Targeting of G-protein coupled receptors in sepsis

Since 1982, more than 80 phase II and phase III clinical trials involving patients with sepsis have been conducted. Despite this, the only interventions consistently shown to have any tangible impact on the survival from sepsis and septic shock have been early administration of appropriate antimicrobials, source control and hemodynamic stabilization. The current treatment of sepsis is centered around limiting the development of end-organ dysfunction by providing rapid source control and hemodynamic stabilization, and when needed, organ support to ensure the recovery of end-organ function. Based on differing results from numerous trials evaluating the use of corticosteroids in sepsis, the Surviving Sepsis Campaign guidelines recommend administration of glucocorticoid therapy only for those patients with septic shock who remain hemodynamically unstable despite adequate fluid resuscitation and vasopressor therapy ( Rhodes, et al., 2017 ). The negative clinical trials in sepsis also warrant attention in that they improved our understanding of its pathophysiology and shed light on the challenges of conducting clinical trials in sepsis. Prior therapeutic strategies in sepsis initially focused primarily on thwarting the vicious circle of inflammation and controlling the cytokine storm that typifies sepsis. However, over the past decade, a paradigm shift occurred in sepsis research as immune paralysis was identified as a central theme leading to mortality in a vast majority of septic patients ( Leentjens, Kox, van der Hoeven, Netea, & Pickkers, 2013 ). This shifted the emphasis from designing pharmacotherapies to block the initial hyper-inflammatory phase to therapies aimed at stimulating the adaptive immune response. Many such immune stimulatory therapies have been explored in randomized controlled trials with mixed results. More recently, randomized trials evaluating the role of immune checkpoint inhibitors for the treatment of sepsis have been started ( NCT02960854 and JapicCTI-173600). Given the encouraging results of immune checkpoint inhibitors in oncological trials, legitimate optimism exists in the critical care community regarding the use of these therapies in patients with sepsis. Table 1 provides a summary of various interventions that have been tested hitherto in clinical trials for sepsis.

GPCRs are the largest family of transmembrane receptors in humans and are involved in signal transduction of numerous chemical and physical stimuli including olfaction, vision, gustation, neurotransmission, hormonal control, metabolic regulation and modulation of local mediators. There are close to ~800 GPCRs encoded by the human genome with the vast majority of them being involved in olfaction. Among the ~369 GPCRs not involved in basic sensations (i.e. gustation, vision and olfaction), another ~150 are orphan receptors in that their ligands and regulatory functions remain uncertain ( Fredriksson, Lagerstrom, Lundin, & Schioth, 2003 ; Glukhova, Draper-Joyce, Sunahara, Christopoulos, & Wootten, 2018 ). GPCRs have regulatory functions in all organ-systems of the body and dysregulation of GPCR signaling has been implicated in the pathogenesis of numerous diseases ( Heng, Aubel, & Fussenegger, 2013 ). Phylogenetically, GPCRs have been classified into five distinct subfamilies: (i) Glutamate; (ii) Rhodopsin; (iii) Adhesion; (iv) Frizzled/Taste2; and (v) Secretin. For the purposes of this review, we will be mainly focusing on GPCRs that are implicated in the pathogenesis of sepsis and have not been previously evaluated as pharmacologic targets in clinical trials for sepsis. The general structure of all GPCRs consists of highly conserved hepta-helical transmembrane α -helix folds, an extracellular N-terminus and an intracellular C-terminal cytoplasmic tail. These cell-surface receptors transduce extracellular signals into intracellular biochemical effector pathways via G-protein (guanine nucleotide binding protein) activation. G-proteins associated with the receptor typically exist in the form of a heterotrimer composed of a G α subunit interacting with a G βγ complex. The first step in signal transduction is the binding of a ligand to the GPCR, which results in conformational changes in the receptor, typically involving disruption of strong ionic interactions between various transmembrane helices ( Rosenbaum, et al., 2007 ). These changes expose an otherwise hidden cytoplasmic surface of the receptor that facilitates guanine nucleotide exchange on the G α protein (release of guanosine diphosphate [GDP] and binding of guanosine triphosphate [GTP]) to which the receptor is coupled. Binding of GTP results in activation of the heterotrimeric G-protein complex and allows the dissociation of both G α subunit and G βγ complex from the receptor ( Rosenbaum, Rasmussen, & Kobilka, 2009 ). Depending on the types of G-protein, the dissociated G α subunit and G βγ complex can then modulate a variety of downstream signaling pathways through either stimulation or inhibition of enzymes and/or ion channels. G α proteins have the capability to both bind and hydrolyze GTP through their intrinsic GTPase activity. Binding of GTP activates the G-protein—a process that may be facilitated by guanine nucleotide exchange factors, while hydrolysis of GTP to GDP inactivates the G-protein—which may be regulated by GTPase-activating proteins. However, hydrolysis of GTP occurs at a relatively slow rate, which allows the activated G-protein (GTP bound) to have a much longer half-life in the cell than the activated receptor itself. This allows amplification of the transduced signal such that the initial ligand-receptor interaction may last for a short of period (e.g. few milliseconds), while the activated G-protein may remain active in the cell for a much longer period of time (e.g. many seconds) ( Homan & Tesmer, 2014 ). Following hydrolysis of GTP to GDP, G α proteins can associate again with G βγ complexes to form heterotrimers and re-couple with GPCRs. Despite this, the response mediated by GPCRs to their ligands tends to attenuate over time, even in the continued presence of the ligand—a phenomenon referred to as desensitization . The chief reason for this is that activation of the GPCR not only activates G-proteins, but it also activates a family of protein kinases called GPCR kinases (GRKs). The intracellular C-terminus of the GPCR contains many serine and threonine residues whose hydroxyl (–OH) groups can be phosphorylated by GRKs ( Gurevich & Gurevich, 2019 ). This phosphorylation leads to diminished receptor–G-protein coupling and increases the affinity of the receptor to bind a protein called β -arrestin. Binding of β -arrestin to the receptor further diminishes the ability of the receptor to bind ligands and promotes endocytosis of the receptor through recruitment of clathrin and adaptor protein-2 that leads to the formation of clathrin-coated pits. Dissociation of ligand from the internalized receptor reduces its affinity for β -arrestin and allows dephosphorylation of the receptor through the action of protein phosphatases ( Bahouth & Nooh, 2017 ). Recycling of the internalized receptor to the cell membrane allows it to bind ligands again and elicit cellular responses ( resensitization ). However, repeated or continued exposure to ligands favors the lysosomal degradation of internalized receptors, thereby leading to down-regulation of receptor density and persistent desensitization. It should be noted here that despite similarities among different GPCRs, individual GPCRs have unique combinations of signal transduction activities involving multiple G-protein-dependent and G-protein-independent signaling pathways along with complex regulatory processes. A variety of G-protein heterotrimers can be coupled to a diverse array of GPCRs to elicit divergent responses in various cells. There are at least 18 different G α proteins to which GPCRs can be coupled. In turn, these G α proteins form heterotrimeric complexes with G β and G γ subunits, both of which have at least 5 and 11 different types respectively ( Kroeze, Sheffler, & Roth, 2003 ). G α proteins have been broadly classified into four families: G α i/o , G α s , G α q/11 and G α 12/13. G α s proteins chiefly stimulate the activity of adenylyl cyclase, while Gα i/o proteins principally inhibit adenylyl cyclase. Adenylyl cyclase is an enzyme responsible for the formation of cyclic adenosine monophosphate (cAMP) from ATP (adenosine triphosphate). cAMP, in turn, can modulate a plethora of functions within different cells through the activation of cAMP-dependent protein kinases ( Lorenz, Bertinetti, & Herberg, 2015 ). On the other hand, G α q/11 proteins mainly act by activating phospholipase C (PLC), whereas G α 12/13 proteins chiefly stimulate the activity of the Rho family of GTPases through RhoGEFs ( Suzuki, Hajicek, & Kozasa, 2009 ). PLC is responsible for breaking down PIP 2 (phosphatidylinositol-4,5-biphosphate), a component of membrane phospholipids, into diacylglycerol (DAG) and inositol-1,4,5-triphosphate (IP 3 ). DAG, which is confined to the cell membrane, can activate protein kinase C (a phospholipid- and calcium-sensitive protein kinase), while IP 3 diffuses through the cytoplasm and binds to ligand-gated calcium channels, thereby triggering the release of calcium ions into the cytosol. Calcium ions can stimulate the activity of calcium-dependent protein kinases through multiple mechanisms, such as by binding to calmodulin ( Mizuno & Itoh, 2009 ). Numerous experimental studies have demonstrated that each GPCR can be activated by a variety of ligands and a single GCPR can couple with multiple G-proteins; such properties of a particular receptor may be partly modified through alternative splicing and/or post-translational modifications ( Markovic, 2013 ). Moreover, biased ligands may preferentially activate a particular signaling pathway by activating certain G-proteins coupled with the receptor ( Rankovic, Brust, & Bohn, 2016 ). Additionally, a single G α protein can modulate the activity of multiple effectors in response to varying concentrations of a single ligand (for instance, G α s can stimulate the activity of PLC in addition to stimulating the activity of adenylyl cyclase) ( Hermans, 2003 ). Besides the canonical pathways of GPCR signaling mentioned here, GPCRs may also interact with a wide variety of GPCR-interacting proteins, such as PDZ-scaffold proteins, receptor-activating modifying proteins and allosteric mediators ( Romero, von Zastrow, & Friedman, 2011 ). Likewise, although many GPCRs exist and function as monomers, GPCRs can form homo- or heterodimers in association with other GPCRs ( Bulenger, Marullo, & Bouvier, 2005 ). All these factors can fine-tune GPCR signaling and influence receptor interactions with their major orthosteric ligands. With this background of GPCRs, we now serially discuss some of the GPCRs that have not been fully explored in clinical trials for sepsis but hold promise as potential therapeutic targets. Adrenergic receptors or adrenoceptors are GPCRs that mediate the effects of catecholamines including adrenalin (epinephrine) and noradrenalin (norepinephrine). Adrenoceptors are expressed on nearly all tissues of the body and mediate a diverse variety of physiologic functions including blood pressure regulation, myocardial contractility, energy mobilization, metabolic processes and bronchial responsiveness ( Ahles & Engelhardt, 2014 ). Adrenoceptors are of two broad classes viz. α - and β -adrenoceptors, each of which has further subtypes ( α 1A , α 1B , α 1D , α 2A , α 2B , α 2C and β 1 , β 2 , β 3 ). α 1 -adrenoceptors couple principally to G q/11 proteins and activate PLC, which in turn leads to an increase in the cytosolic concentration of calcium ions and activation of calcium-dependent protein kinases. α 1 -adrenoceptors are expressed on a wide variety of tissues and cells in the body, such as iris, ciliary body, blood vessels (tunica media), smooth muscles (primarily, internal anal sphincter and urethral sphincter), and cardiac myocytes. Stimulation of α 1 -adrenoceptors on various cells leads to a diverse array of physiologic responses including mydriasis, systemic vasoconstriction, and tightening of anal and urinary sphincters ( Michel & Vrydag, 2006 ). α 2 -adrenoceptors couple primarily to G i proteins and their stimulation leads to inhibition of adenylyl cyclase and a reduction in the intracellular concentration of cAMP. α 2 -adrenoceptors are primarily expressed on presynaptic autonomic neurons and their stimulation leads to a net inhibition of sympathetic outflow. Clonidine, a prototype α 2 agonist, has sympatholytic effects in the body including hypotension, bradycardia and bronchoconstriction ( Jamadarkhana & Gopal, 2010 ). α 2B receptors are mainly involved in neurotransmission at the level of the spinal cord, while α 2C receptors modulate the release of catecholamines from adrenal medullary chromaffin cells ( Gyires, Zadori, Torok, & Matyus, 2009 ). β -adrenoceptors are of three types (β 1 , β 2 and β 3 ) and are widely distributed throughout various tissues and cells of the human body. All β -adrenoceptors couple chiefly to G s proteins and their stimulation leads to an increase in the intracellular concentration of cAMP through activation of adenylyl cyclase ( Baker, 2010 ). β 1 -adrenoceptors are densely expressed on cardiac myocytes and their stimulation leads to positive inotropic (increased contractility), chronotropic (increased heart rate), dromotropic (increased conduction) and lusitropic (rapid relaxation) effects. β 1 -adrenoceptor stimulation has also been implicated in stimulating the release of renin from the juxtaglomerular apparatus. β 2 -adrenoceptors are expressed on hepatocytes, adipocytes, blood vessels, detrusor muscle, intestinal smooth muscle and bronchial smooth muscle. Stimulation of β 2 -adrenoceptors leads to vasodilation in skeletal muscle beds, increased gluconeogenesis, enhanced lipolysis, bronchodilation, urinary retention and constipation. Both β 1 - and β 2 -adrenoceptors are expressed on the ciliary epithelium and their stimulation enhances the production of aqueous humor. β 3 -adrenoceptor expression has been demonstrated in adipose tissue, intestinal smooth muscles, detrusor muscle, prostate gland, and coronary endothelial cells ( Ursino, Vasina, Raschi, Crema, & De Ponti, 2009 ). Stimulation of β 3 -adrenoceptors leads to lipolysis, constipation, urinary retention, and coronary vasodilation. A number of pharmacologically active drugs targeting α - and β -adrenoceptors are currently in use for the treatment of various disorders. Phenylephrine, a selective α 1 -agonist, has potent vasoconstrictor effects and is used as a vasopressor of choice in patients with hypertrophic obstructive cardiomyopathy ( Gajewski & Hillel, 2012 ). Selective α 1-antagonists (prazosin, terazosin, tamsulosin) are used in the treatment of patients with bladder outlet obstruction secondary to prostatic enlargement. Clonidine, an α 2 -agonist, is used for treatment of patients with opioid withdrawal and is also used as an antihypertensive agent. Yohimibine, an α 2 -agonist, is used for the treatment of erectile dysfunction. Another α 2 agonist, mirtazapine, is used as an antidepressant ( Dekeyne & Millan, 2009 ). Selective β 2 -agonists (albuterol, terbutaline, salmeterol) are potent bronchodilators and are the cornerstone of management of asthma and emphysema ( Broadley, 2006 ). Mirabegron, a selective β 3 -agonist, is approved for the treatment of detrusor overactivity ( Cernecka, Sand, & Michel, 2014 ). Likewise, dobutamine, a selective β 1 -agonist, is a potent inotropic agent and is useful in the management of cardiogenic shock. Conversely, selective β 1 -blockers (such as metoprolol and bisoprolol) have anti-arrhythmic, antianginal and anti-hypertensive effects and form the backbone of pharmacotherapy for coronary artery disease and congestive heart failure. Labetalol, a non-selective β -blocker, is used in the management of pre-eclampsia and eclampsia. Moreover, propranolol, another non-selective β -blocker, is utilized for the treatment of a variety of diseases including essential tremor, thyrotoxicosis, portal hypertension, performance anxiety disorder, and migraine headaches ( D. W. Wang, et al., 2010 ). Carvedilol, a non-selective α - and β -antagonist, is commonly used in the management of patients with congestive heart failure. Additionally, ophthalmic preparations of certain non-selective β -blockers, such as timolol, are efficacious in the management of patients with glaucoma ( Winn, Culhane, Gilmartin, & Strang, 2002 ). Given the diverse physiologic processes mediated by adrenoceptors, it is not surprising that pharmacologic agents targeting these receptors have found wide applications in numerous diseases. Adrenergic receptors have been shown to modulate inflammatory and immunological