The Regulation of Neutrophil Migration in Patients with Sepsis: The Complexity of the Molecular Mechanisms and Their Modulation in Sepsis and the Heterogeneity of Sepsis Patients

Sepsis is defined as life-threatening organ dysfunction caused by a dysregulated host response to infection. Common causes include gram-negative and gram-positive bacteria as well as fungi. Neutrophils are among the first cells to arrive at an infection site where they function as important effector cells of the innate immune system and as regulators of the host immune response. The regulation of neutrophil migration is therefore important both for the infection-directed host response and for the development of organ dysfunctions in sepsis. Downregulation of CXCR4/CXCL12 stimulates neutrophil migration from the bone marrow. This is followed by transmigration/extravasation across the endothelial cell barrier at the infection site; this process is directed by adhesion molecules and various chemotactic gradients created by chemotactic cytokines, lipid mediators, bacterial peptides, and peptides from damaged cells. These mechanisms of neutrophil migration are modulated by sepsis, leading to reduced neutrophil migration and even reversed migration that contributes to distant organ failure. The sepsis-induced modulation seems to differ between neutrophil subsets. Furthermore, sepsis patients should be regarded as heterogeneous because neutrophil migration will possibly be further modulated by the infecting microorganisms, antimicrobial treatment, patient age/frailty/sex, other diseases (e.g., hematological malignancies and stem cell transplantation), and the metabolic status. The present review describes molecular mechanisms involved in the regulation of neutrophil migration; how these mechanisms are altered during sepsis; and how bacteria/fungi, antimicrobial treatment, and aging/frailty/comorbidity influence the regulation of neutrophil migration.


Introduction
Sepsis is defined as life-threatening organ dysfunction caused by a dysregulated host response to infection, i.e., the host response to the infection causes organ damage/ failure [1,2]. The organ functions are evaluated by the Sequential/sepsis-related Organ Failure Assessment (SOFA) score that is based on scoring respiration, coagulation, liver function, circulation, consciousness, and renal function. Finally, septic shock is observed when the circulatory dysfunctions are severe enough to substantially increase mortality; these patients can be clinically identified by the fact that they require vasopressor therapy and by their increased serum lactate levels despite adequate volume resuscitation [1,2]. Septic shock patients have a hospital mortality exceeding 40%.

Formylated Peptide Receptors and Their Binding of Bacteria-Derived and Endogenous N-Formylated Peptides and Lipopeptides
Neutrophils can respond to exogenous N-formylated peptides originating from various bacteria and to endogenous peptides derived from damaged cells (e.g., mitochondrial proteins); such peptides can be recognized by three different receptors and the formylated peptide receptors (FPR) 1 and FPR2 which are both expressed by neutrophils [7,36]. FPR1 and FPR2 bind structurally diverse ligands with either agonistic or antagonistic effects [37]. Agonistic ligation of FPR1 increases neutrophil chemotaxis [38]. These formylated peptides are regarded as end target chemoattractant, i.e., when the neutrophils experience opposing chemotactic signaling, they will preferentially migrate along the formylated peptide gradients [7,36]. Intracellular signaling downstream to the FPRs involves various pathways including the p38 kinase and the Rac2 GTPase, and at high concentrations, extracellular signal-regulated kinase 1 (ERK1) and ERK2 also [39,40].
FPR1 can also bind endogenous non-formylated ligands that include peptides derived from the neutrophil mediator cathepsin G and annexin A1/annexin A4 [41,42]. The

