Microbial Recognition and Danger Signals in Sepsis and Trauma

Abstract

Early host recognition of microbial invasion or damaged host tissues provides an effective warning system by which protective immune and inflammatory processes are initiated. Host tissues responsible for continuous sampling of their local environment employ cell surface and cytosolic pattern recognition receptors (PRRs) that provide redundant and overlapping identification of both microbial and host alarmins. Microbial products containing pathogen-associated molecular patterns (PAMPs), as well as damage-associated molecular patterns (DAMPs) serve as principle ligands for recognition by these PRRs. It is this interaction which plays both an essential survival role in response to infection and injury, as well as the pathologic role in tissue and organ injury associated with severe sepsis and trauma. Elucidating the interaction between ligands and their respective PRRs can provide both a better understanding of the host response, as well as a rational basis for therapeutic intervention.

1. Introduction

1.1 Epidemiology

Despite advances in healthcare, sepsis and trauma remain major causes of morbidity, mortality and healthcare costs in the United States. In 2007, severe sepsis accounted for greater than 700,000 hospitalizations, 200,000 deaths and $24 billion dollars in hospital expenses in the United States [1]. Likewise, trauma accounted for greater than 2.5 million hospitalizations, 180,000 deaths and $149 billion dollars in hospital costs in the United States in 2010 [2]. Thankfully, with improvements in early detection of sepsis and better approaches to the immediate resuscitation of trauma and sepsis patients, multi-organ failure and early mortality are declining [3–7]. Long-term mortality, and equally important, long-term functional, cognitive and physiologic deficits have been much more resistant to improvements in care [8].

The early host response to both severe trauma and sepsis results from the activation of innate immunity and inflammatory responses, as well as simultaneous modulation of adaptive immunity. Exposure to microbial pathogens, hemorrhagic shock, tissue injury, cell necrosis and reperfusion injury all elicit a highly reproducible host protective response that is aimed at reducing microbial proliferation and secondary infections through damaged tissue barriers [9]. Simultaneously, down-regulation of antigen presentation capabilities hinders an auto-immune response to the release of cellular constituents not normally recognized by the host. Taken in its entirety, this multicomponent response is beneficial, but when exaggerated or prolonged, the beneficial effects are outweighed by potentially deleterious effects. These include endothelial and epithelial cell injury, a hypercoagulable state, microvascular perfusion defects and organ injury when the inflammatory response is excessive, and persistent immune suppression, chronic inflammation and protein catabolism when the response is prolonged [10]. Thus, understanding the early host protective response to microbial infection and tissue injury is essential to improving outcomes without enhancing adverse effects associated with an inappropriate exaggerated or prolonged response.

1.2 Historical Background

In 1913, Osler first suggested that in most disease, death resulted from a host response rather than the underlying pathogen [11]. Subsequently, Lewis Thomas warned against the ‘angry macrophage’ which could be sufficiently destructive to produce organ injury and death on its own [12]. The concept that an unwelcome innate immune response could lead to tissue injury has evolved over the past fifty years. In 1949, Burnet introduced the concept of ‘self versus non-self’ in which he proposed non-self or foreign elements initiated an immune reaction whereas ‘self’ elements were tolerated [13]. Moving forward to 1989, Janeway proposed that the innate immune system had evolved to differentiate ‘self from infectious non-self’ [14]. He suggested that discrimination occurred via pattern recognition receptors (PRRs) on lymphocytes and antigen-presenting cells. Janeway proposed that these PRRs recognize molecules on the surfaces of microorganisms, such as lipopolysaccharide (LPS) from gram negative bacteria. From this concept, the term pathogen-associated molecular patterns (PAMPs) evolved. Since that time, the identification of other microbial PAMPs has significantly increased, now including not only lipoproteins, glycoproteins and polysaccharides from bacteria and fungi, but also nucleic acids from bacteria and viruses.

In 1994, Polly Matzinger expanded on the work of Janeway and others with the introduction of the ‘danger hypothesis’ [15], which states that immune system focuses on endogenous and exogenous danger signals rather than differentiating ‘self versus nonself.’ Microbial products are obvious danger signals, but the definition could also be now expanded to include host products which are themselves danger signals. These latter compounds have been designated damage-associated molecular patterns (DAMPs) and represent endogenous alarmins that can similarly activate innate immune and inflammatory processes. Many of these DAMPs are products of damaged tissue, such as mitochondrial and nuclear DNA and advanced glycosylated end products (AGEs) [16–18]. Others include compounds actively released from cells as danger signals, such as high mobility group box 1 (HMGB1) and S100 proteins [19–21]. The importance of the ‘danger hypothesis’ is that it ties together the similarities in the early inflammatory host response to both microbial pathogens and to conditions that produce tissue damage but are sterile, such as blunt trauma, hemorrhagic shock and surgical injury.