processes, which makes them potential targets for pharmacotherapy in sepsis ( Hasko & Szabo, 1998 ; Hasko, Szabo, Nemeth, & vizi, 1997 ). The sympathetic nervous system plays an important role in controlling the function of the immune system and modulating inflammation ( Hasko, 2001 ). Neurotransmitters of the sympathetic nervous system—norepinephrine and epinephrine—are released in the vicinity of immune cells in response to various stressful stimuli and fine-tune the immune response by binding to adrenoceptors on immune cells ( Sperlágh, Dóda, Baranyi, & Haskó, 2000 ). Presynaptic adrenoceptors are implicated in inhibiting the release of neurotransmitters and serve as a feedback loop ( Vizi, Orso, Osipenko, Hasko, & Elenkov, 1995 ). Presynaptic α 2 -adrenoceptors have a much greater affinity for their ligands than post-synpatic α 2 -adrenoceptors. Consequently, ligand binding to α 2 -adrenoceptors occurs predominantly on the presynaptic side and the effects of such ligands are principally determined by their interactions with presynaptic receptors. α 2 -adrenoceptor stimulation in vivo can increase the production of pro-inflammatory cytokines (TNF α and IL-12) ( Elenkov, Hasko, Kovacs, & Vizi, 1995 ). Conversely, blockade of α 2 -adrenoceptors can suppress the production of TNF α , MIP-1 α and IL-12, while increasing the production of anti-inflammatory cytokines, such as IL-10 ( Hasko, Elenkov, Kvetan, & Vizi, 1995 ). As mentioned previously, these effects are likely to be mediated through stimulation of presynaptic α 2 -adrenoceptors. In experimental models, inhibition of presynaptic α 2 -adrenoceptors was shown to increase the release of norepinephrine in lymphoid organs (including the thymus and spleen) ( Hasko, Elenkov, & Vizi, 1995 ). The high local concentration of norepinephrine can stimulate β -adrenoceptors on immune cells and alter the cytokine expression profile of these immune cells. These observations further confirm the role of presynaptic α 2 -adrenoceptors as regulators of neuro-immune interactions. β -adrenoceptors are present on immune cells and can modulate a variety of their effector functions including leukocyte trafficking, lymphocyte proliferation, phagocytosis of pathogens and release of cytokines ( Hasko, Shanley, et al., 1998 ; Hasko, Szabo, Nemeth, & Deitch, 2002 ; Pastores, Hasko, Vizi, & Kvetan, 1996 ). Experimental studies in mice showed that stimulation of β -adrenoceptors inhibited the LPS-induced release of IL-12 ( Hasko, Szabo, Nemeth, Salzman, & Vizi, 1998a ). In another study, exogenous and endogenous catecholamines were found to inhibit the release of MIP-1 α induced by LPS infusion in mice. This effect was reversed by propranolol, but not phentolamine, suggesting that the effect was mediated through activation of β -adrenoceptors. Isoproterenol, a non-selective β -agonist, was found to inhibit the production of TNF α and nitric oxide, while stimulating the production of IL-10, in the RAW 264.7 macrophage cell line ( Hasko, Nemeth, Szabo, et al., 1998 ). In vivo , isoproterenol was also found to protect against the development of endotoxin-induced vasoplegia in rats through inhibition of LPS-induced TNF α and nitric oxide release, while stimulating LPS-induced release of IL-10 ( Szabo, et al., 1997 ). Stimulation of β -adrenoceptors leads to an increase in the intracellular concentration of cAMP in immune cells. Consequently, increase in intracellular cAMP concentration by phosphodiesterase inhibitors and dopamine receptor agonists has been demonstrated to have a similar impact on cytokine secretion from immune cells ( Hasko, Szabo, Merkel, et al., 1996 ; Hasko, Szabo, Nemeth, Salzman, & Vizi, 1998b ). Stimulation of β -adrenoceptors also has an impact on the adaptive immune response by inhibiting the release of IFNγ and stimulating the release of IL-10; this inhibits the formation of T H 1 cells and promotes differentiation into T H 2 cells ( Panina-Bordignon, et al., 1997 ). The experimental studies discussed so far suggest a potentially beneficial role of β -adrenoceptor agonists and α 2 -adrenoceptor antagonists in sepsis. More recent experimental evidence has shown discordant effects of α - and β -adrenoceptors activation and blockade in different cells and tissues. In a CLP model of polymicrobial sepsis in rats, dexmedetomidine (an α 2 -adrenoceptor agonist) was shown to reduce mortality and alleviate renal dysfunction by decreasing the expression of IL-1 β , IL-6, TNF α , NFκB and TLR4; the protective effects of dexmedetomidine were blocked by atipamezole (an α 2 -adrenoceptor antagonist) suggesting the involvement of α 2 -adrenoceptors ( Qiu, et al., 2018 ). On the other hand, other experimental studies suggested that the protective effect of dexmedetomidine may be related to its anti-inflammatory effects independent of the α 2 -adrenoceptor as these effects were not antagonized by yohimbine ( J. Zhang, Wang, Wang, Zhou, & Li, 2015 ). Conversely, Hsing and colleagues demonstrated that dexmedetomidine attenuated sepsis-induced acute kidney injury by reducing expression of histone deacetylase-2 and histone deacetylase-5, while increasing expression of acetyl-histone H3 ( Hsing, et al., 2012 ). Interestingly, these effects of dexmedetomidine were antagonized by yohimbine, which suggests the possible involvement of α 2 -adrenoceptors in these effects. Further studies utilizing the CLP model of sepsis in rats showed that α 2A -adrenoceptor blockade reduces levels of HMGB1 (high-mobility group box protein 1) and attenuates sepsis-induced acute lung injury by reducing levels of TNF α and IL-6 ( Ji, et al., 2012 ). Likewise, erythropoietin was shown to reverse sepsis-induced vasoplegia by preserving α 1D -adrenoceptor expression in murine aorta, possibly through inhibition of GRK2 ( Kandasamy, et al., 2016 ). With respect to β -adrenoceptors, the effects of agonists and antagonists are divergent in different tissues. Both epinephrine infusion and propranolol were found to have deleterious effects in mice with polymicrobial sepsis induced by CLP ( Oberbeck, et al., 2004 ). On the other hand, another study showed that propranolol treatment alone was found to have lung protective effects and favorable metabolic effects in rats with sepsis induced by CLP ( J. Wilson, et al., 2013 ). Further experimental studies have shown that blockade of myocardial β 1 -adrenoceptors improves cardiac and vascular function in sepsis ( Kimmoun, et al., 2015 ). On the other hand, stimulation of β 1 -adrenoceptors by dobutamine was found to improve survival in septic rats by improving hepatic microcirculation; this effect was completely abolished by esmolol ( Fink, et al., 2013 ). Isoproterenol was also shown to have anti-inflammatory effects through down-regulation of HMGB1 (high-mobility group box protein 1) in mice with sepsis induced by CLP ( Ha, et al., 2011 ). In other experimental studies, terbutaline (a selective β 2 -adrenoceptor agonist) was shown to improve sepsis-induced organ dysfunction, circulatory failure and diaphragmatic dysfunction ( Ito, Fujimura, Omote, & Namiki, 2006 ; Tsao, et al., 2010 ; Uzuki, Yamakage, Fujimura, & Namiki, 2007 ). Overall, most studies suggest a protective effect of β 1 -adrenoceptor antagonism and β 2 -adrenoceptor stimulation in sepsis ( de Montmollin, Aboab, Mansart, & Annane, 2009 ). The discrepancies between the results of individual studies may be related to inter-specie differences, variations in experimental protocols, different dosages of drugs used, and phenotypic heterogeneity in sepsis induced by CLP. Despite the theoretical promise of targeting adrenoceptors in sepsis, clinical use of drugs targeting adrenoceptors is limited to vasopressor therapy. Norepinephrine is recommended as the first-line vasopressor for patients with septic shock ( Rhodes, et al., 2017 ). Norepinephrine is used preferentially over other vasopressors, such as dopamine and phenylephrine, as the latter are associated with more adverse effects ( De Backer, Aldecoa, Njimi, & Vincent, 2012 ; Vail, et al., 2017 ). Norepinephrine is also preferred over epinephrine as experimental evidence suggests that epinephrine may increase myocardial oxygen consumption with potentially deleterious consequences ( Ducrocq, et al., 2012 ). From a theoretical perspective, this observation is of interest as epinephrine has greater affinity for β 2 -adrenoceptors than norepinephrine, and experimental evidence suggests a beneficial role of β 2 -adrenoceptor stimulation in sepsis ( de Montmollin, et al., 2009 ). Conversely, the high affinity of epinephrine for α 2 -adrenoceptors may also counterbalance its beneficial effects in sepsis. Having said this, it seems unlikely that traditional agonists and antagonists of adrenoceptors will be of much clinical benefit in patients with sepsis and septic shock. Evidence accumulated from recent studies shows that adrenoceptors are down-regulated in patients with septic shock due to activity of GRKs and up-regulation of phosphodiesterases and phospholipases ( Sakai, et al., 2017 ; Thangamalai, et al., 2014 ). Novel methods of targeting adrenoceptors intracellularly through pepducins and aptamers may circumvent these problems and hold theoretical promise for use in sepsis. Adenosine is an endogenous purine nucleoside that is elaborated in response to tissue injury and inflammation ( Hasko & Cronstein, 2004 ). Adenosine is constitutively present in the extracellular space at low concentrations, but, its concentration increases markedly in response to tissue injury. Newby classified adenosine as a ‘retaliatory metabolite’ and postulated that adenosine, which is released in response to a wide variety of stressful stimuli, mediates an auto-regulatory loop that serves to limit end-organ injury ( Newby, 1984 ). Adenosine is believed to exert its protective effects through multiple mechanisms including reduction in the energy demand of tissues (for instance, negative inotropic effects in cardiac muscle), promotion of a more favorable tissue environment (for instance, coronary vasodilation leading to improved nutrient and oxygen delivery) and modulation of the immune response ( Antonioli, Blandizzi, Pacher, & Hasko, 2013 ; Hasko, Deitch, Szabo, Nemeth, & Vizi, 2002 ). Extracellular concentration of adenosine is tightly regulated in tissues through modulation of its production, release and metabolism as well as regulation of intracellular purinergic metabolic pathways. During tissue hypoxia, ATP is degraded to AMP (adenosine monophosphate) and the dephosphorylation of AMP to adenosine by the enzyme 5’-nucleotidase is up-regulated while the re-phosphorylation of adenosine by adenosine kinase is inhibited ( M. D. Nguyen, Ross, Ryals, Lee, & Venton, 2015 ). As the intracellular concentration of adenosine increases, adenosine is exported to the extracellular space by the function of highly specialized equilibrative nucleoside transporters ( Csoka, et al., 2015 ). Another important source of extracellular adenosine is through the action of ectonucleotidases (CD39 and CD73) on extracellular ATP, ADP (adenosine diphosphate) and AMP that is released from cells during tissue hypoxia and inflammation ( Antonioli, Pacher, Vizi, & Hasko, 2013 ). Adenosine is chiefly catabolized to inosine by the enzyme adenosine deaminase, which itself has immunomodulatory and neuroprotective effects ( Hasko, Kuhel, Nemeth, et al., 2000 ; Hasko, Sitkovsky, & Szabo, 2004 ; Liaudet, et al., 2002 ; Liaudet, et al., 2001 ; Marton, et al., 2001 ; Soriano, et al., 2001 ). In experimental models, the principal sources of extracellular adenosine have been determined to be neutrophils, endothelial cells and platelets ( Eltzschig, et al., 2004 ). Adenosine can bind to one of four different GPCRs designated A 1 , A 2A , A 2B and A 3 . These receptors are encoded by the genes ADORA1 (1q32.1), ADORA2A (22q11.23), ADORA2B (17p12-p11.2) and ADORA3 (1p13.2) respectively in humans. A 1 and A 3 receptors are principally coupled to G i proteins that lead to inhibition of adenylyl cyclase with a concomitant decrease in intracellular cAMP concentration. On the other hand, A 2A receptor is chiefly coupled to G s proteins that lead to stimulation of adenylyl cyclase with a concomitant increase in the intracellular concentration of cAMP. A 2B receptor can transduce signals through both G q and G s proteins that lead to activation of both adenylyl cyclase and PLC. This is demonstrated in Figure 2 . In mast cells, increase in intracellular cAMP concentration can up-regulate the accumulation of NFATc1, which in turn can increase the transcription of IL-4 in response to a rise in cytosolic calcium concentration (mediated by PLC β ). Mast cells have a high density of A 2B receptors and activation of these receptors can lead to mast cell degranulation and bronchospasm. A 1 receptors are concentrated in the central nervous system and are implicated in inhibition of neurotransmitter release ( Hasko, Pacher, Vizi, & Illes, 2005 ). A 3 receptors are present in the heart and, together with A 1 receptors, are involved in ischemic preconditioning of myocardium. A 2A receptors are expressed on the surface of platelets, neutrophils and endothelial cells; they are implicated in the pathogenesis of inflammation and sepsis ( Csoka, et al., 2014 ; Hasko, et al., 2011 ; Hasko & Pacher, 2008 ). Adenosine A 2A and A 2B receptors are involved in modulation of the inflammatory response ( Hasko & Cronstein, 2013 ; Hasko, Linden, Cronstein, & Pacher, 2008 ). Endothelial cells are known to be a source of adenosine due to their ability to dephosphorylate AMP and ADP to adenosine ( Eltzschig, et al., 2003 ). In particular, during inflammation, transmigrating neutrophils release AMP which is dephosphorylated by endothelial cells to adenosine ( Lennon, Taylor, Stahl, & Colgan, 1998 ). In experimental studies, adenosine has been shown to inhibit expression of adhesion molecules (vascular cell adhesion molecule-1 and E-selectin) and down-regulate secretion of IL-6 and IL-8 by human umbilical vein endothelial cells, probably through stimulation of A 2A receptors ( Bouma, van den Wildenberg, & Buurman, 1996 ; Ohta & Sitkovsky, 2001 ). In another study, adenosine stimulated IL-8 production from human microvascular cells—but not from human umbilical vein endothelial cells—through stimulation of A 2B receptors ( Feoktistov, et al., 2002 ). This suggests that both A 2A and A 2B receptors are involved in the modulation of the inflammatory response in sepsis, although the net effect of adenosine in a given tissue is likely related to its inflammatory milieu. Adenosine receptors A 2A and A 2B are also expressed on innate immune cells and are important for regulating the immune response in sepsis ( Hasko, Csoka, Nemeth, Vizi, & Pacher, 2009 ; Hasko, Pacher, Deitch, & Vizi, 2007 ). A 2A receptors are present on macrophages and their stimulation can modulate the release of pro-inflammatory cytokines ( Hasko & Pacher, 2012 ; Pinhal-Enfield, et al., 2003 ). Studies on knock-out mice have shown that A 2A receptor activation on macrophages leads to inhibition of TNF α release from macrophages ( Hasko, Kuhel, Chen, et al., 2000 ; Kreckler, Wan, Ge, & Auchampach, 2006 ). Moreover, in Adora2a −/− mice, use of selective A 2B receptor antagonists suggests an additional role of the A 2B receptor in inhibiting the release of TNF α ( Ryzhov, et al., 2008 ). However, A 2B receptors are only operational when their effect is not masked by A 2A receptors as genetic deletion of Adora2b does not influence the secretion of TNF α in the presence of functional A 2A receptors. A 2A receptors also play a key role in augmenting the release of IL-10 from macrophages as has been demonstrated in studies utilizing the RAW264.7 macrophage cell line ( Hasko, Szabo, Nemeth, et al., 1996 ; Németh, et al., 2005 ). Likewise, studies in knockout mice demonstrated a crucial role of A 2A receptors in stimulating the release of IL-10 from Escherichia coli -exposed macrophages ( Csoka, et al., 2007 ). Additionally, A 2B receptors also have been implicated in amplifying the release of IL-10 from LPS-stimulated macrophages through post-transcriptional mechanisms ( Hasko, Kuhel, Nemeth, et al., 2000 ; Koscsó, et al., 2013 ; Koscsó, et al., 2012 ). A 2B receptors have also been demonstrated to stimulate the release of IL-6 from macrophages in Adora2b +/+ but not in Adora2b −/− mice. Moreover, activation of A 2B receptors on macrophages results in the differentiation of macrophages to the anti-inflammatory M2 phenotype, while A 2A receptors also play a complementary role in this process ( Csoka, et al., 2014 ; Csoka, et al., 2012 ; Ferrante, et al., 2013 ; Grinberg, Hasko, Wu, & Leibovich, 2009 ; Koscsó, et al., 2013 ). Adenosine is also an important regulator of neutrophil function as it modulates phagocytosis and the generation of reactive oxygen species in activated neutrophils ( Cronstein, Rosenstein, Kramer, Weissmann, & Hirschhorn, 1985 ). Through A 2A receptors, adenosine decreases the production of TNF α , MIP-1 α (CCL3), MIP-1 β (CCL4), MIP-2 α (CXCL2) and MIP-3 α (CCL20) ( McColl, et al., 2006 ; Szabó, et al., 1998 ). Moreover, adenosine has also been shown to modulate the adhesiveness of neutrophils to vascular endothelium by altering the expression of adhesion molecules; specifically, A 1 and A 2 receptors enhance and reduce (respectively) the adhesiveness of neutrophils to endothelial cells ( Cronstein, et al., 1992 ). Adenosine receptors are expressed on the surface of DC and their activation can regulate the immune response in sepsis. Adenosine A 2A receptors are present on myeloid and plasmacytoid DCs, and their stimulation can lead to a shift from a pro-inflammatory to an anti-inflammatory cytokine profile (i.e. increased IL-10 production and reduced secretion of IL-6, IL-12 and IFN γ ) ( Schnurr, et al., 2004 ). This change in cytokine profile leads to a preferential differentiation of naïve CD4 + T cells to T H 2 cells (humoral immune response) as opposed to T H 1 cells (cell-mediated immune response) ( Panther, et al., 2003 ). Moreover, in animal studies, administration of DCs treated ex vivo with an A 2A receptor agonist protected mice from ischemia-reperfusion injury through suppression of IFN γ production and regulation of DC co-stimulatory molecule expression ( L. Li, et al., 2012 ). Conversely, activation of A 2B receptors on hematopoietic progenitor cells in mice leads to DC differentiation towards a distinct phenotype characterized by the expression of both monocyte and DC surface markers ( Novitskiy, et al., 2008 ). In addition, A 2B receptor stimulation on DCs augmented IL-6 secretion, which resulted in increased T H 17 polarization of naïve T cells ( Wei, et al., 2013 ). Additionally, adenosine A 1 receptors may also play a role in DC maturation as activation of A 1 receptor inhibits vesicular MHC class I cross-presentation by resting DCs ( L. Chen, Fredholm, & Jondal, 2008 ). Likewise, stimulation of adenosine A 3 receptors has been demonstrated to have anti-inflammatory effects through inhibition of IL-6 and TNF α release ( Vincenzi, et al., 2013 ). In another study, agonists of A 3 receptors were found to be protective in endotoxemic mice by decreasing levels of IL-12 and IFN γ ( Hasko, Nemeth, Vizi, Salzman, & Szabo, 1998 ). These studies suggest that adenosine plays a complex role in the differentiation and functioning of DCs and, depending on the state of the DC and the type of receptor activated, adenosine may induce differential responses in effector cells. Adenosine can indirectly affect lymphocyte function through modulation of DC maturation as discussed previously. However, adenosine can also act directly on lymphocytes by binding to adenosine A 2A receptors on the surface of lymphocytes. Activation of A 2A receptors on the surface of naïve CD4 + T cells leads to inhibition of IL-2 secretion, which suppresses proliferation of T lymphocytes ( Naganuma, et al., 2006 ). Moreover, A 2A receptor activation can also lead to up-regulation of negative co-stimulatory molecules (viz. PD-1 [programmed death protein-1] and CTLA-4 [cytotoxic T lymphocyte antigen 4]), down-regulation of CD40L and suppression of IFN γ and IL-4 release; all these actions culminate in overall suppression of the adaptive immune system ( Csoka, et al., 2008 ). At the same time, A2A receptor activation on T cells suppresses both Th1 and Th2 differentiation and activation-induced cell death ( Himer, et al., 2010 ). A 2A receptors are also expressed on natural killer (NK) cells and regulatory T (T reg ) lymphocytes. Stimulation of A 2A receptors inhibits the cytolytic activity of IL-2 activated NK cells ( Raskovalova, et al., 2005 ). Moreover, stimulation of A 2A receptors on T reg cells leads to enhanced immunosuppressive effects through the amplification of FOXP3 expression, which drives the co-expression of CD39 and CD73—both of which are involved in the generation of adenosine from dephosphorylation of exogenous ADP and AMP ( Deaglio, et al., 2007 ). Lastly, invariant natural killer T cells are also receptive to the effects of adenosine in that stimulation of A 2A receptors on invariant natural killer T cells inhibits the release of pro-inflammatory cytokines, principally IFN γ ( Lappas, Day, Marshall, Engelhard, & Linden, 2006 ). Experimental studies exploring the role of adenosine receptors in the CLP model of sepsis have shown somewhat discordant results as compared to other experimental models. In one study, the combination of an adenosine A 2A receptor agonist and P2X7 antagonist was hepatoprotective during the acute phase of sepsis ( Savio, et al., 2017 ). Likewise, A 2A and A 2B receptors were shown to attenuate ischemia-reperfusion injury in septic rat hearts ( Busse, et al., 2016 ). On the other hand, A 2A receptor antagonism was observed to afford protection against sepsis-induced lymphopenia ( Riff, et al., 2017 ). Moreover, A 2A receptor blockade and A 2B receptor stimulation increased survival in polymicrobial sepsis induced by CLP ( Cohen & Fishman, 2019 ; Csoka, et al., 2010 ). In another study, A 2B receptor blockade was shown to enhance macrophage-mediated bacterial phagocytosis and improve survival in polymicrobial sepsis induced by CLP ( Belikoff, et al., 2011 ). Moreover, the A 1 receptor antagonist L-97–1 was shown to protect against renal dysfunction and improve survival from sepsis ( C. N. Wilson, Vance, Lechner, Matuschak, & Lechner, 2014 ). Experimental studies have also demonstrated that A 3 receptor stimulation can decrease renal and hepatic injury in mice with sepsis induced by CLP, thereby leading to a reduction in mortality ( H. T. Lee, et al., 2006 ). Adenosine receptors are widely expressed on numerous cell types and have pleiotropic effects on the human body. A 1 receptor stimulation can cause both cardiovascular and pulmonary adverse effects, while A 3 receptor stimulation appears to be safe ( Conti, Monopoli, Gamba, Borea, & Ongini, 1993 ; Fishman, Bar-Yehuda, Liang, & Jacobson, 2012 ). These considerations and the protective role of A 2A receptor blockade and A 3 receptor stimulation in animal models of sepsis indicate that selective A 2A receptor antagonists (pbf-509 and v81444) and selective A 3 receptor agonists (piclidenoson [cf101] and namodenoson [cf102]) hold great promise for use in sepsis ( Antonioli, et al., 2014 ; Cohen & Fishman, 2019 ; Koscsó, Csóka, Pacher, & Haskó, 2011 ; Németh, et al., 2005 ) (see Table 2 ). Complement receptors are expressed on multiple blood cells (including erythrocytes, platelets, neutrophils, monocytes, macrophages, eosinophils, mast cells and lymphocytes) and can be broadly classified into two categories: (a) receptors that bind fluid-phase cleavage products of complement proteins (e.g. receptor for C5a); and (b) receptors that bind to complement products deposited on the surface of other cells (e.g. CR1), essentially forming a bridge that links the target cell to the receptor ( Karsten & Köhl, 2012 ). Of the first category, the most well-characterized receptor is the receptor for C5a (C5aR1 or CD88). C5aR1 is a GPCR that is expressed on neutrophils, monocytes and macrophages. Activation of the C5aR1 on neutrophils and macrophages promotes chemotaxis. Some experimental studies suggest that C5aR1 may interact cooperatively with Fc γ receptors on macrophages to enhance phagocytosis and microbial killing ( Atkinson, 2006 ). Another receptor for C5a is C5L2—a G-protein independent receptor that may serve as a decoy receptor for C5a with regulatory functions ( R. Li, Coulthard, Wu, Taylor, & Woodruff, 2013 ). The receptor for C3a (C3aR1) is expressed on B cells, mast cells, adipocytes and endothelial cells. C3aR1 has been implicated in activation of the adaptive immune response and vascular changes characteristics of acute inflammation ( Mathern, K. Horwitz, & Heeger, 2018 ). Moreover, evidence from experiments in mice suggests that both C3aR1 and C5aR1 play vital roles in the maturation and differentiation of T reg lymphocytes ( Kwan, van der Touw, Paz-Artal, Li, & Heeger, 2013 ; Strainic, Shevach, An, Lin, & Medof, 2013 ). The second category of complement receptors includes receptors for cleavage products of C3 and C4 (CR1, CR2, CR3, CR4 and CRIg) and C1qR. C1qR is a carbohydrate-rich protein expressed on the surface of lymphocytes and phagocytes. Activation of C1qR on these cells modulates phagocytosis, cytotoxicity and release of pro-inflammatory cytokines ( Ross & Medof, 1985 ). C1qR can be activated by a number of ligands including C1q, MBL, surfactant protein A and conglutinin. CR1 (receptor for C3b/C4b) is expressed on erythrocytes, neutrophils, monocytes, lymphocytes and follicular DCs. CR1 has been shown to be involved in clearance of immune complexes, ingestion of C3b/C4b-bearing particles and modulation of lymphocytic function ( J. G. Wilson, Andriopoulos, & Fearon, 1987 ). CR2 (receptor for C3d and C3dg) is present on the surface of B lymphocytes and follicular DCs. Association of CR2 with CD19 in B cells plays an important role in the activation of B cells in response to complement activation ( Matsumoto, et al., 1991 ). CR2 also plays a role in targeting immune complexes to lymphocyte-rich areas in the spleen and lymph nodes. Both CR3 and CR4 are members of the integrin family and can bind to iC3b (implicated in the alternate complement pathway); CR3 can also bind to C3b and C3dg. CR3 is implicated in neutrophil adhesion, while both CR3 and CR4 are involved in phagocytosis of microbes ( Myones, Dalzell, Hogg, & Ross, 1988 ). CRIg can bind to C3b and iC3b, and is expressed on the surface of macrophages, especially Kupffer cells. This receptor can block the activity of C3 and C5 convertases, thereby inhibiting the complement cascade ( Wiesmann, et al., 2006 ). Modulation of the complement cascade in sepsis can be a double-edged sword with over-activation leading to microbial eradication at the expense of worsening inflammation and multi-organ dysfunction, while inhibition may limit host tissue damaged at the cost of unchecked proliferation of microbial pathogens. This is substantiated by evidence from experiments where inhibition of C5a signaling improved survival ( Ward, 2008 ), while C3 deficiency was associated with worsening mortality from sepsis ( Fischer, et al., 1997 ). These seemingly paradoxical effects may be explained by the fact that different levels of complement activity are needed during the progression of sepsis: complement activation in the early phases is necessary to curtail the spread of microbes and limit microbial invasion; however, in later stages, complement over-activity in concert with the cytokine storm may lead to host tissue damage and multi-organ dysfunction. Given the pivotal role of the complement cascade in diverse physiologic activities, a number of therapeutic targets have been explored in clinical trials for various diseases (including sepsis, paroxysmal nocturnal hemoglobinuria, thrombotic microangiopathy, C3 glomerulopathy, neuromyelitis optica, antineutrophil cytoplasmic antibody-associated vasculitis, macular degeneration and others) ( Morgan & Harris, 2015 ). Most notably, infusion of C1 esterase inhibitor was shown to improve survival in patients with sepsis who had the lowest C1-esterase inhibitor activity levels ( Igonin, et al., 2012 ). Further studies continue to explore the potential utility of C1 esterase inhibitor in the treatment of patients with sepsis and septic shock ( Bobkov, Tikhonov, Shuster, Poteryaev, & Bade, 2017 ). With respect to complement receptors, a number of agonists and antagonists are currently being explored in clinical trials. Avacopan (CCX168), an oral C5aR1 antagonist, is currently being tested in phase II and III trials as a treatment for antineutrophil cytoplasmic antibody-associated vasculitis, hidradenitis suppurativa, IgA nephropathy and C3 glomerulopathy ( NCT03301467 , NCT03852472 , NCT02994927 and NCT02384317 ). Two other antagonists of the C5aR1 receptor have also been described in the literature viz. NDT9513727 (a non-peptide inverse agonist) and PMX53 (a peptide antagonist) ( H. Liu, et al., 2018 ). Another potent, selective and non-competitive antagonist of C5aR1, DF2593A, has been studied in mice for its anti-inflammatory and anti-nociceptive effects ( Moriconi, et al., 2014 ). These agents hold promise for use in patients with sepsis as C5a blockade in rats and mice in the cecal ligation and puncture (CLP) model of sepsis was highly effective in diminishing the severity of sepsis ( Rittirsch, et al., 2008 ). CCX140, an inhibitor of CCR2, is currently being explored in randomized trials for diabetic nephropathy and focal segmental glomerulosclerosis ( NCT03703908 and NCT01440257 ). Table 3 lists the pharmacologic agents currently being developed against various complement proteins. Chemokines are a family of cytokines that primarily promote leukocyte chemotaxis to the site of inflammation. The general structure of chemokines consists of a C-terminal α -helix, three β -pleated sheets and a short N-terminus that plays an important role in chemokine receptor activation. Variation in the configuration of cysteine residues closest to the N-terminal domain confers specificity to a particular chemokine and has been used to classify chemokines into four different classes: CC, CXC, CX3C and XC chemokines ( Nomiyama, Osada, & Yoshie, 2013 ). CXC chemokines harbor a single amino acid between the two cysteine residues, while the two cysteine residues are directly juxtaposed in CC chemokines. Likewise, CX3C chemokines have three amino acids between the two cysteine residues, while XC chemokines lack the first and third cysteines of the motif. Moreover, a diverse array of genes encode for CC and CXC chemokines with considerable copy number variation and allelic isoforms, which creates significant genetic diversity and influences susceptibility to various pathologies ( Guergnon & Combadiere, 2012 ). Chemokines bind to both conventional chemokine receptors (cCKRs) and atypical chemokine receptors (ACKRs) to elicit a variety of responses including chemotaxis (directional movement towards a chemical gradient in a soluble medium), chemokinesis (random movement induced by a chemical gradient in a soluble medium), haptotaxis (directional movement up a gradient of cellular adhesion molecules) and haptokinesis (random movement while adhering to cellular adhesion substrates). The nomenclature for chemokine receptors is based on the predominant type of chemokine they bind followed by the letter ‘R’ and a number denoting the order of their discovery ( Bachelerie, et al., 2014 ). Signal transduction through cCKRs typically involves G α i subunits and β -arrestins, although down-stream signaling pathways may also involve cross-talk and overlap with JAK-STAT pathways ( Kufareva, Salanga, & Handel, 2015 ). Moreover, in the case of chemotaxis, G βγ subunits also play an important role as compared to G α i subunits ( Neptune & Bourne, 1997 ). Additionally, G 12/13 subunits may also play important roles in regulating actin-dependent functions including endocytosis and cytoarchitectura l remodeling ( Kim, Kim, & Ko, 2010 ). Once chemokines tether to the extracellular loops and N-terminal domain of their cognate cCKR, the N-terminus of the cCKR interacts with its heptahelical bundle and induces conformational changes in the receptor that leads to its activation and intracellular signal transduction. ACKRs are structurally related to cCKRs but do not couple to the same signal transduction pathways as cCKRs. Although ACKRs can bind to chemokines with high affinity, it remains controversial whether chemokine–ACKR interaction actually leads to transduction of any intracellular signals at all ( Nibbs & Graham, 2013 ). Having said that, all ACKRs do play an important role in regulating chemokine abundance, distribution and localization; this can indirectly influence interactions between chemokines and cCKRs, and regulate their physiologic and pathophysiologic responses ( Nibbs & Graham, 2013 ). Additionally, although cCKRs exist as homodimers, they can aggregate with ACKRs, other cCKRs and non-chemokine-binding GPCRs (e.g. opioid receptors) to form functionally distinct heterodimers ( Hauser, et al., 2016 ). While leukocyte movement and migration were initially thought to be the dominant responses mediated by chemokines, pleiotropic effects of chemokines on a variety of cells (such as endothelial cells, epithelial cells, mesenchymal cells, neurons and astrocytes) have been demonstrated in numerous studies ( López-Cotarelo, Gómez-Moreira, Criado-García, Sánchez, & Rodríguez-Fernández, 2017 ). Chemokines mediate multiple homeostatic and inflammatory responses in sepsis and chemokine receptors can serve as potential therapeutic targets for pharmacotherapy. The homeostatic functions of chemokines include cell survival, proliferation, endocytosis, actin