Formylated Peptide Receptors and Their Binding of Bacteria-Derived and Endogenous N-Formylated Peptides and Lipopeptides
Neutrophils can respond to exogenous N-formylated peptides originating from various bacteria and to endogenous peptides derived from damaged cells (e.g., mitochondrial proteins); such peptides can be recognized by three different receptors and the formylated peptide receptors (FPR) 1 and FPR2 which are both expressed by neutrophils [7,36]. FPR1 and FPR2 bind structurally diverse ligands with either agonistic or antagonistic effects [37]. Agonistic ligation of FPR1 increases neutrophil chemotaxis [38]. These formylated peptides are regarded as end target chemoattractant, i.e., when the neutrophils experience opposing chemotactic signaling, they will preferentially migrate along the formylated peptide gradients [7,36]. Intracellular signaling downstream to the FPRs involves various pathways including the p38 kinase and the Rac2 GTPase, and at high concentrations, extracellular signal-regulated kinase 1 (ERK1) and ERK2 also [39,40].
FPR1 can also bind endogenous non-formylated ligands that include peptides derived from the neutrophil mediator cathepsin G and annexin A1/annexin A4 [41,42]. The same is true for FPR2 which is more promiscuous and can bind formylated peptides as well as serum amyloid A, β-amyloid, annexin A1, humanin, and possibly lipoxin A 4 [41,42]. FPR2 can mediate both pro-and anti-inflammatory effects depending on the ligand, and various ligands seem to differ with regard to their effects on neutrophil migration [37,41]. Possible explanations for these ligand-dependent variations are the different dimerization states of the receptor, the altered receptor conformation, and/or differences in downstream signaling depending on the ligand [41]. Lipopeptides are short endogenous or microbial-derived peptides with a fatty acid linked at the N-terminal end [37]. These molecules have both activating and inhibitory effects on G-protein-coupled receptors, and they can modulate FPR-2-initiated intracellular signaling [37]. Due to their lipid domain, the molecules pass through the cell membrane and bind to the cytosolic parts of receptor complexes. Their protein part is usually identical or similar to one of the intracellular loops of the targeted receptor, although the receptor specificity is not always complete, e.g., a CXCR4 targeting lipoprotein was also found to target FPR2 [41][42][43]. This last observation suggests that lipopeptides represent a mechanism for crosstalk between various chemotactic mechanisms [43].

Complement Factors
The complement factor C5 is important for neutrophil chemotaxis, and the proteolytic fragment C5a is the most potent complement chemoattractant [7,44]. C5a is formed by C5 cleavage in response to both classical, lectin, and alternative complement activation [7], and its chemotactic effect is mainly mediated through the C5a Receptor 1 (C5aR1) and its downstream mediator p38 [10,11]. Neutrophil activation through C5aR1 can also lead to the release of the chemoattractant leukotriene B 4 , i.e., C5aR1 crosstalks with lipid chemoattractants [45]. Finally, the role of complement factor C3 in inflammation is complex, but it does not seem to have any major effect on neutrophil migration [44].

The Interacting Lipid Chemoattractants Leukotriene B 4 and Platelet Activating Factor
Eicosanoids derived from arachidonic acid should be regarded as the main lipid chemoattractants [7]. Leukotriene B 4 is the strongest lipid chemoattractant for neutrophils; it is mainly released by myeloid cells and binds to the G-protein-coupled receptors leukotriene B 4 receptor 1 (LTB4R1) and 2 (LTB4R2) [46,47]. LTB4R1 is mainly expressed by neutrophils, and receptor ligation results in neutrophil polarization followed by migration [9]. The functional effects of LTB4R2 ligation are less well characterized [7].
Another lipid chemoattractant is platelet-activating factor (PAF) which is derived from phosphatidylcholine [7,48,49]. It binds to a specific G-protein-coupled receptor on the neutrophils and can thereby activate and polarize neutrophils through intracellular signaling involving protein kinase C (PKC) [39,50]. PAF can also induce the release of leukotriene B 4 by neutrophils [48], and there is an additional crosstalk between these two main chemotactic mediators through leukotriene B 4 modulation of intracellular signaling downstream to PAF receptors [51]. However, animal models suggest that PAF may also inhibit neutrophil chemotaxis in certain biological contexts [50].

Classification of Chemokines and Chemokine Receptors
The chemokines constitute a family of low molecular weight cytokines (8-12 kDa, 70-120 amino acid residues) with chemotactic activity and, based on their chemical structure, they are classified into two main subsets: CCL and CXCL chemokines [52,53]. These two subsets bind to CCR and CXCR chemokine receptors, respectively. Several of these receptors are promiscuous with regard to ligand binding, and several chemokines are promiscuous with regard to receptor binding. The CXCL chemokines can be further subclassified into chemokines with and without a glutamate-leucine-arginine (ELR) motif before the CXC motif. CXCL chemokines with the ELR motif are CXCL1-3 and CXCL5-8; these chemokines are most important for neutrophil chemotaxis and can bind to the CXCR1/CXCR2 receptors [52][53][54]. Finally, neutrophil recruitment can also be guided by CCR1 ligands (CCL3 and CCL4), whereas CCR2 (receptor for CCL2) is most important during chronic inflammation. Finally, collagen fragments can also have chemotactic activity by binding to the CXCR2 receptor [52,53].
CCR2, CXCR1, and CXCR2 chemokines show similarities in downstream intracellular signaling. The PI3K pathway is important together with the additional activation of phospholipase C and increased intracellular calcium levels [55]. Receptors CXCR1/CXCR2 and Their Signaling: Biased  Intracellular Signaling Is Initiated Both at the Ligand, Receptor,