2. Recognition, Signaling and Cellular Response

2.1 Recognition

Epithelial, endothelial and immune cells are continuously sampling their environment and are capable of recognizing PAMPs and DAMPs via pattern recognition receptors (Table 1). The identities of these receptors are increasing, but include: toll-like receptors (TLRs); C-type lectin receptors (CLRs); nucleotide-binding oligomerization domain (Nod) like receptors (NLRs); retinoic-acid-inducible gene I (RIG-I) like receptors (RLRs); C-type lectin receptors (CLRs); and receptor for advanced glycation end products (RAGE). PRRs can be classified based not only on their ligands, but also on their location, i.e. cell surface or cytoplasmic. In addition to membrane-associated and intracellular receptors, there are cellular proteins of the innate immune system that circulate and also serve as pattern recognition molecules. A common example of these circulating proteins is complement but also includes the bactericidal/permeability-increasing protein (BPI) [22].

Table 1. A listing of commonly studied PRRs and their associated PAMPs and DAMPs.

Microbial products and endogenous danger signals are recognized by a conserved and overlapping group of pattern recognition receptors leading to a common host response.

PRRs	Location	PAMPs	DAMPs
TLR1	Plasma membrane	Lipoproteins
TLR2	Plasma membrane	Lipoproteins	HMGB1
		Peptidoglycan	Hsp70
		Lipoteichoic acid	S100
TLR3	Endosome	Viral dsRNA
TLR4	Plasma membrane	Lipopolysaccharide	HMGB1
			Hsp70
			S100
			Hyaluronic acid
TLR5	Plasma membrane	Flagellin
TLR6	Plasma membrane	Lipoproteins
TLR7	Endosome	Viral ssRNA
TLR8	Endosome	Viral ssRNA
TLR9	Endosome	Pathogen DNA	mtDNA
			nDNA
Dectin	Plasma membrane	Beta-glucans
Mincle	Plasma membrane		SAP130
NOD1	Cytoplasm	Peptidoglycan
NOD2	Cytoplasm	Peptidoglycan
NALP3	Cytoplasm	Viral dsDNA	nDNA
		Viral ssRNA	ATP
			Uric acid
RIG-1	Cytoplasm	Viral dsRNA
MDA5	Cytoplasm	Viral dsRNA
RAGE	Plasma membrane		AGEs
			HMGB1
			S100

Toll-like receptors (TLRs) are among the most studied PRRs and were first identified in mammals in the 1990s [23]. Thirteen mammalian TLRs have been identified, including ten receptors in humans. Each TLR recognizes a specific set of molecular patterns, but there is considerable pattern recognition overlap among these receptors. TLR1, TLR2, TLR4, TLR5, TLR6 and TLR10 are all expressed on the extracellular cell membrane and sample the extracellular environment. TLRs located on the cell membrane largely detect external microbial components and circulating damage signals. For example, TLR4 recognizes lipopolysaccharide (LPS) from gram negative bacteria, structural proteins from viruses, mannan (a cell wall polysaccharide) from fungi, glycoinositolphospholipids from Trypanosoma as well as endogenous high mobility group box (HMGB) nuclear proteins from distressed cells [24, 25]. Alternatively, TLR3, TLR7, TLR8 and TLR9 are predominantly expressed within intracellular vesicles of innate immune cells and probe for intracellular signals of infection or cellular stress. TLRs expressed within the intracellular vesicles mainly detect free nucleic acids, distinguishing endogenous from microbial. For example, TLR3 and TLR7 are necessary for recognition of double stranded and single stranded RNA viruses, respectively [26–28]. Likewise, viral and bacterial DNA activate TLR9 and lead to production of inflammatory cytokines, especially in plasmacytoid dendritic cells [29–31].

C-type lectin receptors (CLRs) are transmembrane proteins expressed on the surface of antigen presenting cells, and these receptors interact with several bacterial, fungal and viral carbohydrates. Yamasaki and colleagues demonstrated that other membranes of the CLR family can recognize DAMPs [32]. Specifically, macrophage-inducible, calcium-dependent lectin (Mincle) recognizes spliceosome-associated protein 130 (SAP130), a nuclear protein that is released from necrotic and dying cells, and leads to the production of inflammatory cytokines and chemokines [32].

Nod-like receptors (NLRs) are cytoplasmic receptors capable of recognizing PAMPs in the cytosol such as microbial toxins, bacterial peptidoglycans and viral RNA [33]. Likewise, these receptors can respond to DAMPs such as host cell-free nuclear DNA [34]. These pathogen and damage signals lead to the formation of large protein complexes in the cytosol, termed inflammasomes. Inflammasomes play an important role in the promotion and amplification of the inflammatory response through the activation of caspase-1 and ultimate production of interleukin-1 beta (IL-1β) and interleukin-18 (IL-18) [33]. In addition, inflammasome mediated caspase-1 activity leads to pyroptosis, an inflammatory form of cell death [35].

RIG-I-like receptors (RLRs) are cytoplasmic receptors and respond primarily to the presence of RNA viruses. Three mammalian RLRs have been identified: retinoic-acid-inducible gene I (RIG-1); melanoma differentiation-associated gene 5 (MDA5); and laboratory of genetics and physiology gene 2 (LGP2). RIG-I and MDA5 recognize viral dsRNA. RIG-I leads to increased type I interferon (IFN) production in response to the Rhabdovirus, Paramyxovirus, Orthomyxovirus, Flavivirus, Filovirus and Reovirus [36]. It is suggested that the Reovirus, Flavivirus and Picornavirus can also trigger MDA5, leading to the activation of the antiviral innate immune response independent of RIG-I. LGP2 is believed to act as a negative feedback regulatory of antiviral immune response by inhibiting RIG signaling and by sequestering viral dsRNA [37, 38].