polymerization, cytoskeletal remodeling, integrin activation, cell-cell adhesion, chemotaxis, chemokinesis, chemorepulsion, haptotaxis, haptokinesis, haptorepulsion and trans-endothelial migration. On the other hand, the inflammatory functions of chemokines include NET formation, respiratory burst stimulation, phagocytosis, degranulation and exocytosis. Cells of the innate immune system (namely, neutrophils, monocytes, macrophages, DCs and NK cells) express cCKRs that regulate inflammatory responses. CXCR1 and CXCR2 receptors on neutrophils promote the formation of NETs ( Hazeldine, et al., 2014 ). Moreover, CXCR1 and CXCR2 receptors on both monocytes and neutrophils amplify the respiratory burst ( Walz, Meloni, Clark-Lewis, von Tscharner, & Baggiolini, 1991 ). Likewise, CCR4 expressed on the surface of macrophages up-regulates the respiratory burst in these cells ( Ness, Ewing, Hogaboam, & Kunkel, 2006 ). Bactericidal protease release can be enhanced by a variety of chemokine receptors on neutrophils (CXCR1, CXCR2 and CCR5), monocytes (CCR1 and CCR5), macrophages (CCR4), NK cells (CCR5) and dendritic cells (CCR1, CCR2, CCR3 and CCR5) ( Chabot, et al., 2006 ; Jin, Batra, Douda, Palaniyar, & Jeyaseelan, 2014 ; Matsukawa, et al., 2000 ; Sallusto & Lanzavecchia, 2000 ). Additionally, eosinophils express the CCR2 and CCR3 receptors, which promote degranulation and the respiratory burst in these cells ( Badewa, Hudson, & Heiman, 2002 ). Mast cells also express CCR1 and CCR2 receptors, which promote their activation and recruitment during inflammatory responses ( Lippert, et al., 1998 ). Apart from innate immune cells, chemokines can also influence the adaptive immune response. CCL19 can induce proliferation of CD4 + T cells and also stimulates monocytes to elaborate IL-10, which inhibits T H 1 cells and has anti-inflammatory effects ( Byrnes, et al., 1999 ; Ploix, Lo, & Carson, 2001 ). Likewise, leukocyte and lymphocyte differentiation, survival and cytokine expression profiles can be influenced by a variety of chemokines through CCR1, CCR2, CCR3, CCR5, CXCR2, CXCR3, CXCR6 and CX3CR1 receptors ( Locati, et al., 2002 ). Experimental studies suggest that chemokine receptors play important roles in sepsis and their pharmacologic stimulation or inhibition may be potentially useful for treatment. Castanheira and colleagues studied the role of ACKR2 in a CLP model of polymicrobial sepsis ( Castanheira, et al., 2018 ). They demonstrated that ACKR2 deficient mice had more severe lung and kidney lesions as compared to wild-type mice, which suggested a protective role for ACKR2 in sepsis. In another experimental study, Castanheira et al. showed that CCR5 −/− knockout mice have increased severity of systemic inflammatory responses suggesting that CCR5 may also have a protective role in sepsis ( Castanheira, et al., 2019 ). In a case-control study, Klaus et al . showed that plasma concentrations of CCL20 and CCR6 correlated with the severity of illness in patients with sepsis and septic shock ( Klaus, et al., 2016 ). In another case-control study, Xiu and others showed that increased expression of CCR2 on monocytes and DCs was a reliable marker of the development of sepsis in patients with burns ( Xiu, Stanojcic, Wang, Qi, & Jeschke, 2016 ). Likewise, CX3CR1 was shown to be protective in a murine model of septic acute kidney injury and humans with the pro-adhesive I249 CX3CR1 allele had a lower incidence of acute kidney injury in the context of sepsis ( Chousterman, et al., 2016 ). Although multiple chemokine receptors are potential targets for pharmacotherapy in sepsis, no pharmacological agent is currently being tested for sepsis in human randomized trials. Three CCR1 antagonists (CP-481,715, BX-471 and MLN-3897) had been evaluated in randomized clinical trials for rheumatoid arthritis and multiple sclerosis with negative results ( Clucas, Shah, Zhang, Chow, & Gladue, 2007 ). Reparixin—an allosteric inhibitor of CXCR1 and CXCR2—is currently being investigated in phase II trials as a treatment for metastatic triple-negative breast cancer ( NCT02370238 ). Table 4 provides a list of various pharmacologic agents currently being developed that target chemokine receptors. PARs belong to a family of highly conserved GPCRs that are expressed on a variety of cells. A characteristic feature of PARs is that they are activated upon proteolysis by the action of extracellular proteases and the distinct structural conformation adopted upon receptor proteolysis dictates the direction and type of intracellular signal transduction ( Nieman, 2016 ). The first member of the PAR to be discovered was the thrombin receptor (PAR-1) on platelets. This was followed by the identification of three other PARs viz. PAR-2, PAR-3 and PAR-4 ( Rezaie, 2014 ). Each PAR shares significant structural and sequence homology with other PARs. In general, the structure of a PAR consists of an extracellular N-terminal domain linked to a heptahelical transmembrane structure, which is in turn linked by intra- and extracellular loops to a cytoplasmic C-terminal domain (tail) ( Coughlin, 2005 ). Proteolysis of the N-terminal domain at defined protease-specific sites by various proteases results in an exposed amino acid sequence (the so-called ‘tethered ligand’) that can interact with the extracellular loops (in the main body of the receptor) and induce conformational changes and elicit intracellular signal transduction. Each protease has distinct requirements for activating a PAR including cleavage sites and co-factors ( H. Lin, Liu, Smith, & Trejo, 2013 ). Moreover, individual PARs can be cleaved by multiple different proteases at various cleavage sites, which in turn allow the transduction of a multitude of signaling events and modulation of various physiologic processes. In fact, one of the notable features of PARs is their ability to stimulate opposing signaling pathways depending on the proteolytic stimulus (‘biased signaling’) ( Zhao, Metcalf, & Bunnett, 2014 ). Another remarkable feature is the ability of PARs to physically interact with other PARs and result in their direct transactivation through the formation of heterodimers; this allows one type of PAR to influence signaling through other PARs and adds a whole new dimension to PAR signal transduction ( Gieseler, Ungefroren, Settmacher, Hollenberg, & Kaufmann, 2013 ). For instance, direct transactivation can occur through the formation of heterodimers between PAR1 and PAR4. Thrombin bound to PAR1 within the heterodimer can ‘reach over’ and cleave PAR4 with subsequent calcium influx inside platelets ( Leger, et al., 2006 ). Likewise, formation of a PAR1-PAR2 heterodimer on endothelial cells can switch the effect of thrombin from a pro-inflammatory mediator (promoting increased vascular permeability) to an anti-inflammatory factor (preserving the endothelial barrier). In sepsis, significant cross-talk occurs between the processes of coagulation and inflammation as coagulation factors can promote inflammation and vice versa. Cleavage of PAR1 by thrombin and other proteases plays an important role in triggering DIC—a phenomenon that may be seen in 30%–60% of patients with sepsis ( Tom van der Poll, 2019 ). Thrombin activates PAR1 by cleaving a peptide bond between Arg-41 and Ser-42 that liberates a ‘tethered ligand’ leading to activation of PAR1 and intracellular signal transduction through G α 12/13 , G α q and G α i subunits ( Tiruppathi, et al., 2000 ). Phosphorylation of the C-terminal domain of PAR1 by G-protein coupled receptor kinase (GRK)-3 or GRK-5 leads to signal termination. Moreover, thrombin-mediated PAR1 activation is significantly prolonged in β -arrestin 1-deficient murine fibroblasts, which suggests a key role of β -arrestin 1 in PAR1 desensitization after activation of PAR1 by thrombin ( Paing, Stutts, Kohout, Lefkowitz, & Trejo, 2002 ). Activation of PAR1 by thrombin on platelets leads to platelet aggregation, release of platelet granules, activation of adhesion proteins and morphological changes. In endothelial cells, activation of PAR1 by thrombin leads to exocytosis of Weibel-Palade bodies, expression of adhesion proteins, loss of barrier function and induction of angiogenesis. Moreover, neurons, immune cells, fibroblasts, smooth muscle cells and epithelial cells all undergo substantial changes in response to thrombin-mediated PAR1 activation ( Petäjä, 2011 ). Apart from thrombin, many other proteases can also activate PAR1 including APC, endothelial protein C receptor and matrix metalloproteinases (MMPs) with multiple pleiotropic effects. It is also important to note that PAR1 activation can have dual effects depending on the cleavage site; activation of PAR1 by thrombin and MMP-1 elicits a pro-inflammatory response (increased vascular permeability), while cleavage of PAR1 by APC and endothelial protein C receptor leads to anti-inflammatory effects (endothelial barrier protection) ( Roy, Ardeshirylajimi, Dinarvand, Yang, & Rezaie, 2016 ). MMP-1 has been found to be implicated in DIC and can disrupt the endothelial barrier through activation of PAR1; blockade of MMP1-PAR1 interaction can potentially attenuate these adverse consequences in sepsis ( Tressel, et al., 2011 ). Development of drugs and agents that specifically target PARs has been challenging in that the receptor ligand is tethered to the receptor itself and cannot diffuse away. Nevertheless, cell-penetrating peptides (pepducins), small molecules and therapeutic proteases have been used experimentally to successfully target PARs ( Flaumenhaft & De Ceunynck, 2017 ). With respect to endothelium, regulation of vascular permeability and expression of tight junction linkers between endothelial cells is dependent on multiple signaling mechanisms and factors. One of these factors is the relative expression of two G-protein-linked GTPases—RhoA and Rac1 ( Radeva & Waschke, 2018 ). RhoA is a GTPase that can induce actin filament breakdown and internalization of VE-cadherin, thereby leading to the breakdown of endothelial barrier. Rac1 has opposing effects in that it stabilizes the actin cytoskeleton and protects against endothelial cell apoptosis. The differential activity of RhoA and Rac1 can be regulated through the activation of PARs on the surface of endothelial cells ( Klarenbach, Chipiuk, Nelson, Hollenberg, & Murray, 2003 ). In sepsis, thrombin generation leads to the activation of PAR1 on endothelial cells, which promotes RhoA signaling and increases vascular permeability through the breakdown of endothelial barrier function. Conversely, activation of PAR2 by a variety of proteases can have opposing effects through Rac1 signaling and protection of the endothelial barrier. Using a pepducin approach, Kaneider and colleagues showed that PAR1 switched from being a vascular disruptive receptor to a vascular protective receptor during progression of sepsis in mice ( Kaneider, et al., 2007 ). This switch in the behavior of PAR1 required transactivation of PAR2 signaling pathways, which suggests that pharmacotherapies selectively activating PAR1-PAR2 complexes may be potentially efficacious in the treatment of sepsis. Cannabinoid (CB) receptors CB 1 and CB 2 were identified as members of the GPCR family more than two decades ago ( Howlett & Abood, 2017 ). These receptors mediate the effects of Δ 9 -tetrahydrocannabinol, an exogenous ligand derived from the plant Cannabis sativa . Endogenous ligands (called endocannabinoids) can also stimulate these receptors and have been found to be involved in a wide variety of physiologic processes ( Araque, Castillo, Manzoni, & Tonini, 2017 ). The two principal endocannabinoids are anandamide and 2-arachidonoyl glycerol, both of which are derivatives of arachidonic acid. Anandamide can be catabolized by the enzyme fatty acid amide hydrolase to arachidonic acid, while 2-arachidonoyl glycerol can be degraded to arachidonic acid through the action of monoacylglycerol lipase ( Biro, Toth, Hasko, Paus, & Pacher, 2009 ; Cao, et al., 2013 ; Di Marzo & Piscitelli, 2015 ). CB 1 and CB 2 receptors are widely distributed throughout different cells and tissues of the body ( Stella, 2010 ). CB 1 receptors are most abundant in the brain with high expression noted in the basal ganglia, hippocampus, cerebellum and cortex. Within the nervous system, CB 1 receptors are chiefly localized on the terminals of central and peripheral neurons. This distribution correlates with the function of these receptors in memory, cognition, analgesia and mood. Outside of the nervous system, CB 1 receptors have been detected in numerous tissues including heart, lung, liver, prostate, vas deferens, uterus, ovary, adrenal glands, bone marrow, thymus and tonsils. CB 2 receptors have been found to be expressed heavily on macrophages, neutrophils and lymphocytes in the spleen, thymus and tonsils. In contrast with CB 1 receptors, CB 2 receptor expression in the healthy central nervous system is minimal, although CB 2 receptors are up-regulated in various diseased states ( Pal Pacher, Steffens, Haskó, Schindler, & Kunos, 2018 ). A whole body of literature suggests that the endocannabinoid system plays important roles in inflammatory processes including sepsis ( Csoka, et al., 2009 ; Mukhopadhyay, Horváth, et al., 2011 ; M Rajesh, et al., 2008 ). CB 1 receptors are expressed by neurons in the hypothalamus and these receptors are involved in the initiation of LPS-induced hypotension ( Varga, Wagner, Bridgen, & Kunos, 1998 ). Experimental evidence suggests that LPS-induced hypotension involves a process in which an inflammatory signal is conveyed from the periphery to the brain via autonomic sensory nerves, which then precipitates vasoplegic shock through a central mechanism requiring activation of neurons in the preoptic/anterior hypothalamic area ( Villanueva, et al., 2009 ). Given that rimonabant (a CB 1 inverse agonist) can attenuate the fall in arterial pressure evoked by LPS infusion in mice and also lower plasma concentrations of pro-inflammatory cytokines, this suggests a crucial role played by endocannabinoids in mediating LPS-induced hypotension ( Godlewski, Malinowska, & Schlicker, 2004 ). Moreover, experimental evidence suggests that vasopressin release from the hypothalamus may be reduced in septic shock, possibly through the action of endothelin-1 on endothelin A receptors. Interestingly, endocannabinoids are involved in mediating the inhibitory effects of endothelin on vasopressin release, primarily through stimulation ofCB1 receptors ( M. C. Leite-Avalca, et al., 2016 ; Vercelli, Aisemberg, Billi, Wolfson, & Franchi, 2009 ). In another study, anandamide was found to be implicated in mediating LPS-induced nitric oxide production, which was antagonized by CB 1 and CB 2 receptor antagonists ( Gardiner, March, Kemp, & Bennett, 2002 ). Other experiments have shown that endocannabinoids exert vascular tone regulatory effects through an increase in sympatho-adrenal activity (mediated centrally by CB 1 receptors), which could be abolished through antagonism of β -adrenoceptors ( Gardiner, March, Kemp, & Bennett, 2005 ). In fact, hyporesponsiveness to norepinephrine observed in an experimental model of polymicrobial sepsis could be reversed through inhibition of CB 1 receptors by AM-251 (a selective CB 1 receptor antagonist). Apart from the effects on blood vessels, anandamide may also mediate LPS-induced hypothermia in rats which can be abolished by rimonabant ( Steiner, et al., 2011 ). In another study, endocannabinoids were found to be implicated in LPS-induced septic ileus in mice, which was sensitive to blockade with selective CB 1 and CB 2 receptor antagonists (HU210 and JWH133 respectively) ( Y. Y. Li, et al., 2010 ). In a mouse model of colitis, inhibition of monoacylglycerol lipase by JZL 184 (with consequent increase of 2-arachidonoyl glycerol levels) resulted in reduction of histologic evidence of colitis and decreased levels of pro-inflammatory cytokines ( Alhouayek, Lambert, Delzenne, Cani, & Muccioli, 2011 ). This effect was abolished by co-administration of CB 1 and CB 2 receptor antagonists (SR141716A and AM630), which confirmed the involvement of cannabinoid receptors. In the CLP model of sepsis, rimonabant decreased markers of septic multi-organ dysfunction ( M. Leite-Avalca, et al., 2019 ). Likewise, antagonism of the CB 1 receptor was found to protect against hepatic ischemia-reperfusion injury in experimental endotoxemia ( Caraceni, et al., 2009 ), mainly through prevention of endotoxin-related hypotension and inhibition of neutrophil recruitment (in turn driven by a reduction in levels of CXCL1 and MIP-2 α ) ( Smith, Denhardt, & Terminelli, 2001 ). In another study, stimulation of CB 1 receptors by CB receptor agonists decreased levels of TNF α and IL-12, and increased level of IL-10 in an experimental model of endotoxemia in mice ( P Pacher & Hasko, 2008 ; Pertwee, 2012 ; Smith, Terminelli, & Denhardt, 2000 ; Tschöp, et al., 2009 ). Moreover, stimulation of CB 2 