and Signal Initiation Levels
The chemokine system is characterized by biased signaling. This means that either (i) the relative strength of the intracellular pathway activation by two receptors with similar cellular density differs even when binding the same ligand (i.e., different ligand affinity of the receptors) or (ii) the relative activation of various pathways downstream to a certain chemokine receptor depends on the binding ligand. CXCR1/CXCR2 show biased signaling, which can be due to various molecular mechanisms [52].
CXCR1 versus CXCR2 affinity of individual chemokines. All CXCR2 ligands bind to this receptor with a similar high affinity, whereas the CXCR1 affinity differs between ligands with CXCL8 showing the highest affinity [56]. Thus, the balance between CXCR1/CXCR2 signaling depends on the availability of various ligand in the microenvironment.
Chemokine receptors have two different ligand-binding sites. The chemokines interact with CXCR1 and CXCR2 through two different binding sites, i.e., the chemokine N-terminal loop can either bind to N-terminal receptor residues (site 1) or to extracellular loop residues (site 2). Furthermore, an initial Site-1 interaction causes conformational receptor changes that are essential for binding to Site 2; differences between two chemokines in regard to the affinity for the two binding sites may therefore lead to differences in downstream signaling [57,58].
Chemokine monomers and dimers differ in receptor affinity. Chemokines exist as monomers and dimers, and some of them even exist as high-order oligomers or heterodimers [52, [57][58][59]. Chemokine monomers show biological activity at the nanomolar level, but at higher concentrations, this varies. For example, the CXCL8 dimers show an altered receptor affinity with lower CXCR1 affinity but similar high CXCR2 affinity compared with the monomer [58][59][60]. Thus, the local chemokine concentration and the equilibrium/ratio between monomeric and dimeric states will influence the balance between CXCR1 and CXCR2 ligation/signaling at least for certain CXCL chemokines [59,61,62].
Modulation of chemokine ligation by glycosaminoglycan binding. Chemokines can bind to extracellular glycosaminoglycans, and for certain chemokines (i.e., CXCL1/5/7/8) this binding is stronger for the dimers than for the monomers [52]. Chemokine binding to glycosaminoglycans on endothelial cells is also involved in the regulation of extravasation.
Downstream receptor signaling through G-proteins versus β-arrestin. The intracellular signaling of chemokine receptors is initiated immediately downstream to the receptor by either G-proteins or β-arrestin [52]. G-protein-mediated signaling is rapid in onset and is followed by gradual waning, whereas β-arrestin signaling has a slower onset with sustained duration [63]. β-arrestin is additionally involved in termination of the G-protein initiated signaling through recruitment of adaptor proteins and the enhancement of endocytosis of the chemokine/receptor complex [52,59].
Biased downstream signaling due to ligation of the same receptor by different chemokines. Several examples illustrate that ligation of the same receptor by different chemokines leads to different downstream effects. First, CXCR2 ligation by CXCL1 leads to an influx of extracellular calcium, but this is not observed after CXCL8 ligation [64]. Second, CXCL8 bound to CXCR1 but not to CXCR2 primes neutrophils for superoxide release, and this priming effect is also observed for CXCR2 after CXCL1 and CXCL5 binding [65,66]. Third, both G-protein and β-arrestin are important for the induction of the dynamic actin remodeling that is fundamental for neutrophil migration and for the integrin activation needed for transendothelial migration, but the balance between G-protein and β-arrestin signaling varies [52,59,[67][68][69]. For example, CXCR2 ligation by CXCL6 is relatively G-protein biased whereas the CXCL8 seems to be relatively β-arrestin biased with regard to cytoskeletal effects compared with other CXCR2 ligands [70]. Finally, CXCL1 monomers are more active than dimers with regard to several functional effects (e.g., calcium mobilization, chemotaxis, and β-arrestin recruitment), whereas such differences are not observed between CXCL2 monomers and dimers [60]. Finally, monomers and dimers may also differ with regard to the initiation of CXCR1/2 internalization/recycling [60,71].
To summarize, binding of the same receptor (CXCR1 and/or CXCR2) by different ligands can influence the downstream regulatory signaling of neutrophil migration.