Finally, RAGE is a protein expressed on human endothelial cells, monocytes and lymphocytes, and serves as a receptor for the products of nonenzymatic glycation and oxidation of proteins/lipids (AGEs), HMGB1 and S100 proteins [39–41]. AGEs engagement with RAGE on endothelial cells promote generation of reactive oxygen species (ROS) [41]. HMGB1-RAGE interactions play a vital role in monocyte migration [40]. Engagement of S100 protein with RAGE triggers cellular activation and generation of tumor necrosis factor alpha (TNFα) [39].

2.2 Signaling Pathways

Stimulation of PRRs by molecular patterns activates complex, downstream signaling events resulting in the recruitment and phosphorylation of intracellular intermediates and ultimately, activation of immediate response genes (Figure 1). Signaling by TLRs primarily involves five adaptor proteins: myeloid differentiation primary response 88 (MyD88); toll-interleukin 1 receptor domain containing adaptor protein (TIRAP); toll-interleukin 1 receptor domain containing adaptor protein inducing interferon beta (TRIF); TRIF-related adaptor molecule (TRAM); and sterile-alpha and armadillo motif containing protein (SARM) [42]. In general, TLR signaling can be classified as being either MyD88- or TRIF-dependent. TLR1, TLR2 and TLR5–9 utilize MyD88-dependent pathways exclusively, while TLR3 utilizes only TRIF-dependent pathway. TLR4 utilizes both MyD88- and TRIF-dependent pathways, depending upon the status of receptor interacting protein [43, 44]. TIRAP functions as a sorting adaptor to bridge MyD88 to TLR1, TLR2, TLR6 and TLR4 [45]. Similarly, TRAM functions to recruit TRIF to TLR4 [46]. The adapter protein SARM acts as a negative regulator of both MyD88 and TRIF activation [42].

Figure 1. TLR receptors and intracellular signaling pathways for the recognition of microbial products and endogenous danger signals.

TLR1, 2, 4, 5 and 6 are located on the transmembrane whereas TLR 3, 7, 8 and 9 are located in endosomal vesicles. TLR stimulation allows binding with MyD88, TIRAP, TRAM and TRIF adapter proteins. Downstream signaling occurs via IRAK, TRAF3, TRAF6, TAK1, TBK1 and IKK complexes. Activation of MAPK, NFKB and IRF results in the induction of inflammatory cytokines and type I IFN.

Upon TLR engagement and adapter protein recruitment, downstream signaling occurs via IRAK, TRAF3, TRAF6, TAK1, TBK1 and IKK complexes. Ultimately, the activation of mitogen-activated protein kinase (MAPK), nuclear factor- κB (NF-κB) and interferon regulatory factor 3/7 (IRF3/7) leads to transcriptional changes and the secretion of inflammatory cytokines, type I IFN, chemokines and antimicrobial peptides which together function to target and kill pathogens.

2.3 Cellular Response

Pattern recognition receptor stimulation and downstream signaling events lead to both common (non-specific) and pathogen-specific host cellular responses. Early responses tend to be non-specific, focused on early inflammatory pathways aimed at suppressing microbial replication, tissue invasion and dissemination from the site of infection. The near immediate release of cytokines and chemokines elicit an endothelial cell-target hypercoagulable state meant to reduce blood loss and trap microbial pathogens. Coagulopathy is likely driven by multiple pathways, and can be characterized by an early hypercoagulable state followed in some patients, especially those with adverse outcomes, by a consumptive coagulopathy and uncontrolled bleeding [47]. Under normal conditions of health, the vascular endothelium has a net anticoagulant state due to the constitutive expression of thrombomodulin, protein C and heparin [48]. Simultaneously, expression of selectins and adhesion molecules on endothelial cells and leukocytes is suppressed in healthy states. This pattern changes dramatically in the presence of microbial infection or trauma driven by complement activation and PAMP/DAMP mediated signaling. For example, platelet activating factor triggers activation and accumulation of platelets at local sites of injury as well as the synthesis of cytokines and activation of innate immune cells [49]. Likewise, TNFα and other inflammatory cytokines lead to a rapid activation of the common pathway of the coagulation system, upregulation of adhesion molecules, PAI-1 and Tissue Factor, and down regulation of thrombomodulin and protein C [48, 50]. The resultant hypercoagulable state reduces microvascular perfusion while the increased endothelial permeability permits the diapedesis and the recruitment of inflammatory cells out of the vascular compartment and to the site of infection or injury. These cells not only serve to phagocytose pathogens and damaged host tissues but release reactive nitrogen and oxygen species, and proteases, creating a local milieu lethal to microbes but isolated from healthy tissues. Neutrophil extracellular traps (NETs) ensure a high local concentration of antimicrobial components that bind, disarm and kill microbes independent of phagocytic uptake [51]. In addition to their antimicrobial properties, NETs contribute to the physical barrier that prevents further spread of the pathogens. Furthermore, delivering the granule proteins into NETs reduces healthy tissue damage by preventing potentially injurious proteins such as proteases from diffusing away from the site of inflammation.