receptors was also found to protect against ischemia-reperfusion injury in an in vivo mouse model by decreasing inflammatory cell infiltration and reducing levels of TNF α , MIP-1 α (CCL3) and MIP-2 α (CXCL2) ( Batkai, et al., 2007 ; Mukhopadhyay, Rajesh, et al., 2011 ; Mohanraj Rajesh, Pan, et al., 2007 ). Additionally, CB 2 receptor activation was shown to attenuate TNF α -induced human endothelial cell activation and transmigration of monocytes in a mouse model of LPS-induced hypotension ( Mohanraj Rajesh, Mukhopadhyay, et al., 2007 ; M Rajesh, et al., 2008 ). Despite the theoretical promise of targeting cannabinoid receptors, pharmacologic targeting of cannabinoid receptors for their anti-inflammatory effects has not been materialized hitherto. The main reason for this stagnation is that systemic antagonism of CB 1 and CB 2 receptors is fraught with neuropsychiatric adverse effects. Rimonabant was initially introduced in the European market as an anorectic drug; however, the drug was withdrawn later after serious psychiatric adverse effects were reported. This discouraged pharmaceutical companies from developing further drugs targeted at the cannabinoid receptors. However, there are multiple different pharmacological strategies that could still be theoretically used to design drugs targeting these receptors. Such strategies could include the use of drugs that do not cross the blood-brain barrier, use of partial CB 1 receptor agonists, use of selective CB 2 receptor agonists, combined use of a selective CB 1 receptor antagonist with a selective CB 2 receptor agonist, modulation of local endocannabinoid metabolism, and/or targeting CB 1 or CB 2 receptor signaling in conjunction with drugs targeting other intracellular signaling pathways ( Klein, 2005 ). Endothelin (ET)-1 is a hypoxia inducible peptide that plays important roles in circulatory regulation. Endothelin receptors (ET A , ET B1 and ET B2 ) are GPCRs that mediate vasoconstriction (ET A and ET B2 receptors) and vasodilation (ET B1 receptors) in certain microcirculatory beds ( Inscho, Imig, Cook, & Pollock, 2005 ). In experimental models of endotoxemia, ET-1 has been shown to be implicated in alteration of respiratory mechanics, pulmonary hypertension, myocardial depression and systemic hypotension through differentially expressed ET A and ET B receptors in various tissues ( Forni, et al., 2005 ). Likewise, in patients with sepsis, plasma ET-1 concentration was correlated with myocardial depression and overall severity of sepsis ( Pittet, et al., 1991 ). Rutai and colleagues recently showed that in a CLP model of experimental sepsis, a combination of IRL-1620 (ET B1 receptor agonist) and ETR-p1/fl peptide (ET A receptor antagonist) improved sepsis-induced hypotension, normalized microcirculatory perfusion and restored mitochondrial oxidative phosphorylation ( Rutai, et al., 2019 ). This suggests that selective ET A receptor antagonists and selective ET B1 receptor agonists may be potentially useful in the treatment of sepsis. Given that ET receptor antagonists (such as ambrisentan, macitentan and bosentan) are already approved for use in pulmonary hypertension, these agents should be explored in randomized trials for patients with sepsis ( Kowalczyk, Kleniewska, Kolodziejczyk, Skibska, & Goraca, 2015 ). Lysophospholipid receptors are another group of GPCRs that could be potentially targeted in patients with sepsis. As the name suggests, endogenous ligands for these receptors are lysophospholipids viz. lysophosphatidic acid (LPA), sphingosine-1-phosphate (S1P), lysophosphatidylinositol (LPI) and lysophosphatidylserine (LysoPS). The receptors for these lysophospholipids are LPA 1–6 , S1P 1–5 , LPI 1 and LysoPS 1–3 respectively. Lysophospholipids have been implicated in a diverse array of physiological and pathophysiological processes including fibrosis, pain, inflammation, fertility, osteogenesis, immune modulation and angiogenesis ( Kihara, Mizuno, & Chun, 2015 ). Specifically, S1P receptors are involved in lymphocyte development, maturation and recirculation ( Tsai & Han, 2016 ). S1P 2 receptors have also been shown to modulate vascular development and remodeling as well as alter cell motility ( Rosen, Stevens, Hanson, Roberts, & Oldstone, 2013 ). S1P 2 receptors couple to G α 12/13 proteins and activate Rho signaling while inhibiting Rac signaling. This enhances actin filament breakdown and promotes increased vascular permeability by breaking down VE-cadherin between endothelial cells (as discussed previously in section 4.5 ). In the experimental CLP model of sepsis, elevated S1P 2 receptor levels in peripheral blood mononuclear cells were positively correlated with the severity of sepsis; S1P 2 receptor signaling was found to suppress macrophage phagocytosis ( Hou, et al., 2015 ). Moreover, in another experimental study using the CLP model of sepsis, ileal expression of sphingosine kinase 1 (enzyme responsible for synthesis of S1P by phosphorylation of sphingosine) was up-regulated six-folds in septic mice; pharmacological inhibition of sphingosine kinase 1 alleviated symptoms of sepsis ( Ugwu & Ho, 2019 ). Likewise, S1P 1 receptor agonists were found to be beneficial in improving renal microcirculation in mice with sepsis induced by CLP ( Z. Wang, Sims, Patil, Gokden, & Mayeux, 2015 ). In another experimental study, S1P was found to be implicated in thymic involution and lymphocyte apoptosis in mice with sepsis secondary to CLP ( Kuchler, et al., 2017 ). Additionally, the S1P analogue, FTY720, was found to reduce vascular permeability and plasma extravasation in mice with CLP-induced sepsis ( Lundblad, Axelberg, & Grande, 2013 ). Fingolimod, a non-selective S1P receptor agonist and selective down-regulator of S1P 1 receptors, is already approved for use in patients with remitting-relapsing multiple sclerosis ( Chaudhry, Cohen, & Conway, 2017 ). Other selective agonists and antagonists of S1P receptors may be potentially useful in the treatment of sepsis. Apart from S1P receptors, LPI 1 receptor (GPR55) is another GPCR that may be a potential target for pharmacotherapy in sepsis. This receptor was first described as a novel endocannabinoid receptor because of its sequence homology to cannabinoid receptors CB 1 and CB 2 ( Yang, Zhou, & Lehmann, 2016 ). Later, LPI was found to be the endogenous ligand binding to this receptor, which led to its re-classification as a receptor for LPI (i.e. LPI 1 receptor). GPR55 is expressed on a variety of tissues including endothelium, adrenal glands, intestines, spleen and leukocytes ( Henstridge, et al., 2011 ). GPR55 mediated signaling is found to be implicated in energy metabolism as well as innate immunity through stimulatory effects on neutrophils, monocytes, NK cells and mast cells ( Chiurchiu, Lanuti, De Bardi, Battistini, & Maccarrone, 2015 ; Simcocks, et al., 2014 ). In an experimental model of colitis, GPR55 knockout mice exhibited less severe inflammation ( Schicho & Storr, 2012 ). Moreover, increased level of GPR55 expression was noted in the gastrointestinal tracts of septic mice ( X. H. Lin, et al., 2011 ). Experiments in mice with sepsis induced by endotoxemia, GPR55 inhibition (using antagonists CID16020046 or O-1918) resulted in decreased levels of pro-inflammatory cytokines (TNF α and IL-6) and reduced leukocyte adherence in submucosal venules ( Zhou, Yang, & Lehmann, 2018 ). These experimental studies suggest that selective targeting of the GPR55 may be of value in sepsis, although further research is needed to understand the pleiotropic effects mediated by this receptor and to design selective agonists and antagonists for this receptor. Melatonin (5-methoxy- N -acetyltryptamine) is the primary chronobiotic hormone produced by the pineal gland and plays a role in diverse physiologic processes including regulation of sleep and circadian rhythms, pubertal development, seasonal adaptation, learning, memory, glucose metabolism, hemostasis, antioxidant defenses and modulation of the innate immune system ( Carrillo-Vico, Lardone, Alvarez-Sanchez, Rodriguez-Rodriguez, & Guerrero, 2013 ). Melatonin exerts its physiological effects through two different GPCRs viz. MT 1 and MT 2 receptors. Both MT 1 and MT 2 receptors couple to G i and G q/11 proteins, and inhibit adenylyl cyclase, stimulate phosphorylation of MAPK and extracellular signal-regulated kinase, and increase potassium conductance through inwardly rectifying potassium channels ( Emet, et al., 2016 ). Like other GPCRs, MT 1 and MT 2 receptors can form homo-dimers or hetero-oligmers, which modifies the physiologic and pharmacological properties of these receptors. MT 1 and MT 2 receptors are expressed on a variety of tissues including the brain (principally hypothalamus), retina, heart, blood vessels, testes, ovary, skin, liver, kidney, adrenal cortex, immune cells, pancreas and spleen ( Slominski, Reiter, Schlabritz-Loutsevitch, Ostrom, & Slominski, 2012 ). Melatonin has been shown to be elaborated by human lymphocytes and induces the secretion of IL-2 ( Carrillo-Vico, et al., 2004 ). Moreover, daily rhythms of melatonin and IL-2 are transiently lost in inflammatory diseases with the recovery of IL-2 rhythm following restoration of daily melatonin rhythm ( Pontes, Cardoso, Carneiro-Sampaio, & Markus, 2007 ). These observations suggest the existence of a pineal gland–immune system axis that modulates the immune response. Sepsis has been shown to disrupt circadian rhythms resulting in abnormalities in melatonin secretion ( Bellet, et al., 2013 ). Chronodisruption, in turn, has been associated with alterations of the immune system that could potentially worsen outcome from sepsis ( Acuna-Castroviejo, et al., 2017 ). Experimental evidence suggests that mice may be at an elevated risk of sepsis at night as compared to during daytime because of variations in melatonin levels and its effects on the immune system ( K. D. Nguyen, et al., 2013 ). In the LPS model of experimentally induced sepsis, melatonin inhibited the inflammatory response induced by LPS infusion in mice in a dose-dependent manner ( Escames, Lopez, Ortiz, Ros, & Acuna-Castroviejo, 2006 ). Moreover, melatonin was shown to alleviate sepsis-induced liver damage in mice through inhibition of the NFκB pathway ( Garcia, et al., 2015 ). In the CLP model of experimental sepsis, melatonin was also shown to have anti-oxidant effects and direct effects on the mitochondria that boosts the production of ATP and impedes the activation of the NLRP3 (Nucleotide-binding oligomerization domain-like receptor family, pyrin domains-containing protein 3) inflammasome ( Escames, et al., 2006 ). Likewise, melatonin was also shown to enhance the antibacterial activity of neutrophils in the CLP model of experimentally induced sepsis ( Xu, et al., 2019 ). Moreover, melatonin has also been shown to have stimulatory effects on nearly all innate immune cells including monocytes, NK cells and macrophages ( Calvo, Gonzalez-Yanes, & Maldonado, 2013 ). These results suggest that melatonin signaling may be a potential therapeutic target in sepsis and pharmacotherapies that increase the local concentrations of melatonin may be beneficial for patients with sepsis. At present, melatonin receptor agonists (ramelteon, agomelatine and tasimelteon) are already approved for the treatment of sleep and mood disorders. A phase II clinical trial (Eudract # 2008–006782-83) is currently evaluating the anti-inflammatory effects of an injectable formulation of melatonin (PCT/ES2015070236) for patients with sepsis. Therapies directed at melatonin signaling may be potentially useful in the management of patients with sepsis. Resolution of acute inflammation was traditionally thought to be a passive process with dilution of pro-inflammatory mediators and local chemo-attractants. Evidence published over the past two decades has shown that inflammation is a tightly regulated process and, its initiation and termination is governed by fine-tuned chemical mediators including lipoxins and specialized pro-resolving mediators (SPMs) ( Serhan, 2014 ). SPMs are lipid derivatives derived from polyunsaturated fatty acids which play important roles in resolving tissue inflammation (termed catabasis ). Catabasis consists of a number of discrete steps including removal of cellular debris and dead microbes by phagocytes (termed efferocytosis ), restoration of vascular integrity and regeneration of tissues. SPMs are divided into four classes viz. D-series resolvins (R V D), E-series resolvins (R V E), protectins and maresins ( Basil & Levy, 2016 ). R V Ds are derivatives of docosahexaenoic acid, while R V Es are derivatives of eicosapentaenoic acid. R V Ds act through GPCRs and actively promote resolution of inflammation through enhanced efferocytosis and restoration of tissue integrity. R V D 1 acts through the formyl peptide receptor 2 (ALX/FPR2) and GPR32 receptor—also known as R V D 1 receptor. FPR2 receptor is expressed on a variety of cells including monocytes, neutrophils, epithelial cells, hepatocytes and astrocytes ( Schmid, Gemperle, Rimann, & Hersberger, 2016 ). Pro-resolving effects mediated through the FPR2 receptor involve suppression of Ca 2+ -calmodulin-dependent protein kinase and subsequent inhibition of p38 MAPK phosphorylation. R V D 1 receptor is expressed on macrophages and is activated by a number of D-series resolvins viz. R V D 1 , R V D 3 and R V D 5 . Activation of the R V D 1 receptor on macrophages results in enhanced efferocytosis and differentiation of macrophages into M2 phenotype ( Schmid, et al., 2016 ). Moreover, activation of R V D 1 receptor on T cells results in decreased differentiation into T H 1 and T H 17 phenotypes ( Chiurchiu, et al., 2016 ). R V D 2 acts through the GPR18 receptor—now termed the DRV2 receptor. Interaction of RvD 2 with DRV2 receptor results in inhibition of neutrophil chemotaxis, decreased monocyte adhesion to adipocytes, and induces efferocytosis of apoptotic neutrophils ( Spite, et al., 2009 ). In an experimental model of sepsis induced by CLP, RvD 2 significantly improved survival through activation of DRV2 receptors and enhanced phagocytosis-mediated bacterial clearance ( Chiang, de la Rosa, Libreros, & Serhan, 2017 ). In patients with sepsis, resolvins were also found to be predictive of the development of acute respiratory distress syndrome and overall survival ( Dalli, et al., 2017 ). R V E 1 acts as a full agonist of the chemokine-like receptor 1 for which reason this receptor is often referred to as the ERV1 receptor. R V E 2 also acts as a partial agonist of the same receptor. Interaction of R V E 1 with ERV1 receptor on neutrophils leads to neutrophil apoptosis and efferocytosis ( El Kebir, Gjorstrup, & Filep, 2012 ). Macrophages derived from mice deficient in the ERV1 receptor have an enhanced ability to produce pro-inflammatory cytokines, which is consistent with a pro-resolving effect of R V E 1 ( López-Vicario, et al., 2017 ). In human monocyte-derived macrophages, these pro-resolving effects of R V E 1 were abolished by an antibody directed against ERV1 ( Ohira, et al., 2010 ). It is worth noting here that binding of different ligands to the same resolvin receptors can elicit pro-inflammatory responses. For instance, binding of chemerin to ChemR23 and binding of serum amyloid A to FPR2 can have pro-inflammatory effects. Protectins and maresins are two other classes of SPMs that are both derived from docosahexaenoic acid. Protectins are docosatrienes derived from docosahexaenoic acid, while maresins (macrophage mediators in resolving inflammation) are produced in macrophages through 14-lipoxygenation of docosahexaenoic acid ( Serhan, et al., 2009 ). Maresin 1 has been shown to limit neutrophil infiltration in murine peritonitis models and enhance efferocytosis by human macrophages by binding to ALX/FPR2 ( Serhan, et al., 2012 ). Moreover, maresin 1 was also shown to attenuate sepsis induced in mice by CLP and potentially alleviated mitochondrial dysfunction in patients with sepsis ( Gu, et al., 2018 ). Protectin D1 (also referred to as neuroprotectin D1) has been shown to play important roles in efferocytosis, wound healing, neovascularization, neuro-protection, and corneal protection from thermal injury ( Serhan, Dalli, Colas, Winkler, & Chiang, 2015 ). Neuroprotectin D1 exerts its actions by binding to the GPCR GPR37. Given that pro-resolving effects of SPMs are mediated through GPCRs, these receptors represent potential targets for pharmacotherapy in sepsis. Obestatin and ghrelin are two peptides encoded by the GHRL gene and produced by proteolytic cleavage of pre-proghrelin. These peptide hormones bind to the ghrelin receptor (also known as GPR39 or growth hormone secretagogue receptor [GHS-R]) and exert antagonistic activities with regards to satiety and hunger (obestatin signals satiety, while ghrelin stimulates hunger). GHS-R is expressed in a variety of organs including the gastrointestinal tract, pancreas, liver, adipose