Other Receptors or Mediators Involved in the Regulation of Neutrophil Migration: Molecular Mechanisms That Function as Modulators of the Main Chemotactic Pathways
Several other receptors and soluble mediators are also involved in the regulation of neutrophil chemotaxis, including: • Toll-like receptors 2 and 4 (TLR2/TLR4). These receptors have both exogenous (e.g., microbial molecules) and endogenous agonists [74,75]. Ligation of TLR2 and TLR4 alters the neutrophil expression of CXCR1 and CXCR2 [76]. However, this effect of TLR4 may be age-dependent because it differs between newborns and adults [77]. TLR4 also increases neutrophil expression of immunostimulatory HLA-DR and the immunosuppressive T cell checkpoint molecule PD-L1 (see later Sections 9.4 and 10.1), but these two effects may differ between neutrophil subsets [78]; • TLR9. Ligation of this intracellular receptor in neutrophils by exogenous or endogenous ligands triggers CXCR2 downregulation [79], and TLR9 ligation in other immunocompetent cells seems to have additional indirect effects on neutrophil chemotaxis [80][81][82]; • NOD-like receptors. Several cytosolic nucleotide-binding oligomerization domains (NOD) and NOD-like receptors (NLRs) have been described. NOD2 and NLRP3 are expressed by neutrophils and ligation of both receptors facilitates migration [83]. These receptors can be activated by the peptidoglycans of most bacteria [84,85] Lysophosphatidylcholines. Lysophosphatidylcholines can modulate neutrophil migration/chemotaxis. These lipids mediate the direct effects on neutrophils by ligation of the G2A receptor, a member of the proton-sensing G-protein-coupled receptors [97]. The G2A ligation causes the recruitment of clathrin that is important for the signal transduction [97], modulates p38 signaling [98], and finally increases neutrophil expression of the αM/CD11b integrin and formylated peptide receptors [99]. In vivo animal studies have shown that these lipids can reduce neutrophil migration [100]; • Sphingolipids. Sphingolipids and sphingolipid metabolism are involved in the regulation of neutrophil migration [9,[101][102][103][104][105][106]. First, lactosylceramide forms lipid raft microdomains coupled with the Src family kinase Lyn in neutrophils, and it can thereby initiate lipid raft-mediated regulation of neutrophil polarization and migration [10,104]. Second, sphingosine 1-phosphate (Sph-1-P) inhibits CXCL8-and formylated peptideinduced chemotactic migration of human neutrophils [103]. Third, glycosphingolipids seem to support slow neutrophil rolling on endothelial cells mediated by E-selectin and the transition to firm neutrophil adherence [105]. Finally, the neutral sphingomyelinase (N-SMase) is a plasma membrane enzyme that converts sphingomyelin to ceramide; it preferentially distributes towards the leading edge of neutrophils and is important for proper orientation in chemotactic gradients [30,104,106]. Thus, sphingolipids influence neutrophil migration through various molecular mechanisms.