Simultaneously, antigen presenting cells either present or recruited to the site are processing microbial antigens, altering host tissue degradation products and delivering secondary signals at both the site of infection and in draining lymph nodes. The secondary release of cytokines, chemokines and lipid mediators all modulate the recruitment of additional inflammatory cells and serve to polarize the inflammatory milieu appropriate for recruitment and expansion of adaptive immunity initiators. Interestingly, much of the early response to either microbial invasion or sterile tissue death associated with trauma or hypoxia are more similar than disparate [9]. Although much of the latter innate and adaptive immune responses are ultimately pathogen- or damaged tissue-specific, the immediate or early responses driven by engagement of PRRs induce a similar set of response elements via activation of common, conserved early response pathways. In human ex vivo studies, macrophages induce a shared pattern of genes in response to bacterial components (e.g. LPS, lipoteichoic acid) and endogenous alarmins (e.g. Hsp65, Hsp70) [52]. These shared genes include inflammatory cytokines, chemokines and adhesion molecules [52].

3. PAMPs

3.1 Background

Pathogen recognition is the first step in mounting an immediate or early innate immune response. As such, PAMPs from bacteria, fungi and viruses play an essential role in the understanding and management of infection. Commonly studied PAMPs include lipopolysaccharide, lipoproteins, peptidoglycans, lipoteichoic acid and nucleic acids.

3.1 Lipopolysaccharide

Gram negative bacteria possess an outer membrane containing large quantities of lipopolysaccharide (LPS) and phospholipids. These highly charged molecules bind large numbers of proteins and nucleic acids. LPS has been investigated for many years, and much of what we have learned about pathogen-recognition comes from the investigations of LPS. Importantly, much of our understanding of the pathogenesis of septic shock is derived from studies utilizing LPS. Septic shock is essentially an inappropriate, exaggerated early inflammatory response to microbial invasion and can be replicated in its entirety by binding of LPS to the TLR4 receptor complex comprised of CD14, TLR4 and MD2 [53, 54]. Mice deficient in either CD14, TLR4 or MD2, all components of the TLR4 receptor complex, are protected from LPS mediated septic shock [53, 55–58]. Although LPS can replicate in its entirety the pathological changes associated with septic shock, LPS alone does not explain the pathogenicity of gram negative bacteria. For example, mice deficient in TLR4 or MD2 are not resistant to the lethal effects of live or heat-killed gram negative bacteria [56, 57], demonstrating that living or heat-killed bacteria contain additional PAMPs, other than LPS recognized by PRRs. Although sepsis is a complex phenomenon that is difficult to entirely recapitulate in animal studies [59], research involving LPS has played a vital role in our understanding of the pathology of the syndrome.

Although TLR4 signaling occurs through both MyD88- and TRIF-dependent pathways, the inflammatory host response to LPS is regulated primarily through TLR4 signaling via MyD88-dependent pathways [55, 60]. This was illustrated when macrophages from TLR4 ^−/− mice as well as MyD88^−/− mice fail to produce significant levels of inflammatory cytokines in response to Escherichia coli LPS [55, 60]. When compared to wild-type mice, MyD88^−/− mice are resistance to LPS-induced shock, similar to TLR4 and CD14 null animals [60]. TRIF signaling from the TLR4 receptor complex, however, plays an essential role in the antimicrobial responses to bacterial and viral infections. Mediated primarily by type I interferon signaling, TRIF signaling via the TLR4 complex or TLR3 dimerization with dsRNA activates type I interferon-dependent antimicrobial properties in T and B cells, NK cells and myeloid cell populations.

There are a number of individuals with impaired TLR4 signaling. Children with congenital MyD88 deficiencies are known to have increased susceptibility to gram positive infections, especially Staphylococcus [61]. Subjects with IRAK4 deficiencies, another signaling intermediate of TLR4 complex and MyD88, have recurring respiratory infections with both gram negative and positive bacteria; however, the incidence of viral and fungal infections are not increased [61]. In contrast, subjects with congenital TRIF deficiencies have a much higher prevalence of viral infections, many lethal, particularly Herpes simplex encephalopathy [62].

There is a strong developmental component to TLR4 signaling. For example, neonates have reduced numbers of TLR4 receptors, impaired TLR signaling in response to LPS [63] and impaired LPS-induced cytokine production [64]. LPS stimulated polymorphonuclear leukocytes (PMNs) from healthy adults readily form neutrophil extracellular traps (NETs) which bind and kill bacteria extracellularly [51, 65]. In contrast, PMNs isolated from preterm and term neonates fail to form NETs following LPS stimulation and have deficient extracellular bacterial killing compared to PMNs isolated from adults [65].