tissue, thyroid, kidney, lung and heart. Ghrelin and obestatin regulate a wide variety of physiologic responses including glucose metabolism, energy expenditure, food intake, growth hormone secretion, fat deposition, muscle build-up and bone formation ( Pradhan, Samson, & Sun, 2013 ). Moreover, ghrelin has also been shown to modulate the cardiac sympathetic nerve activity and may improve prognosis after a myocardial infarction by reducing the incidence of arrhythmias ( Mao, et al., 2012 ). GHS-R is expressed on innate immune cells including macrophages and couples to G q/11 proteins. Experimental studies have shown that GHS-R may be involved in the pathogenesis of many diseases including metabolic syndrome, depression and colitis ( Petersen, et al., 2011 ; Sunuwar, Medini, Cohen, Sekler, & Hershfinkel, 2016 ). In an experimental model of sepsis induced by LPS, GHS-R expression was up-regulated in peritoneal macrophages and stimulation of GHS-R enhanced LPS-induced IL-10 secretion ( Muneoka, et al., 2018 ). In this study, oral administration of a GHS-R agonist afforded improved survival in mice with LPS-induced sepsis through its anti-inflammatory effects. Likewise, ghrelin was shown to attenuate the release of IL-6 in response to LPS stimulation in the murine dopaminergic SN4741 cell-line ( Beynon, et al., 2013 ). In the CLP model of experimentally induced sepsis, ghrelin had protective effects mediated in part by the restoration of CD4 + T cell proliferation ( M. Zhou, et al., 2018 ). Experimental studies on knockout mice have shown that GHS-R deletion affects macrophage polarization and alters the ratio between M1 and M2 phenotypes ( L. Lin, et al., 2016 ). Ghrelin has also been shown to inhibit the NFκB pathway of inflammation in rats with sepsis-induced acute lung injury ( Wu, et al., 2007 ). Studies in patients with sepsis have revealed that elevated ghrelin levels portend a favorable prognosis in sepsis ( Koch, et al., 2010 ). Moreover, expression of GHS-R in rat aorta, heart and small intestine was markedly up-regulated in sepsis along with increased vascular sensitivity to ghrelin ( Wu, Zhou, Cui, Simms, & Wang, 2004 ). Results of these studies suggest that GHS-R may play diverse roles in the inflammatory response in various tissues and may be a potential target for pharmacotherapy in sepsis. Short-chain fatty acids (SCFAs) are produced by colonic bacteria as byproducts of fermentation of non-digestible and partially digested polysaccharides. SCFAs are a subset of fatty acids containing six or less carbon molecules, such as acetate, butyrate and propionate. The quantity and repertoire of SCFAs produced in the colon are in part determined by composition of the gut microbiome and reflect the importance of the gut microbiome in physiologic and pathophysiologic processes ( Morrison & Preston, 2016 ). SCFAs are taken up by colonic enterocytes and either metabolized locally or transported across the gut epithelium into the portal circulation. SCFAs have been implicated in a variety of physiologic and pathophysiologic processes including metabolism, inflammation and immune responses to infection. SCFAs are believed to signal through two main mechanisms viz. inhibition of histone deacetylases and GPCRs. A number of GPCRs are involved in SCFA signaling including GPR41 (free fatty acid receptor 3 [FFAR3]), GPR43 (free fatty acid receptor 2 [FFAR2]) and GPR109A (hydroxycarboxylic acid receptor 2 [HCAR2]) ( Tan, et al., 2014 ). FFAR2 is the primary receptor for acetate in that acetate is the most selective ligand for this receptor, while the most potent ligand for this receptor is propionate. FFAR2 is expressed throughout the gastrointestinal tract on a variety of cells including enterocytes, innate immune cells, pancreatic islet cells and neuroendocrine cells ( Tolhurst, et al., 2012 ). FFAR2 is expressed on neutrophils, monocytes, eosinophils, macrophages and B lymphocytes. Intracellular signal transduction through FFAR2 occurs primarily through G q/11 proteins and phospholipase C stimulation, while G i/o proteins also play a secondary role in inhibiting the activity of adenylyl cyclase ( Nilsson, Kotarsky, Owman, & Olde, 2003 ). FFAR2 stimulation on neutrophils results in their recruitment and activation ( Le Poul, et al., 2003 ). Experimental studies have shown that FFAR2 knock-out mice suffer from non-resolving or prolonged inflammation in models of colitis, asthma and arthritis ( Maslowski, et al., 2009 ). A phase I trial of GLPG0974 (FFAR2 antagonist) in patients with ulcerative colitis found GLPG0974 to be safe and well-tolerated ( Namour, et al., 2016 ), although results from subsequent phase II trials did not demonstrate clinical benefits ( NCT01829321 ). A possible explanation for this apparently negative result could be a compensatory increase in FFAR3 expression as seen in FFAR2 knock-out mice ( Bjursell, et al., 2011 ). FFAR3 is expressed widely on immune cells including T cells, B cells, monocytes and macrophages ( Brown, et al., 2003 ). Although FFAR3 is highly related to FFAR2 (52% similarity) and is activated by similar ligands, it has differing specificity for SCFA of different carbon lengths; for instance, pentanoate is the most potent ligand for this receptor. FFAR3 chiefly transduces signals through G i/o proteins and inhibits adenylyl cyclase to modulate cytoplasmic cAMP concentration in innate immune cells ( Xiong, et al., 2004 ). Activation of FFAR3 in conjunction with FFAR2 mediates the anti-inflammatory effects of SFCAs by reducing IL-6 and IL-8 production ( M. Li, van Esch, Henricks, Folkerts, & Garssen, 2018 ). HCAR2 is the target of niacin (nicotinic acid) and is sometimes referred to as the nicotinic acid receptor, although niacin is not believed to be the natural ligand for this receptor; butyrate is instead the natural ligand for this receptor ( Thangaraju, et al., 2009 ). Expression of HCAR2 has been demonstrated in splenic macrophages, neutrophils, Langerhans cells, adipocytes, retinal pigment epithelial cells, keratinocytes and intestinal epithelial cells. Stimulation of HCAR2 in the colon by butyrate (produced by gut microbes) suppresses intestinal inflammation by inducing differentiation of T reg cells, and inhibiting colonic macrophages and DCs ( Singh, et al., 2014 ). Intracellular signaling through HCAR2 is mediated through G i/o proteins, which inhibit adenylyl cyclase and decrease the intracellular concentration of cAMP. Additionally, HCAR2 can also stimulate phospholipase A 2 , activate the MAPK cascade and inhibit the Akt/mTOR signaling pathway ( Z. Li, et al., 2017 ; Richman, et al., 2007 ). Activation of HCAR2 by β-hydroxybutyrate on monocytes and macrophages affords neuroprotection in a stroke model in mice ( Rahman, et al., 2014 ). Moreover, HCAR2 stimulation suppressed IL-23 production by colonic DCs and inhibited colonic inflammation in a mouse model of colitis ( Bhatt, et al., 2018 ). In experimental models of sepsis induced by CLP, HCAR2 knockout mice were noted to have distinct gut microbiota composition, altered intestinal permeability and increased mortality ( G. Chen, et al., 2018 ). Interestingly, blood–mucosal barrier was reconstituted in HCAR2 knockout mice after these mice received gut microbiota from wild-type mice. These findings suggest that HCAR2 regulates the gut microbiota and plays a crucial role in maintaining the integrity of intestinal epithelial barrier. All these studies indicate that receptors for SCFAs may be attractive targets for potential pharmacotherapy in sepsis. Urotensin II is an 11-amino acid peptide that is known to be the most potent vasoconstrictor. Urotensin was named so as it was originally isolated from the urophysis (endocrine organ) of teleost fish ( Ames, et al., 1999 ). Urotensin II receptor (UTR) is a GPCR that transduces intracellular signals primarily by coupling to G q/11 proteins and leading to activation of phospholipase C with an increase in the cytoplasmic concentration of calcium ions, although inhibition of adenylyl cyclase and decrease in intracellular cAMP concentration may also be implicated ( Ziltener, Mueller, Haenig, Scherz, & Nayler, 2002 ). Expression of UTR is nearly ubiquitous and has been detected in cardiac myocytes, fibroblasts, endothelial cells, skeletal muscle cells, neurons and innate immune cells. Numerous experimental studies suggested the role of urotensin II and UTR in the pathogenesis of a variety of cardiovascular disorders including hypertension, heart failure, atherosclerosis, pre-eclampsia, diabetes mellitus and cerebrovascular disease ( Maryanoff & Kinney, 2009 ). UTR is also believed to be implicated in inflammation, principally leukocyte recruitment and migration ( Castel, et al., 2017 ). UTR is expressed on the surface of B lymphocytes, NK cells, monocytes and macrophages. Urotensin II acts as a chemoattractant for human monocytes ( Maguire, Kuc, Wiley, Kleinz, & Davenport, 2004 ) and induces increased vascular permeability in rats and mice ( Vergura, et al., 2004 ). Urotensin II induces the secretion of pro-inflammatory cytokines (such as IL-6), while pro-inflammatory cytokines (TNFα and IFNγ) up-regulate the expression of UTR ( Segain, et al., 2007 ). Moreover, urotensin II also increases the synthesis of pro-thrombotic and inflammatory markers (intercellular adhesion molecule-1, tissue factor and plasminogen activator inhibitor-1) in human coronary endothelial cells ( Cirillo, et al., 2008 ). In an experimental model of sepsis induced by CLP, urotensin II aggravated sepsis-induced lung injury in diabetic mice through UTR, which was antagonized by palosuran (UTR antagonist) ( Ugan, Cadirci, Halici, Toktay, & Cinar, 2018 ). Our knowledge of urotensin II, UTR and their role in the pathophysiology of sepsis is still evolving. As our understanding of the urotensinergic system improves, it may become a potentially feasible target for pharmacotherapy in sepsis ( Svistunov, et al., 2018 ). Having said that, the UTR antagonist palosuran has been tested in phase I human trials and found to be well-tolerated at a dose of 500 mg twice daily in healthy volunteers ( Sidharta, van Giersbergen, & Dingemanse, 2018 ).

GPCRs play diverse physiologic roles in the body and are implicated in the pathogenesis of sepsis. Traditional pharmacologic approaches of targeting GPCRs employed orthosteric ligands, which has a number of shortcomings. Novel pharmacologic approaches can target GPCR signaling intracellularly through the use of aptamers, intrabodies and pepducins. This has opened a fresh avenue of pharmacological possibilities that were not previously feasible with traditional methods of drug discovery. On the other hand, numerous clinical trials in sepsis have failed to show a survival benefit for any particular drug or intervention. A number of reasons may partly account for the myriad number of failed trials in sepsis. Firstly, sepsis is a heterogeneous syndrome caused by a wide spectrum of diverse infectious entities. The traditional approach of enrolling patients in sepsis trials who meet the broadly defined criteria of the sepsis syndrome is likely contributing to failed trials. The description of various phenotypes and molecular endotypes of sepsis have provided new insights and potential opportunities for precision medicine in sepsis. Secondly, the natural course of sepsis is biphasic in that early sepsis is characterized by a hyperinflammatory response followed by a delayed state of immuno-paralysis. Interventions that are tested in clinical trials need to be tailored in a time-sensitive manner. Certain interventions that may be beneficial in the hyperinflammatory phase of sepsis may be detrimental in the immuno-paralysis phase of sepsis and vice versa. Moreover, the hyper- or hypoactive immune response in sepsis is highly heterogeneous and accurate diagnostics that dynamically capture the state of the immune response in real-time are crucial for developing tailored, individualized therapies. Thirdly, not all preclinical models of sepsis accurately parallel the pathogenesis of sepsis in humans, and certain interventions that appear beneficial in preclinical studies have been found to have no real benefit in human trials. In this regards, the endotoxemia model of experimentally induced sepsis is particularly problematic and does not accurately recapitulate the true pathogenesis of polymicrobial sepsis. The CLP model of experimental sepsis is often considered the gold standard method for studying sepsis. To further address this issue, an expert consensus group has developed the MQTiPSS (Minimal Quality Thresholds in Preclinical Studies for Sepsis) guidelines for conducting pre-clinical studies in sepsis ( Osuchowski, et al., 2018 ). This will help to improve the quality of preclinical studies in sepsis and may provide better experimental models for testing pharmacotherapy in sepsis. Fourthly, multiple redundant pathways are concurrently activated in sepsis and testing singular interventions in traditional randomized controlled trials may be a reason for their failure. Consequently, newer trials should test a “cocktail” or package of multiple pharmacological interventions concomitantly. Finally, traditional design of randomized trials may not be the best method to test interventions in sepsis. Newer adaptive trial designs that incorporate Bayesian probabilities to modify and evolve the conduct of the trial as data is generated, may be better suited to identify beneficial interventions in sepsis. In this regard, the present review identified a number of GPCRs that may be potentially useful targets for pharmacotherapy in sepsis ( Table 5 ). Novel approaches that utilize pepducins, aptamers and intrabodies to target these GPCRs may provide an unprecedented opportunity for altering the trajectory of morbidity and mortality from sepsis.

Historically, the term “sepsis” was first used by Hippocrates (470–360 BC) to describe a disease-producing process by which flesh rotted and wounds festered . The modern concept of sepsis is an abnormal host response to infection that results in multi-organ dysfunction. This condition is associated with a high rate of mortality and morbidity with an estimated 5.3 million deaths per year being attributable to sepsis ( Adhikari, Fowler, Bhagwanjee, & Rubenfeld, 2010 ). In the United States, sepsis was estimated to account for more than $20 billion (5.2%) of the total hospital costs in the year 2011 ( Angus, et al., 2001 ). Sepsis is reported to have a prevalence of 12% in hospital-derived cohorts in the United States and may be even higher in lower-to-middle income countries ( Mayr, Yende, & Angus, 2014 ). Sepsis was made a global health priority by the World Health Organization in 2017 and a resolution was adopted by both the World Health Organization and the World Health Assembly to improve the diagnosis, treatment and prevention of sepsis ( Reinhart, et al., 2017 ). According to the Third International Consensus Definitions (Sepsis-3) ( Singer, et al., 2016 ), sepsis is defined as life-threatening end-organ dysfunction caused by a dysregulated host response to infection. Organ dysfunction in patients with sepsis is assessed using the Sequential Organ Failure Assessment (SOFA) score ( J.-L. Vincent, et al., 1996 ). Presence of an infection coupled with an increase in SOFA score by 2 points is sufficient to diagnose sepsis. In a small subset of patients, sepsis results in profound cellular and circulatory abnormalities severe enough to substantially increase mortality; these patients are deemed to have septic shock and have a reported in-hospital mortality of 40–60% ( Cecconi, Evans, Levy, & Rhodes, 2018 ). Despite significant advances in our understanding of the pathobiology of sepsis, therapeutic interventions have not translated into tangible differences in overall outcomes for patients. The mortality from sepsis remains at 25%, while that for patients in septic shock is even higher ranging from 30% to 50% in most series ( Fleischmann, et al., 2016 ). This may well be a conservative estimate as evidence suggests that mortality from sepsis and septic shock is even higher in the developing world where close to 87% of the world’s population lives ( Reinhart, et al., 2017 ). Scarcity of data from the developing world makes it difficult to make accurate epidemiological estimations regarding the incidence and mortality of sepsis and septic shock. However, it is estimated that mortality from sepsis is likely around 50–60% and that for septic shock is close to 80% in the developing world ( Jawad, Lukšić, & Rafnsson, 2012 ). In this review, we summarize our current understanding of the pathology of sepsis followed by a review of prior unsuccessful therapeutic strategies in sepsis. We then turn our attention to G-protein coupled receptors (GPCRs) and discuss those that could be potentially targeted in sepsis. Finally, we shed light on novel methods of intracellularly targeting those receptors in sepsis.