Neutrophils Priming during Chemotaxis Modulates Neutrophil Migration
Neutrophil priming means that a primary agonist enhances the response to a second activator/event [107,108]. Priming may be transient or more durable, and the mechanisms mediating the priming effect involve various signaling pathways (e.g., p38, ERK1/2, and PI3K kinase), transcriptional regulators including NFκB, increased antiapoptotic signaling, granule mobilization, and the expression of surface molecules including the αMβ2 integrin and CD66 [109]. The ligand affinity of integrins as well as the cellular shape/deformability and polarization can be altered by priming [109]. Thus, neutrophil priming can modulate molecular mechanisms involved in neutrophil migration.
Neutrophils can be primed before (i.e., in the circulation), during, or following transmigration, and priming mechanisms that facilitate neutrophil migration involve endothelial cells, extravascular stromal cells, extracellular matrix proteins [110][111][112][113], platelets [110,111], and other immunocompetent cells [112]. Furthermore, the priming agents include the soluble mediators C3a and C5a complement components, interferon γ, CXCL8, and tumor necrosis factor (TNF) α together with pathogen-derived agents such as formylated peptides, peptidoglycan, and Staphylococcus aureus toxins [107][108][109]114]. Thus, neutrophils can be primed during chemotaxis, and this priming is at least partly caused by the chemotactic agents. However, there seems to be a possibility for depriming and recovery of primed neutrophils, but this process is less well characterized [109].

The Coagulation System: Regulation of Neutrophil Migration by Protease Receptors
Both activation of the coagulation system and thrombocytopenia occurs in a large number of sepsis patients [21,[115][116][117]. Coagulation factors contribute to the regulation of inflammation through effects on protease-activated receptors 1-4 (PAR1-4) that can be activated by various proteases. Thrombin can activate PAR1/3/4, Granzyme A activates PAR1/2, cathepsin G activates PAR1/4, and matrix metalloprotease 1 (MMP1) activates PAR1 [118]. Several other proteases that are activated during coagulation and/or inflammation can also activate various PARs (e.g., plasmin, FVIIa, FXa, and C4a) [118]. The PARs are G-protein-coupled membrane receptors that are expressed by a wide range of cells involved in the regulation of inflammation and neutrophil migration, e.g., endothelial cells, platelets, and various leukocytes [119,120].
PARs are activated when thrombin binds to the receptor and cleaves the N-terminal domain to unmask a new N-terminus that acts as a ligand through its binding to the receptor body [120]. PAR2 agonists can then alter neutrophil expression of certain adhesion molecules (e.g., decreased L-selectin and increased α4β1 integrin), but at the same time these agonists also reduce the transendothelial migration of neutrophils [121]. This apparent discrepancy may be caused by indirect PAR effects, e.g., mediated by endothelial cells or platelets [121]. Furthermore, PAR2 agonists can also increase the neutrophil release of proinflammatory IL1β, CXCL8, and IL6 [122]. Finally, in vivo animal studies suggest that both PAR1 and PAR4 are involved in the regulation of neutrophil migration by increasing neutrophil rolling, migration, and chemotaxis [123,124].
Phospholipids are important for the anchoring, orientation, and signaling of plasma membrane proteins [142], whereas gangliosides (i.e., glycosphingolipids linked to sugar chains) have effects on β1 integrins and cytoskeletal proteins that are involved in cell migration [142]. Furthermore, the modulation of fatty acid saturation is important to control the stiffness and elasticity of the cell membrane and for regulation of receptorinitiated signaling [143,144]. These lipid-associated modulations of cell signaling can be seen with relatively small changes in membrane lipid composition [145]. Finally, both the proportion of polyunsaturated fatty acids and the cholesterol level of the plasma membrane can be influenced by dietary interventions [146,147]. Although these studies were based on experimental in vitro models or animal models, the observations are possibly also relevant for neutrophil G-protein-coupled receptors.
Experimental models suggest that electrostatic interactions between the N-terminal parts of the CXCR1 receptor and the lipid parts of the cell membrane are important for the three-dimensional folding of the receptor chain [148][149][150][151]. These interactions increase the affinity of CXCR1 for CXCL8 [148]. Furthermore, after CXCR1 ligation by CXCL8, the receptor relocalizes to lipid rafts that facilitate its G-protein-mediated signaling [152]. The leukotriene B 4 receptor BLT1 is also localized to lipid rafts in neutrophils, and cholesterol deprivation disrupts the lipid rafts and thereby reduces downstream BLT1 signaling [153,154]. In contrast, the FPR-1 receptor is localized outside the lipid rafts and its downstream signaling is therefore not altered by lipid raft disruption [153].