3.2 Lipoproteins

Lipoproteins are a component of all bacteria and share a common chemical motif [66]. Lipoproteins play a key role in pathogen recognition and induction of host innate immunity against bacterial infections. Lipoproteins induce inflammatory responses via TLR2 signaling [66]; however, the ligation of lipoproteins with TLR2 is complex. Lipoproteins can bind to TLR2 homodimers, as well as TLR1-TLR2, TLR6-TLR2 and CD14-TLR2 heterodimers. Interestingly, the signaling pathways and cellular responses to ligation of the TLR2 homo- and heterodimers are similar, so the diversity in TLR2 complexes serves to primarily expand the recognition of diverse lipoproteins. Signaling via all of these complexes is primarily MyD88-dependent. Staphylococcus aureus variants lacking lipoproteins escape recognition by the immune system and can cause lethal infections with disseminated abscess formation compared to wild-type S. aureus containing lipoprotein [67]. Pretreatment with bacterial lipoproteins protects wild-type mice from polymicrobial sepsis-induced lethality via increased bacterial recognition and bactericidal activity [68].

3.3 Peptidoglycan

Peptidoglycan (PGN) is a unique component of the cell wall of virtually all bacteria, but especially gram positive bacteria, and is not present in eukaryotes. As such, PGN is an excellent target for recognition by the innate immune system. Indeed, higher eukaryotes, including humans, have several PGN recognition molecules, including CD14, TLR2, nucleotide oligomerization domain (Nod)-containing protein-1 and −2, a family of peptidoglycan recognition proteins (PGRPs) and PGN-lytic enzymes (lysozyme and amidase). Signaling via the Nod receptors bypasses MyD88 and induces an inflammatory response primarily through RIP2 (RICK) signaling of NF-κB and MAP kinases. In addition, peptidoglycan induces inflammasome assembly and caspase-1 activity [69].

3.4 Lipoteichoic Acid

Lipoteichoic acid (LTA) is also located within the cell wall of most gram positive bacteria and is an important virulence factor in gram positive infections. LTA is bound to the cell membrane through diacylglycerol and projects through the peptidoglycan layer of most gram positive bacteria. LTA is released following bacteriolysis induced by lysozyme and cationic peptides from leukocytes [70]. LTA induces NF-κB, MAP kinase and phosphoinositide 3-kinase activation via TLR2-CD14 and TLR2 homodimer complexes [71].

3.5 Nucleic Acids

Pathogen-derived nucleic acids are an important group of PAMPs. Most are recognized intracellularly during either bacterial phagocytosis, viral replication or reverse transcriptase. Distinguishing host DNA or RNA from either intracellular bacterial or viral DNA/RNA is a key role in determining an infectious state; fortunately, bacterial DNA is easily distinguished from eukaryotic DNA by their differences in methylation patterns. Bacterial DNA is principally methylated on the adenine nucleotide, whereas eukaryotic DNA is primarily methylated on the cytosine nucleotide. The absence of methylation of CpG DNA fragments are recognized as microbial and triggers a potent inflammatory response via TLR9 [29] and is important for single stranded DNA (ssDNA) and double stranded (dsDNA) viral infections. Although epigenetic DNA from viruses is generally not methylated, once integrated into the host genome, methylation of viral DNA is similar to the endogenous eukaryotic DNA. Thus, infections by DNA-containing viruses are generally detected early by differences in their methylation pattern prior to host integration. Once integrated into host DNA, the viral DNA becomes hidden from host immune responses until viral reactivation.

In vivo experiments have demonstrated that macrophages from TLR9^−/− mice have impaired activation of NF-κB and fail to produce detectable levels of inflammatory cytokines in response to exposure to bacterial DNA [29]. This failed inflammatory response protects TLR9^−/− mice from bacterial DNA-induced shock syndrome compared to wild-type mice. Viral DNA is recognized by TLR9 and the inflammasome [30, 31] [72]. TLR7 and TLR8 recognize viral ssRNA [27, 28, 73], while TLR3 and RIG-I-like receptors recognize viral dsRNA [26, 36].

4. DAMPs

4.1 Background

Cell damage in sepsis and trauma results in a release of damage-associated molecular patterns (DAMPs) locally and into the systemic circulation. DAMPs reflect products of injured or dying tissues as well as compounds actively released in response to cellular stress (Figure 2). HMGB nuclear proteins, heat shock proteins, S100 proteins, hyaluronic acid degradation products and host DNA fragments are some of the most well studied DAMPs in sepsis and traumatic injury.

Figure 2. Mechanism of endogenous danger signal release.

In response to tissue damage and ischemia, cells actively release HMGB1, Hsps and S100 proteins. Necrotic cells also passively release HMGB1, Hsps proteins, S100 proteins, mitochondrial DNA and nucleic DNA. In contrast, apoptotic cells do not promote inflammation.