Of the ~800 GPCRs identified within the human proteome, about ~369 are implicated in the pathophysiology of various diseases and represent potential targets for pharmacotherapy. The current market share of drugs targeting GPCRs is estimated to be approximately 40%, even though the overall number of GPCRs targeted by current drugs is ~30 ( Wise, Jupe, & Rees, 2004 ). This suggests that the true potential of targeting GPCRs has not been fully realized to this date. The general structure and features of signal transduction through GPCRs have been discussed previously in section 4 . As mentioned before, GPCRs can couple to a limited number of G proteins and transduce signals through these proteins. Despite marked structural diversity at the intracellular regions, GPCRs couple to only about 18 different subtypes of G proteins (belonging to the four major families i.e. G s , G i , G q and G 12/13 ). These G proteins can be activated by a wide variety of different cationic α -helical structures, although the selectivity of GPCR coupling to particular G proteins seems to be encoded by a combination of two functional domains at the intracellular regions ( Hedin, Duerson, & Clapham, 1993 ). The first activation domain can activate multiple G protein subtypes, while the second selectivity domain restricts coupling to a single signaling pathway. Slight changes in the structural conformation at these functional domains can affect signal transduction through the GPCR by altering G protein selectivity. This helps to explain how GPCRs translate a diverse array of extracellular inputs into a limited number of intracellular biochemical signals ( S. K.-F. Wong, 2003 ). Studies employing multiscale computational approaches have revealed that GPCRs can exist in numerous inactive and active conformations, and the balance between these conformations is altered upon binding of orthosteric ligands by changing the thermodynamic stability of the molecular system ( Niesen, Bhattacharya, & Vaidehi, 2011 ). Consequently, traditional pharmacologic approaches to GPCR targeting have focused on designing orthosteric agonists and antagonists that alter the conformational state of the receptor. This has proven challenging in view of the fact that many GPCRs have a high degree of sequence and structural homology, particularly within the same receptor. Moreover, orthosteric modulators that tonically inhibit or stimulate signaling are more likely to have side effects. Allosteric modulators that bind to less conserved regions of GPCRs may be more selective in their action and potentially have fewer adverse effects by modulating a narrow range of physiologic responses ( X. Liu, et al., 2017 ). Fortunately, recent studies have reported the crystal structures of a few GPCRs complexed with allosteric modulators bound to the cytoplasmic side of the receptor ( Oswald, et al., 2016 ; Zheng, et al., 2016 ). These studies have improved our understanding of the structure and mechanistic basis for intracellular signal transduction through GPCRs, thereby providing a potential framework for targeting GPCRs intracellularly (see Figure 3 ). In the following lines, we briefly discuss some of the strategies that may be potentially useful for targeting GPCRs intracellular. Pepducins are a group of cell-penetrating lipidated peptides that can be used to modulate the activity of GPCRs by acting at the intracellular receptor–effector interface ( P. Zhang, Covic, & Kuliopulos, 2015 ). These molecules have a peptide backbone that is typically derived from a sequence of the cytoplasmic part of the target GPCR and a lipid moiety is conjugated with the peptide to make the molecule permeable. Because of this lipid moiety, pepducins can penetrate through the intact plasma membrane and anchor in the cytosolic interface, thereby modulating the interaction of the target receptor with its effector G proteins (as illustrated in Figure 4 ) ( Carr & Benovic, 2016 ). Pepducins are different from conventional GPCR agonists and antagonists in that they are allosteric modulators and affect receptor–effector G protein interactions at the intracellular interface. For instance, ICL3–9, a pepducin based on the third intracellular loop of the β 2 -adrenergic receptor, can stimulate interactions between the β 2 -adrenergic receptor and G α s proteins. However, the receptor conformation induced by ICL3–9 is different from the activated conformational state induced by isoproterenol—the orthosteric agonist for β 2 -adrenergic receptor ( Carr, et al., 2014 ). Moreover, intracellular activation of G proteins by pepducins is typically not subject to desensitization by β -arrestin or GRKs. In fact, certain pepducins can directly stimulate or inhibit G α proteins independent of GPCRs ( Carr, et al., 2016 ). Pepducins can also act as biased agonists or antagonists of one particular class of G α proteins. For example, the CXCR4 pepducin ATI-2431, derived from the first intracellular loop of CXCR4, selectively activates G α i signaling but not G α 12/13 signaling ( Quoyer, et al., 2013 ). Likewise, the PAR2 pepducin P2pal-18S, based on the third intracellular loop of PAR2, was strongly biased towards inhibiting PAR2-G α q and PAR2-G α i signaling, but had no effect on PAR2-ligand activated endocytosis ( Sevigny, et al., 2011 ). Although the precise details of how pepducins affect GPCR–G protein interactions remain to be elucidated, a number of pepducins have been designed against a variety of GPCRs. F2Pal16 is a pepducin that acts as an agonist of FPR2. This pepducin is composed of a peptide that has a sequence identical to the third intracellular loop of FPR2 and has a palmitic acid (16-carbon) attached to the peptide ( Forsman, et al., 2013 ). F2Pal16 can activate FPR2 in phagocytes and transfected HL-60 cells, similar to conventional FPR2 agonists. Another pepducin, F1Pal16, was composed of a peptide with sequence identical to the third intracellular loop of FPR1 and linked to palmitic acid. Surprisingly, this pepducin was found to have no effect on FPR1 signaling, but inhibited FPR2-mediated cellular responses ( Winther, Gabl, Welin, Dahlgren, & Forsman, 2015 ). A shorter variant pepducin, F2Pal10, was shown to act as a partial agonist for the FPR2 receptor, but acted as a full agonist for cross-talk triggered FPR2 activation mediated by platelet activating factor and ATP (P2Y 2 ) receptors ( Gabl, et al., 2014 ). PZ-128 (P1pal-7) is a pepducin based on the third intracellular loop of PAR1 that can inhibit the interaction of PAR1 with its effector G proteins ( Leger, et al., 2006 ). PZ-128 is highly efficacious in blocking PAR1-dependent platelet aggregation as it inhibits p38 MAPK activation and blocks G α 12/13 -Rho kinase activation. In experimental studies, PZ-128 had an onset of action within 15 minutes of intravenous administration and suppressed PAR1-mediated platelet aggregation in guinea pigs and baboons ( P. Zhang, et al., 2012 ). PZ-128 was the first pepducin to be tested in a human clinical trial ( NCT01806077 ) and it was found to have a rapid, specific and dose-dependent effect on PAR1-mediated platelet aggregation ( Gurbel, et al., 2016 ). Moreover, PZ-128 was also shown to reduce atherosclerotic plaque burden in patients with coronary artery disease by inhibiting MMP1-PAR1 signaling ( Rana, et al., 2018 ). Larger clinical trials assessing the safety and efficacy of PZ-128 in coronary artery disease are currently being planned. Given that both thrombin- and MMP1-mediated PAR1 activation is implicated in the pathogenesis of sepsis ( Tressel, et al., 2011 ), PZ-128 holds promise for use in patients with sepsis. Thrombin-mediated activation of PAR4 is mechanistically different from that of PAR1 and PAR1-mediated platelet aggregation is generally transient, unless additional signals from PAR4 or P2Y 12 receptors strengthen it ( Covic, Singh, Smith, & Kuliopulos, 2002 ). Consequently, a pepducin was designed based on the third intracellular loop of PAR4, namely P4pal-10. P4pal-10 was found to be a dual inhibitor of PAR1 and PAR4, and inhibited 85% of human platelet aggregation in response to thrombin ( Covic, Misra, Badar, Singh, & Kuliopulos, 2002 ). Given that PAR1 and PAR4 form heterodimers in human platelets, pepducins with dual inhibitory effects on PAR1 and PAR4 may also be of therapeutic value for treatment of sepsis ( Leger, et al., 2006 ). PZ-235 (P2pal-18S) is a pepducin designed against the PAR2 and is based on the third intracellular loop of PAR2. PZ-235 acts as a full antagonist of PAR2 and was evaluated for its protective effects in a mouse model of nonalcoholic steatohepatitis ( Shearer, et al., 2016 ). PZ-235 significantly suppressed hepatic fibrosis, inflammatory cytokine release, reactive oxygen species production, stellate cell proliferation, and nonalcoholic fatty liver disease activity scores by 50–100%. Given that PAR2 plays an important role in the pathogenesis of atopic dermatitis, PZ-235 was also evaluated in laboratory models of atopic dermatitis ( Barr, et al., 2019 ). PZ-235 significantly suppressed total leukocyte and T-cell infiltration, epidermal thickness and total lesion severity scores in filaggrin-deficient mice exposed to dust mite allergens. Moreover, PZ-235 also inhibited PAR2-mediated expression of inflammatory factors by human mast cells and keratinocytes. Given the role played by PARs in the pathogenesis of sepsis, targeting of PARs by pepducins in patients with sepsis may be potentially beneficial. Chemokine receptors have also been targeted successfully by pepducins in experimental studies. CXCR1 and CXCR2 receptors share an identical third intracellular loop, and the pepducin x1/2pal-i3, derived from the third intracellular loop, targets both of these receptors. In human neutrophils, x1/2pal-i3 completely inhibited IL-8–induced calcium influx and blocked neutrophil migration toward chemotactic gradients of IL-8 ( Kaneider, Agarwal, Leger, & Kuliopulos, 2005 ). Another pepducin x1/2LCA-i1, derived from the first intracellular loop of CXCR1 and CXCR2, blocked chemotactic responses of human and mouse neutrophils by inhibiting CXCR1-and CXCR2-mediated signaling ( Kaneider, et al., 2005 ). When tested in the CLP model of sepsis in mice, both x1/2pal-i3 and x1/2LCA-i1 pepducins afforded marked protection against death from sepsis ( Kaneider, et al., 2005 ).These results suggest that targeting of chemokine receptors by pepducins may be a potential therapeutic strategy for patients with sepsis. Pepducins targeting CXCR4 have also been designed. ATI-2341 is a pepducin based on the first intracellular loop of CXCR4 and induced CXCR4-dependent signaling and chemotaxis in leukocytes ( Tchernychev, et al., 2010 ). In mice and cynomolgus monkeys, AT-2341 dose-dependently increased the release of granulocyte-macrophage progenitor cells from the bone marrow. Conversely, the pepducin x4pal-i1, also based on the first intracellular loop of CXCR4, inhibited CXCR4 signaling and blocked CXCL12-mediated migration of lymphocytes ( O’Callaghan, et al., 2012 ). These studies suggest that targeting of chemokine receptors through pepducins is feasible and may be potentially useful in the treatment of sepsis. S1P receptors play important roles in the pathogenesis of sepsis and are potential therapeutic targets as discussed previously in section 4.8 . KRX-725 is a pepducin that activates S1P 3 receptors and is based on the second intracellular loop of the S1P 3 receptor ( Licht, Tsirulnikov, Reuveni, Yarnitzky, & Ben-Sasson, 2003 ). KRX-725 causes activation of S1P 3 receptors on mouse aortic rings, which induces Gi-dependent ERK activation and endothelium-dependent vasodilation mediated by nitric oxide. Severino and colleagues synthesized a pepducin (peptide sequence Myr-GRPYDAN-NH 2 ) that antagonized S1P 3 receptors ( Severino, et al., 2013 ). Given the role played by S1P in sepsis, pepducins targeting S1P receptors may be potentially useful for patients with sepsis. A number of peculiarities regarding pepducins should be noted here. Firstly, it has been observed that pepducins are not entirely specific for their “designated” target receptor ( Winther, et al., 2017 ). As an example, two pepducins (P2Y 2 PalIC2 and P2Y 2 PalIC3) containing sequences from the second and third intracellular loops (respectively) of the ATP (P2Y 2 ) receptor were found to be agonists for FPR2 on neutrophils ( Gabl, et al., 2016 ). Interestingly, this phenomenon involved cross-talk between ATP bound-P2Y 2 receptor and P2Y 2 PalIC2 bound-FPR2 receptor. Likewise, a pepducin designed as an agonist for the CXCR4 receptor, ATI-2341, was found to have stimulatory effects on neutrophils through activation of FPR2 ( Holdfeldt, Winther, Gabl, Dahlgren, & Forsman, 2016 ). Secondly, small substitutions in the amino acid sequence of certain pepducins leads to complete abrogation of their targeting activity ( Gabl, et al., 2016 ). Additionally, FPR2 targeted pepducins have no effect on FPR1 despite significant similarity in the amino acid sequences of intracellular loops of FPR1 and FRP2 ( He & Ye, 2017 ). These observations suggest that the intracellular targeting of GPCRs by pepducins may be related to their ability to target specific dimeric or oligomeric forms of the target GPCR. The precise details of how pepducins intracellularly interact with their cognate receptors have not been fully elucidated ( Carr & Benovic, 2016 ). Despite this, pepducins hold great promise for targeting GPCRs as these cell-penetrating peptides can access receptor conformations that are not otherwise accessible by orthosteric targeting. The interaction of small molecule allosteric modulators with GPCRs is most well-described for chemokine receptors. The structure of CCR9 crystallized in complex with vercirnon has been described, which revealed that the binding site of vercirnon (CCX282) is on the cytoplasmic face of the receptor ( Oswald, et al., 2016 ). Vercirnon has been shown to be efficacious for treatment of inflammatory bowel disease in phase II clinical trials, and is currently being tested in phase III clinical trials ( Wendt & Keshav, 2015 ). Likewise, the crystalline structure of CCR2 complexed with the allosteric modulator CCR2-RA-[ R ] has also been described ( Zheng, et al., 2016 ). CCR2-RA-[ R ] binds to a highly druggable pocket that is the most intracellular allosteric site observed in any class A GPCR. Apart from chemokine receptors, the crystal structure of the β 2 -adrenergic receptor complexed with an allosteric inhibitor cmpd-15A has also been published ( X. Liu, et al., 2017 ). These studies describing the crystalline structures of GPCRs complexed to allosteric modulators have provided important insights with implications for future drug discovery. One important observation is the fact that these allosteric antagonists appear to act through steric hindrance as they preclude the interaction of the GPCR with G-proteins and β -arrestin. Another important observation is that the intracellular ligand-binding pocket harbors a balanced combination of polar and hydrophobic moieties, which makes it a potentially druggable target. Lastly, and most importantly, these crystal structures provide high-resolution details of the intracellular ligand-binding pockets, which may serve as a platform for virtual ligand screening for identifying other allosteric modulators. Nanobodies is the name given to the variable fragment of single-chain antibodies. Human immunoglobulins are composed of heavy and light chains, and differ from single-chain antibodies that are typically produced by members of the camelid family. Pardon and colleagues generated a set of nanobodies against the β 2 -adrenergic receptor by immunizing Ilamas with an agonist-bound purified β 2 -adrenergic receptor ( Pardon, et al., 2018 ). Staus et al. expressed a set of these nanobodies in the cytoplasm of HEK293 cells as intrabodies and assessed their effects on β 2 -adrenergic receptor-mediated signaling ( Staus, et al., 2014 ). Many of the intrabodies were found to inhibit cAMP accumulation, β -arrestin recruitment, GRK-mediated receptor phosphorylation, and/or receptor endocytosis with a preference for either active (agonist occupied) or inactive (antagonist occupied) conformation of β 2 -adrenergic receptors. The ability of these intrabodies to fine-tune ligand-induced GPCR trafficking and signaling opens a unique avenue of opportunities to intracellularly target GPCRs in a precise manner. However, the main barrier to their use in clinical settings would be the delivery or expression of these intrabodies in live cells. A number of putative methods including self-internalizing peptides, cationic liposome encapsulation and nanoparticle-mediated delivery are currently being developed, but, these methods are not yet ready for prime time ( Cardinale, Merlo, Giunchedi, & Biocca, 2014 ). The word “aptamer” comes from the Latin word aptus meaning fit and the Greek word meros meaning part. Aptamers are actually single-stranded molecules of RNA or DNA that can bind to specific interfaces on proteins with high specificity and affinity that is determined by their secondary and tertiary structures ( Nimjee, White, Becker, & Sullenger, 2017 ). These oligonucleotides are typically produced by the SELEX (systematic evolution of ligands by exponential enrichment) method. This iterative method entails incubating a target protein with a large library of nucleic acid molecules and separating nucleic acid molecules that bind to the target protein. The bound RNA molecules are then amplified by real time-polymerase chain reaction and resulting DNA templates are transcribed. The new pool of nucleic acids is again incubated with the target protein and the whole cycle is repeated 8–12 times until an RNA pool with a high affinity for the target protein isolated. Sullenger and colleagues demonstrated that these aptamers could be isolated to virtually any target and may serve as potential therapeutic agents ( Sullenger, Gallardo, Ungers, & Gilboa, 1990 ). Since then, numerous aptamers have been designed against a variety of proteins and are in preclinical or clinical phases of development. Of note, an aptamer against vascular endothelial growth factor, pegaptanib, was approved by the FDA in 2000 for use in patients with wet age-related macular degeneration. Kahsai and colleagues employed the SELEX (systematic evolution of ligands by exponential enrichment) method to identify RNA aptamers that bind to allosteric sites of the β 2 -adrenergic receptor with nanomolar affinity ( Kahsai, et al., 2016 ). They started with an RNA library containing 10 15 unique sequences and used an iterative selection process employing next-generation sequencing and comparative bioinformatics to isolate candidate aptamers with desirable binding properties. These aptamers further underwent nine rounds of positive selection against unliganded and agonist-bound β 2 -adrenergic receptors in order to isolate high-affinity aptamers binding at structurally relevant sites. At the end of the selection process, the pool of final aptamers was able to stabilize unliganded and ligand-specific conformations of the β 2 -adrenergic receptor with nanomolar affinities. In particular, aptamers A1, A2 and A13 significantly inhibited agonist-induced cAMP accumulation. This study demonstrated that aptamers could be potentially developed as pharmacological agents for the modulation of GPCR-mediated signaling.