Modulation of Neutrophil Migration in Patients with Sepsis: Reverse Migration Due to Inflammation-Induced Modulation of Normal Chemotactic Mechanisms
Most neutrophil studies in sepsis patients (Sections 8-10) are based on circulating neutrophils; it is stated in the text when the cells were derived from other compartments.
The local clearance of neutrophils at inflammatory sites is caused by local necrosis or the induction of apoptosis with subsequent phagocytosis by macrophages, but neutrophils can additionally migrate back to the circulation [155]. Animal models have shown that up to 70% of recruited neutrophils can return to the circulation from inflammatory sites [156,157]. Reverse migrated neutrophils are characterized by a ICAM-1 high CXCR1 low phenotype, and this is different from the majority of circulating ICAM-1 low CXCR1 high and tissue-resident ICAM-1 high CXCR1 high neutrophils [158,159]. The multistep process of reversed neutrophil migration is regarded as a programmed process characterized by defined molecular events [158][159][160][161][162][163][164][165][166][167][168][169][170]: • Modulation of chemotactic responses. Neutrophils from sepsis patients show attenuated chemotaxis towards formylated peptides, leukotriene B4, and CXCL8 [170]. Furthermore, neutrophil C5aR expression reaches a peak early during sepsis and thereafter gradually declines [170]; • Altered microvascular permeability modulates chemokine gradients. The endothelial junctions are altered during inflammation, the microvascular permeability is thereby increased and the leakage of chemokines (e.g., CXCL1) from the circulation to the perivascular environment alters the chemokine gradients and contributes to reversed neutrophil migration [160][161][162]. Animal models suggest that CXCL8 and CXCR2 are also important for reversed neutrophil migration [164]; CXCL8 is a chemoattractant at lower concentrations but it can function as a chemorepellent at higher levels and thereby facilitate reversed migration [165]; • A possible role of altered receptor trafficking. Chemokine receptors can be inactivated by internalization and degradation, and end-target chemoattractants use this mechanism to desensitize neutrophils to intermediary chemoattractants [168]. Receptor levels can be quickly altered by the modulation of receptor internalization/degradation compared with mechanisms that require altered gene transcription. Animal models suggest that altered CXCR1/CXCR2 balance due to continued CXCR2 signaling together with reduced CXCR1 levels/signaling facilitates reverse neutrophil migration [153,158,159,168]; • CXCR4/CXCL12 signaling. Disruption of CXCR4/CXCL12 signaling would facilitate reversed neutrophil migration, but animal models suggest that CXCR4 is upregulated when reversed migrated neutrophils migrate to new peripheral organs [166]. These neutrophils show increased expression of the early apoptotic marker Annexin V, and the CXCR4 upregulation may facilitate their later migration back to the bone marrow [166]. Thus, at least some reverse migrated neutrophils possibly migrate via other peripheral organs to the bone marrow where they undergo apoptosis; • Local release of neutrophil elastase alters endothelial cell junctions. Animal studies suggest that neutrophils can exhibit extravascular-to-luminal migration through endothelial cell junctions, and this process depends on the reduced expression and/or function of the endothelial junctional adhesion molecule-C (JAM-C) [167]. Neutrophil elastase cleaves JAM-C and thereby promotes neutrophil-reversed migration [167]. Furthermore, the local release of neutrophil elastase is promoted by leukotriene B 4 ; this chemotactic signal thereby seems to promote reverse transendothelial migration of neutrophils during the late steps of inflammation [167]. Finally, neutrophil elastase seems to be presented to JAM-C by the neutrophil αMβ2 integrins (CD11b/CD18), but ICAM-1 (expressed at high levels by reversed migrating neutrophils) also seems to be important for the cleavage of JAM-C [167]; • Arachidonic acid metabolism. Leukotriene B 4 is an early proinflammatory derivative of arachidonic acid that seems to facilitate the later resolution of inflammation through the release of neutrophil elastase (see above). Furthermore, neutrophils seem to shift to the production of the alternative anti-inflammatory arachidonic acid derivative Lipoxin A 4 when they are localized in an inflammatory environment [168], and Lipoxin A 4 enhances reversed migration [169]; • Proteases. The levels of proteases other than neutrophil elastase may also be involved in the regulation of reversed migration, e.g., Cathepsin C reduction reduces reversed neutrophil migration [166]; • Hypoxia-inducible factor 1 α (HIF-1α). The transcription factor HIF-1α is expressed by activated neutrophils; it is targeted for degradation by oxygen-dependent propyl oxygenase but this enzyme is suppressed in response to bacterial infection and HIF-1α will therefore accumulate [168]. Activated HIF-1α promotes neutrophil survival and reduces reversed migration [166,168], but it is not known whether or how HIF-1α is involved in the regulation of reverse neutrophil migration; • Macrophages. Concomitant macrophage infiltration can also facilitate reversed neutrophil migration [155].
To conclude, the molecular mechanisms responsible for reversed neutrophil migration are complex and involve the modulation of several mechanisms that are important for the early steps of proinflammatory extravasation/tissue migration, i.e., arachidonic acid metabolism, CXCR1/2 signaling, neutrophil/endothelial adhesion, and microvessel permeability. Whether modulation of neutrophil retention signaling (e.g., CXCR4/CXCL12, HIF-1α, and semaphorins) also contributes needs to be further investigated [165,166,168,170].