4.2 High Mobility Group Box

4.2.1 Role of HMGB

High mobility group box (HMGB) nuclear proteins are normally found in the nucleus of eukaryotic cells [74]. Nuclear HMGB protein is involved in histone formation and gene regulation, but it is unique in that under cellular stress, it is actively exported to the cytoplasm where it is acetylated. Acetylation prevents the re-entry of HMGB1 to the nucleus, assuring its ultimate release. HMGB1 protein is unique in that it is both an alarmin and a proinflammatory mediator or cytokine. HMGB1 was first shown to be produced late after a lethal endotoxin or bacterial challenge, and blocking its activity could protect against mortality [20]. HMGB1 release was shown to be not only induced by LPS and TNFα, but its production could also induce the release of TNFα. It was subsequently demonstrated that HMGB1 is released both from dying cells, but perhaps more importantly, can be actively secreted by monocytes and macrophages as part of the early inflammatory response [19, 20]. Actively secreted HMGB1 requires processing via the inflammasome, as does the release associated with pyroptosis and cell death [75]. This form of HMGB1 is often hyperacetylated and in a reduced form. In contrast, the release of HMGB1 from programmed cell death or apoptosis is generally in the fully oxidized form. Interestingly, HMGB1’s capacity to bind to its receptor is determined in large part by the oxidation status of its three cysteine residues. It has been reported that the fully reduced form of HMGB1 binds to SDF1 (CXCL12) and this complex is recognized by its PRRs: RAGE, TLR2 and TLR4 [76]. This difference in binding and oxidation status explains in large part why necrotic or pyroptotic cell death is proinflammatory while apoptotic cell death is anti-inflammatory [77].

4.2.2 HMGB in Sepsis

Clinical studies have demonstrated elevated HMGB1 levels in patients diagnosed with severe sepsis and septic shock [78, 79]. Furthermore, HMGB1 plasma concentration correlates with degree of organ dysfunction and mortality in critically ill septic patients [20, 78]. HMGB1 diminishes neutrophil associated bacterial killing and decreases activation of NADPH oxidase via RAGE-dependent mechanisms [80]. Therefore, it is proposed that HMGB1 potentiates sepsis associated organ dysfunction and mortality by inhibiting neutrophil mediated bacteria killing.

4.2.3 HMGB in Trauma

HMGB1 release has also been observed following severe trauma in humans [81]. Plasma levels also correlate with severity of trauma and are significantly higher in non-survivors than survivors [81]. In murine models, intratracheal administration of HMGB1 results in lung neutrophil accumulation, increased lung edema and increased lung tissue levels of IL-1β, TNFα and macrophage inflammatory protein 2 (MIP-2) [82]. Treatment of mice with anti-HMGB1 antibody in mice with hemorrhagic shock improved survival at 24 hours compared to controls [83].

4.3 Heat Shock Proteins

4.3.1 Role of Hsp

Heat shock proteins (Hsps) are a group of proteins that primarily function as molecular chaperones. However, their expression is markedly increased in response to cellular stress, including heat-shock, cold-shock, UV damage and other forms of cellular stress. In vitro studies have suggested that T and B lymphocytes continually release Hsp70 in the serum and increase release in response to stress such as increased temperature [84]. Once released, extracellular Hsp70 has been hypothesized to bind to the surface of human monocytes via the TLR4/TLR2 complex resulting in NF-κB activation and upregulation of pro-inflammatory cytokines TNFα, IL-1β and IL-6 [85].

It is debated as to whether heat shock proteins in general or Hsp70 in particular is a DAMP. Hsp70 is proposed to be a ligand for the TLR4 and TLR2 signaling [86] but that has been recently challenged. The problem rests with the highly charged nature of this chaperone protein and its ability to bind other charged compounds, including endotoxin and nucleic acids. This has raised the possibility that the alarmin properties of Hsp70 may not be due to the protein itself but the compounds that it chaperones. Contamination of Hsp70 proteins with LPS or DNA could potentially generate responses falsely attributed to Hsp70.

The role of Hsp70 in the host response appears to be complex. Recent data has suggested that Hsp70 may indeed be an alarmin, but not signaling via TLR receptors, and perhaps, not simply as a proinflammatory agent. Rather, Hsp70 may be a ligand for siglec-5 and siglec-14 [87]. Siglecs are sialic acid binding immunoglobulin (Ig)-like lectins that are membrane attached receptors that bind sialic acid rich glycans. Siglecs have either pro- and anti-inflammatory in the host. Specifically, siglec-5 is thought to be immunosuppressive while siglec-14 is immunoactivating. Thus, binding of Hsp70 to siglec-5 and siglec-14 may regulate both pro- and anti-inflammatory responses simultaneously.

4.3.2 Hsp in Sepsis

The expressions of Hsp27, Hsp60, Hsp70 and Hsp90 are significantly increased in adult patients with sepsis [88]. Likewise, plasma Hsp60 and Hsp70 levels are elevated in children with septic shock [89, 90]. In a murine model of LPS induced shock, Hsp90 inhibitor pretreated mice demonstrate a significant prolonged survival compared to control mice [91]. Hsp90 inhibitors prevent the rise in plasma levels of monocyte chemoattractant protein-1 (MCP-1) and TNFα, and attenuate capillary leak [91]. Conversely, Hsp70^−/− mice have increased mortality after cecal ligation and puncture induced sepsis compared to control mice [92]. McConnell and colleagues demonstrated increased systemic levels of TNFα, IL-6, IL-10 and IL-1β amongst aged Hsp70^−/− mice compared to aged wild type mice following cecal ligation and puncture and concluded Hsp70 prevents mortality and decreases the systemic inflammatory response in aged septic hosts [93]. This supports the anti-inflammatory role of Hsp70 in response to sepsis.