Sepsis is a dysregulated host response to infection and is initiated by hyper-activation of the innate immune system. Cells of the innate immune system (macrophages, monocytes, granulocytes, natural killer cells and dendritic cells [DCs]) continuously sample the local tissue environment at various microbial portals of entry. Multiple microbial products and their derivatives (pathogen-associated molecular patterns [PAMPs]) as well as endogenous danger signals (damage-associated molecular patterns [DAMPs]) can bind to cell surface receptors (pattern recognition receptors) on innate immune cells and trigger increased expression of pro-inflammatory cytokines. Simultaneous binding of multiple PAMPs and DAMPs to Toll-like receptors, mannose-binding lectin (MBL), scavenger receptors, nucleotide-binding oligomerization domain (NOD)-like receptors, retinoic acid-inducible gene (RIG)-like receptors, C-type lectin receptors and other receptors leads to complementary activation and amplification of multiple signaling pathways ( Gentile & Moldawer, 2013 ). Intracellular phosphorylation of Janus kinases, signal transduction and activators of transcription, mitogen-activated protein kinases (MAPKs) and other intermediates leads to recruitment of interferon regulatory factor and nuclear factor-κB (NFκB) ( Kumar, 2018 ). Interferon regulatory factor increases the expression of interferon (IFN), while NFκB and activator protein 1 (AP-1) increase the transcription of tumor necrosis factor (TNF), interleukin (IL)-1, IL-12 and IL-18 (among other cytokines) as well as endothelial adhesion molecules ( Wiersinga, 2011 ). Irrespective of the initial repertoire of PAMPs and DAMPs implicated, redundant and complementary signaling pathways are activated in sepsis that converge on the enhanced expression of a common pool of pro-inflammatory cytokines. These cytokines have further diverse downstream effects including activation of complement and coagulation pathways, endothelial barrier dysfunction, alteration of cellular metabolism, and suppression of the adaptive immune system ( T. van der Poll & Opal, 2008 ). Mitochondrial dysfunction has also been identified as a core pathophysiologic feature of sepsis-induced organ dysfunction in more recent studies ( Joseph, et al., 2017 ). Chemokines (including IL-6, IL-8, IFN γ , CXC-chemokine ligand 10 [CXCL10], CC-chemokine ligand 2 [CCL2] and CC-chemokine ligand 3 [CCL3]) induce the chemotaxis and recruitment of phagocytes ( Schulte, Bernhagen, & Bucala, 2013 ). A shift in the endothelial expression of various procoagulant proteins (von Willebrand factor, thrombomodulin, tissue factor and activated protein C [APC]) results in the transformation of a healthy (anticoagulant) endothelium to a prothrombotic endothelium in sepsis ( Ince, et al., 2016 ). Moreover, internalization of the vascular endothelial (VE)-cadherin, as a consequence of pro-inflammatory protease activity leads to a leaky endothelium with increased vascular permeability. PAMPs and DAMPs can lead to activation of the complement cascade. The complement cascade is an integral part of the innate immune response and acts as a bridge between innate and acquired immunity. This system consists of a series of proteins that mediate responses to inflammatory triggers through a co-ordinated and sequential enzyme cascade, eventually leading to clearance of foreign cells through pathogen recognition, opsonization and lysis. The complement system also possesses anti-inflammatory functions in that it binds to immune complexes and apoptotic cells, and assists in their removal from the circulation. This important system is involved in the eradication of invading microbes, but, also contributes to the inflammatory response during sepsis. The complement cascade in humans can be activated through three distinct pathways (as illustrated in Figure 1 ): (a) the classical pathway; (b) the alternate pathway; and (c) the lectin pathway ( Lupu, Keshari, Lambris, & Coggeshall, 2014 ). C1q in the classical pathway acts as a pattern recognition receptor and can bind to PAMPs or DAMPs, thereby resulting in activation of the classical pathway. PAMPs or DAMPs can also activate the lectin pathway by binding to MBL or ficolins, which in turn can activate MBL-associated serine proteases and lead to the formation of C3 convertase . Cleavage of C3 by C3 convertase leads to the formation of C3a (an anaphylatoxin) and C3b (an opsonin). C3b then participates in the formation of C5 convertase , which cleaves C5 into C5a (another anaphylatoxin) and C5b. C5a is one of the most potent inflammatory peptides produced in the complement pathway and results in chemotaxis of phagocytic cells. C5a also amplifies the production of pro-inflammatory cytokines by innate immune cells and triggers the oxidative burst within neutrophils. Production of free radicals by neutrophils leads to widespread tissue injury and contributes to the multi-organ dysfunction typical of sepsis. Moreover, C5a can cause endothelial activation, vasodilation and activation of the coagulation cascade that may initiate disseminated intravascular coagulation (DIC). DIC further compromises tissue perfusion through the formation of microthrombi and aggravates multi-organ dysfunction. Complement proteins C3 and C5 can also be cleaved by thrombin and plasmin (produced in the coagulation cascade) to activate the complement cascade (so-called “extrinsic protease pathway”), and in turn, complement proteins C3a and C5a can bind to their receptors on platelets to regulate platelet activation and thrombus formation. This cross-talk between the coagulation and complement cascades represents an important link between coagulation and inflammation in the setting of sepsis, and both processes can perpetuate each other leading to worsening multi-organ dysfunction ( Foley & Conway, 2016 ; Marcel Levi & van der Poll, 2017 ). The normal endothelium under homeostatic conditions has an anticoagulant property to allow smooth blood flow and permit the exchange of various solutes between the blood, endothelium and interstitium ( Hack & Zeerleder, 2001 ). Integrity of the endothelial barrier is essential to maintain a state of homeostasis as the underlying subendothelial collagen is highly pro-thrombotic. This barrier is maintained consistently by a number of features including the glycocalyx, actin cytoskeleton and intercellular tight junctions maintained by adhesion molecules (VE cadherins) ( Aird, 2001 ). Nitric oxide is synthesized in endothelial cells by the activity of nitric oxide synthase and plays an important role in maintaining the physiologic barrier function of the endothelium ( J. L. Vincent, Zhang, Szabo, & Preiser, 2000 ). Endothelial cells also regulate the composition of the interstitium—the milieu for different types of cells across various tissues—by maintaining a dynamic and interactive interface between the blood and the interstitium ( W. L. Lee & Slutsky, 2010 ). In patients with sepsis, release of pro-inflammatory mediators increases the expression of selectins and integrins on endothelial cells, which promotes leukocyte adhesion and migration ( Ince, et al., 2016 ). Injury to the glycocalyx, actin cytoskeleton and intercellular tight junctions results in increased vascular permeability and a leaky endothelium ( Ait-Oufella, Maury, Lehoux, Guidet, & Offenstadt, 2010 ). This allows the interaction of tissue factor with circulating factor VII, thereby initiating the tissue factor pathway of the coagulation cascade. Likewise, subendothelial collagen fibers bind to von Willebrand factor, which in turn binds platelets (through the glycoprotein Ib/IX receptor) and leads to thrombosis ( M. Levi, ten Cate, & van der Poll, 2002 ). Additionally, endothelial dysfunction leads to systemic vasodilatation and loss of intravascular proteins and plasma fluids into the extravascular space with disruption of capillary flow and tissue hypoperfusion ( Fernandez-Sarmiento, Salazar-Pelaez, & Carcillo, 2020 ). Nitric oxide production markedly increases in sepsis as a result of increased expression of inducible nitric oxide synthase and contributes to endothelial dysfunction ( J. M. Wong & Billiar, 1995 ). Widespread activation of the coagulation cascade leads to the formation of platelet microthrombi in the circulation (DIC), which further compromises tissue perfusion and manifests as septic shock. As mentioned before, increased vascular permeability and exposure of subendothelial collagen can lead to binding of von Willebrand factor, which in turn can bind platelets. Additionally, many pro-inflammatory cytokines and chemical mediators released in sepsis can lead to activation of platelets and the coagulation cascade. Release of tissue factor from the endothelium in response to sepsis can activate the tissue factor pathway and result in formation of fibrin. Glycoprotein IIb/IIIa receptors on platelets can bind fibrin and cross-link multiple platelets together. Formation of fibrin strands and their cross-linking along with aggregation of platelets leads to the formation of a stable thrombus. Moreover, formation of thrombin through the tissue factor pathway leads to activation of the contact pathway, which further amplifies the coagulation process. There is significant “cross-talk” between the coagulation, kinin and complement cascades—activation of the contact pathway produces factor XIIa, which can form kallikrein. Kallikrein not only produces bradykinin but can also amplify the coagulation cascade by producing more factor XIIa. Likewise, kallikrein can cleave plasminogen to form plasmin, which can activate the complement cascade. Many coagulation factors (specifically, factors XIa, Xa and IXa) can also cleave complement proteins C3 and C5, thereby activating the complement cascade ( Amara, et al., 2010 ). Conversely, complement activation can lead to cleavage of thrombinogen to form thrombin, further amplifying the coagulation cascade. Thrombin itself can amplify the overall inflammatory response by multiple mechanisms, such as induction of tissue ischemia (through thrombosis), production of down-stream mediators (for instance, APC) and signaling through protease-activated receptors (PARs) which leads to release of inflammatory mediators and, recruitment and chemotaxis of inflammatory cells. Platelets themselves are a source of P-selectin, which serves as an attachment for neutrophils. Neutrophils in turn can form neutrophil extracellular traps (NETs) through the process of NETosis, which provides a scaffold for further clot formation ( Delabranche, et al., 2017 ). Inflammation is closely linked to coagulation and both processes can amplify each other in the setting of tissue injury. When such tissue injury is localized, these processes can act synergistically to wall off the site of injury and protect the rest of the host from infection. However, in the setting of sepsis, widespread activation of coagulation and inflammatory processes lead to DIC and multi-organ failure. Sepsis is a dysregulated host response to infection that results in multi-organ dysfunction. Traditionally, this dysregulation has been perceived as a hyper-activation of the innate immune system with a cascade of pro-inflammatory cytokines and mediators that result in uncontrolled inflammation. However, a whole body of evidence suggests that immunosuppression occurs both in the early and delayed phases of sepsis ( Gentile, et al., 2012 ). Studies of sepsis survivors have shown that patients who survive the initial “hyper-inflammatory” phase of sepsis go on to develop a prolonged state of “immune paralysis” and chronic inflammation (termed persistent inflammation/immunosuppression and catabolism syndrome). This delayed phase of sepsis is associated with profound changes in functioning of the immune system ( Rubartelli & Lotze, 2007 ; Walton, et al., 2014 ) including a predominance of immature neutrophils, recruitment of myeloid-derived suppressor cells, peripheral lymphopenia, increased proportion of T reg cells (CD4 + /CD25 + /FOXP3 + phenotype), impaired antimicrobial activity of innate immune cells, preferential differentiation to the macrophage M2 phenotype, elevated levels of anti-inflammatory cytokines (chiefly IL-10 and transforming growth factor- β ) and reduced expression of MHC (major histocompatibility complex)-II molecules on DCs ( Boomer, et al., 2011 ; Taneja, Sharma, Hallett, Findlay, & Morris, 2008 ). Experimental studies have also demonstrated increased expression of programmed death ligand 1 (PD-L1) on antigen presenting cells and stromal cells, which can interact with the programmed death protein 1 (PD1) receptor on T cells, thereby leading to broad T cell anergy ( Drewry, et al., 2014 ). Similarly, studies from patients with sepsis identified profound apoptosis of DCs, T cells and B cells ( Hotchkiss, et al., 1999 ). In fact, the degree of apoptotic loss of lymphocytes has been shown to be correlated with the severity of sepsis ( Drewry, et al., 2014 ). Pharmacological approaches that block the interaction of PD-L1 with PD1 and reduce lymphocytic apoptosis have been shown to be beneficial in experimental models of sepsis ( Patil, Guo, Luan, & Sherwood, 2017 ). Immune checkpoint inhibitors that block PD-L1 have shown promising results in cancer immunotherapy trials and hold great promise for use in the treatment of sepsis ( van Ton, Kox, Abdo, & Pickkers, 2018 ). Sepsis is known to be an extremely heterogeneous condition with variations in the type and severity of host response depending on the repertoire of PAMPs and DAMPs implicated in its pathogenesis. This poses significant challenges in designing randomized trials and assessing response to various therapeutic modalities. Consequently, the importance of delineating accurate nosology for designing personalized therapies tailored to the individual patient has been recognized for long. In 2017, the MARS (Molecular Diagnosis and Risk Stratification of Sepsis) consortium published a study describing four molecular endotypes of sepsis (termed MARS1, MARS2, MARS3 and MARS4) based on array-based transcriptomics analysis ( Scicluna, et al., 2017 ). Using a 140-gene expression signature, patients were reliably stratified into one of the four molecular endotypes. When these endotypes of sepsis were combined with clinical data (APACHE [Acute Physiology and Chronic Health Evaluation] scores), they provided robust predictions of 28-day mortality risk. Similar to the MARS consortium study, Sweeney and colleagues identified three distinct clusters of sepsis across multiple datasets using unsupervised machine learning algorithms of transcriptomics data ( Sweeney, et al., 2018 ); the authors termed these clusters as the “Inflammopathic”, “Adaptive” and “Coagulopathic” subtypes of sepsis. The “Inflammopathic” subtype was associated with activation of the innate immune system and higher mortality rate, while the “Adaptive” subtype was associated with activation of the adaptive immune system and a lower mortality rate. The different molecular subtypes of sepsis described in these studies are useful in providing a unifying framework for understanding the molecular heterogeneity of sepsis and applying precision medicine approaches to sepsis. With the discovery of various molecular endotypes of sepsis, efforts were made to describe clinical phenotypes of sepsis that could be identified using routine clinical parameters (such as vital signs and laboratory investigations). Four different clinical phenotypes of sepsis were recently described in the literature ( Seymour, et al., 2019 ). These phenotypes of sepsis (named α , β , γ and δ ) were derived from multiple large datasets by Seymour and colleagues using unsupervised machine learning techniques—most notably, clustering. These phenotypes were associated with mortality and were unique in their defining characteristics (29 clinical variables, such as vital signs and laboratory parameters) when compared to the commonly used severity scales for sepsis (SOFA and APACHE scores). In-hospital mortality for α , β , γ and δ phenotypes were 2%, 5%, 15% and 32% respectively. The recognition of molecular endotypes and clinical phenotypes of sepsis highlighted the importance of considering sepsis as a heterogeneous syndrome (constellation of signs and symptoms) rather than a single disease entity. Inaccurate and vague nosology for a heterogeneous clinical syndrome results in dumping of multiple different pathologic entities into a single basket group. This one-size-fits-all approach partly accounts for the myriad number of negative clinical trials in sepsis as discussed in the next section.