetuin-A (20 -100 mg/kg), beginning 24 h after the onset of sepsis and followed by an additional dose at 48 h post CLP, dose-dependently and significantly increased long-term animal survival rates from 45% to 90% (Li et al. 2011a).

Protective mechanisms
Supplementation of fetuin-A was associated with significant reduction of circulating HMGB1 levels, suggesting that fetuin-A confers protection by inhibiting late-acting proinflammatory mediators (Li et al. 2011a). The mechanisms underlying fetuin-A-mediated suppression of HMGB1 release may be complex. At the concentrations (100 μg/ml) that fetuin-A attenuated LPS-induced HMGB1 release in macrophage cultures, fetuin-A stimulated autophagy and impaired LPS-induced elevation of cytoplasmic and nuclear HMGB1 levels (Li et al. 2011a). It is presently unknown whether fetuin-A reduces cytoplasmic HMGB1 levels by stimulating its degradation in an autophagy-dependent fashion, as what has been shown for other HMGB1 inhibitors such as EGCG, the major catechin of Green tea (Camellia sinensis) (Li et al. 2011b). Accumulating evidence has suggested the possibility that fetuin-A functions as a negative regulator of HMGB1 release during lethal systemic inflammation (Figure 4). First, the timedependent decrease of circulating fetuin-A levels was accompanied by parallel but opposite changes -a time-dependent increase -of circulating HMGB1 levels in animal models of endotoxemia  or sepsis . Second, disruption of fetuin-A expression led to elevation of serum HMGB1 levels in endotoxemia and sepsis (Li et al. 2011a). Lastly, supplementation of fetuin-A resulted in significant reduction of circulating HMGB1 levels during endotoxemia and sepsis (Li et al. 2011a). Nevertheless, the current study can not exclude other alternative mechanisms by which fetuin-A confers these protective effects. For instance, fetuin-A may be capable of binding bacteria (Chmiela et al. 1997;Dubreuil et al. 2002), thereby affecting macrophage-mediated pathogen elimination. Furthermore, fetuin-A may facilitate macrophages-mediated ingestion and elimination of apoptotic neutrophils (Lord 2003;Jersmann et al. 2003), thereby preventing secondary necrosis and passive leakage of injurious molecules (e.g., proteases, reactive oxygen species, and HMGB1) (Bell et al. 2006).

Conclusions
A liver-derived acute phase protein, fetuin-A, appears to be distinctly regulated by different proinflammatory mediators. A previously under-appreciated protective role for fetuin-A in injury and infection has been suggested by recent studies. Fetuin-A is capable of crossing the blood-brain barrier, inhibiting early inflammatory response in animal models of cerebral ischemia, thereby conferring a short-term neuroprotection against ischemic injury. Disruption of fetuin-A expression renders mice significantly more susceptible to lethal endotoxemia or sepsis; whereas repetitive administration of fetuin-A confers a dosedependent and long-lasting protection in animal models of lethal endotoxemia and sepsis. Thus, fetuin-A occupies protective roles against injury-or infection-elicited inflammation.

Acknowledgment
We thank Arvin Jundoria for critical reading of the manuscript. Work in the author's laboratory was supported by the National Institute of General Medical Sciences (R01GM063075 and R01GM070817 to HW) and the National Center of Complementary & Alternative Medicine (R01AT05076 to HW).