4.3.3 Hsp in Trauma

The expressions of Hsp27, Hsp70 and Hsp90 are significantly increased in patients following traumatic injury, and serum levels of Hsp70 correlate with severity of injury [94, 95]. Elevated serum levels of Hsp70 among patients with severe traumatic brain injury (TBI) are associated with a significant increase in mortality [96]. On the other hand, elevated initial serum levels of Hsp72 among patients with severe trauma are associated with a significant increase in survival [97].

4.4 S100

4.4.1 Role of S100

S100 proteins are calcium-binding homodimeric proteins that are highly abundant in the cytoplasm of phagocytes. There are 21 S100 members, and several are actively secreted in response to cell stress, while most are passively released from damaged and ischemic cells [21, 98, 99]. Upon release, the S100 family heterodimer (S100A8/A9) acts as an endogenous ligand of TLR4 and induces TNFα expression via MyD88-dependent pathway [100]. In addition, S100A12 release triggers cellular activation and cytokine generation via RAGE [39].

4.4.2 S100 in Sepsis

Patients with severe sepsis demonstrate elevated circulating levels of S100A8/A9 [101]. In murine models of LPS-induced shock and gram negative bacterial abdominal sepsis, S100A9 appears to promote lethality [100]. Conversely, S100A9 deficiency leads to enhanced bacterial dissemination, increased distant organ damage and reduced survival in Klebsiella pneumoniae pneumonia induced sepsis [102].

4.4.3 S100 in Trauma

Blood samples from human blunt trauma patients demonstrate higher levels of circulating S100A8/A9 in trauma survivors compared to non-survivors [103]. Serum S100A8/A9 levels positively correlated with levels of the chemokine IFNγ inducible protein 10 (IP-10). Based on this murine model, S100A8/A9 induced IP-10 expression is TLR4-TRIF dependent and RAGE independent.

Serum and cerebrospinal fluid (CSF) S100B levels are markedly increased after traumatic brain injury and positively correlate with mortality and poor neurological outcome [104–111]. On the contrary, in vitro experiments with rat neonatal neuronal cultures have demonstrated a protective or reparative role of S100B following traumatic brain injury [112]. When S100 antibody was given pre-injury, there was a significant increase in neuronal injury following a strain injury compared to control mice.

4.5 DNA

4.5.1 Role of DNA

In the event of cellular damage and necrosis, endogenous nuclear and mitochondrial DNA are released into the circulation. Given their evolutionary history, it has been proposed that circulating mitochondria DNA (mtDNA) are recognized by host immune cells through the same mechanisms as prokaryotic DNA, and produce a similar inflammatory response to bacterial nucleic acids. Cell-free host nuclear DNA (nDNA) have been shown to act as DAMPs as well. Cell-free nDNA is believed to obtain immune activity by associating with binding proteins and gaining intracellular uptake by cells of the innate immune system [113]. Specifically, circulating nDNA and mtDNA can activate innate immune cells through TLR9 and cause systemic inflammation [16, 114].

4.5.2 DNA in Sepsis

Plasma levels of nDNA and mtDNA have been shown to be increased in patients with severe sepsis and septic shock [115–118]. Plasma DNA concentrations correlate with severity of illness and are significantly higher in intensive care unit (ICU) non-survivors compared to ICU survivors. Circulating mtDNA levels have a positive correlation with bacteremia in Papio baboons following Bacillus anthracis infusion [119]. Baboons pre-treated with activated protein C (aPC) showed a reduction in mtDNA levels following resolution of bacteremia which was not observed in untreated baboon. The authors suggest aPC prevents persistent endogenous inflammatory tissue injury after resolution of bacteremia. Given its anti-inflammatory and anticoagulant properties, activated protein C was proposed as a promising treatment of sepsis in humans. Despite the initial clinical trial of recombinant human activated protein C demonstrating reduced mortality in patients with severe sepsis [120], subsequent post-marketing trials demonstrated no significant differences in mortality [121–123] and aPC was removed from all markets in 2011.

4.5.3 DNA in Trauma

After acute blunt traumatic injury, nuclear and mitochondrial DNA plasma levels are also markedly elevated compared to healthy controls [16, 17]. Plasma mtDNA concentrations correlate positively to injury severity, and concentrations are higher among non-survivors compared to survivors of trauma [116, 124]. Moreover, plasma mtDNA levels are higher in trauma patients with adverse outcomes compared to those who do not develop complications [17].

5. Therapeutic Strategies

Therapies to modulate the systemic inflammatory response to sepsis have generally been unsuccessful to date. Therapies targeting PAMPs and DAMPs that have been tested in humans have focused primarily on blocking endotoxin associated with gram negative infections. Studies going back several decades have tested antibodies to LPS (HA-1A and E5) and more recently, to small molecule inhibitors (Eritoran) but have all been failures to date [125–130]. Recently, a monoclonal antibody against lipoteichoic acid has also entered clinical trials for the treatment of staphylococcal sepsis in very low birth weight infants [131].

Additional strategies in sepsis have included nonselective immune suppression (corticosteroids, IV immunoglobulin), selective neutralization of microbial products (TLR antagonists), selective neutralization of inflammatory mediators (anti-TNFα, IL-1 ra, PAF receptor antagonist), immune system stimulation (granulocyte colony stimulating factor, interferon-γ, IL-7) and anticoagulation (activated Protein C, thrombomodulin, antithrombin III, heparin) [132]. Additional agents that aim to inhibit the adaptive immune suppression such as anti-programmed death ligand (PD-L1) and recombinant IL-7 are currently in phase II clinical trials for severe sepsis or septic shock (NCT02576457 and NCT02640807, respectively). interferon-γ (IFNγ) has been commonly used in AIDS-associated immune suppression and has proven beneficial at reducing opportunistic fungal and bacterial infections [133–135]. It has also shown some benefit in a very small clinical trial but has not been systematically tested in large randomized clinical trials [136]. Two studies with IFNγ were attempted in patients with blunt trauma, and although overall survival benefit was not seen, reductions in post-trauma infections were observed [137, 138]. With a better understanding of the mechanisms of PAMPs and DAMPs in sepsis and trauma, new therapeutics are likely to develop that inhibit the exaggerated or prolonged inflammatory response while maintaining appropriate host antimicrobial properties necessary to prevent and combat infections.

6. Conclusion

The early innate immune and inflammatory responses to both tissue damage associated with trauma and microbial products associated with infection and sepsis rely on a complex and overlapping network of PAMPs and DAMPs. The natural question is: why are there so many host and pathogen ligands recognized by multiple pattern recognition receptors? With these large numbers of ligands and receptors that show overlapping and complementary functions, there is ultimately a rather limited number of nuclear signaling pathways that lead to activation of innate immunity and expression of early activation genes. In its most simplistic form, an overall view of innate immunity is one in which multiple recognition systems lead to a prototypical immediate host immune and inflammatory response. Studies with knockout mice provide unique insights about the redundancy of the pattern recognition system. This is particularly true for gram negative infections that are recognized by multiple PRRs simultaneously. For example, not surprisingly, TLR4^−/− mice are extremely resistant to LPS administration [55], yet these animals have only modest deficiencies in host responses to polymicrobial infection [139] and Pseudomonas pneumonia [140]. In contrast, TLR2^−/− mice are hyporesponsive to gram positive bacterial cell wall products including Staphylococcus aureus peptidoglycan administration [53] and show significant deficiencies in response to streptococcal [141] and staphylococcal [142] infections. To keep this in perspective, MyD88^−/− mice, which fail to respond to most TLR receptors except TLR3 and TLR4 and lead directly to NF-κB activation, have much greater susceptibility to infections than most TLR^−/− mice [142].

It is not by chance that both DAMPs and PAMPs can play both beneficial and pathological roles in the host response. Neither is it coincidence that some forms of pattern recognition orthologues are found in our distant evolutionary relatives, such as invertebrates, insects and even early Cnidaria. There have obviously been strong evolutionary pressures to maintain, expand and refine this early recognition system. Importantly, the parallel evolution of pattern recognition receptors capable of recognizing microbial pathogens as well as endogenous danger signals has provided a significant advantage to the mammalian host, by reducing both the severity of the original infection and the likelihood of secondary infections with the damage of tissue barriers.

However, clearly this beneficial host recognition and response system can have disastrous effects when the activation is either exaggerated, inadequate, chronic in nature or directed not towards microbial invasion or tissue repair, but against healthy normal tissues. Probably the most evident demonstration of a misguided innate immune response is the sepsis and shock response to massive bacterial infections and severe trauma. In this case, it is the overabundant magnitude of the response that results in organ injury and death. With improvements in early detection and acute ICU management, more and more patients are surviving their initial insult, reestablishing homeostasis and having a relatively uncomplicated course [143]. However, a significant number of patients, commonly the aged, do not recover quickly and frequently develop a protracted and complicated ICU course. Critically ill patients who are maintained in the hospital for extended periods of time show host responses comparable to patients with metastatic, extensive tumor burden. The persistent recognition of endogenous alarmins from tumor growth and tissue injury leads to a chronic inflammatory state characterized by persistent inflammation, immune suppression and protein wasting. Recent evidence has also suggested that there are components of auto-immune diseases, such as rheumatoid arthritis, systemic lupus erythematosus and psoriasis, which are driven by endogenous alarmins and pattern recognition receptors. In this case, host tissues and their products are misidentified by both the innate and adaptive immune systems and viewed as alarmins.

It is this interaction between microbial products, danger signals and pattern recognition receptors which plays both an essential survival role in response to infection and injury, and can also be responsible for the tissue and organ injury associated with severe sepsis and trauma, or chronic inflammatory processes.

Highlights.

• PAMPs and DAMPs play a vital role in recognition of infection and cellular stress

• Early innate immune and inflammatory responses rely on complex, overlapping signals

• Exaggerated or prolonged host inflammatory response can have deleterious effects