Increased Bone Marrow Release of Neutrophils: Effects of Sepsis on G-CSF and CXCL12
Inflammation is often associated with increased levels of circulating neutrophils due to increased bone marrow release [171,172]. The growth factor G-CSF can be released by neutrophils together with a wide range of proinflammatory cytokines (Table 1); it is upregulated in sepsis as a part of the acute phase reaction and promotes the generation and release of mature and immature neutrophils [173,174]. Furthermore, under homeostatic conditions, CXCL12 is highly expressed in bone marrow stromal cells, and CXCL12/CXCR4 serves as a main mechanism for bone marrow retention of neutrophils [175]. Infectious stress causes rapid downregulation of CXCL12 mRNA and protein levels in the bone marrow, and neutrophil release is thereby facilitated [171,172].

Effects of Sepsis on Neutrophil Rigidity and F-Actin Accumulation: Increased Rigidity as a Possible Mechanism That Increases the Risk of Organ Failure
Proinflammatory mediators as well as bacterial products cause increased neutrophil rigidity and priming of the neutrophils at the same time [176,177]. The increased rigidity is associated with F-actin accumulation; it can be induced by proinflammatory TNFα and this TNFα effect seems to be mediated by the transcription factor PPARγ [90,178].
There is also reduced endothelial rolling of neutrophils during sepsis, and increased rigidity/priming combined with decreased rolling probably leads to neutrophil sequestering in the microcirculation as a part of the pathogenesis of organ dysfunction/failure [177,179]. Recent studies investigated the effect of dietary polyunsaturated fatty acids in a murine model of Staphylococcus aureus sepsis [180][181][182][183]. Increased dietary polyunsaturated fatty acids then increased survival, decreased the bacterial load, and increased neutrophil infiltration at the inflammatory site. The cell membrane should be regarded as a dynamic regulator of neutrophil migration [144][145][146]184,185]. A possible explanation for this effect of dietary polyunsaturated fatty acids could therefore be modulation of neutrophil membranes with altered chemotactic signaling, increased neutrophil migration/infiltration, and thereby decreased mortality in sepsis. Alternatively, dietary modulation may alter the membrane stiffness and thus have a general effect on neutrophil migration that reduces the risk of multiorgan failure.

Effects of Sepsis on Neutrophil-Endothelial Interactions: The Importance of Adhesion Molecules, Integrins, and Heparinase
Studies in animal models have demonstrated that endothelial ICAM-1 expression is upregulated during sepsis, and there is also a transient increase in VCAM-1 [186][187][188][189][190]. Sepsis is also associated with increased degradation of the glycocalyx, and a TNFα-induced increase in endothelial expression of heparinase is important for this degradation [191]. A possible effect of this degradation is exposure of adhesion molecules on the endothelial cells leading to increased neutrophil adhesion.
Neutrophils upregulate β1 and β2 integrins in response to proinflammatory mediators; these adhesion molecules interact with endothelial ICAM-1/VCAM-1 and studies in animal models suggest that this upregulation enables firm adhesion to endothelial cells and thereby inhibition of the later extravascular neutrophil migration [95,96,192]. Animal models suggest that the combined upregulation of integrins in circulating neutrophils and endothelial adhesion molecule is seen especially for gram-positive bacteria, whereas for other bacteria the integrin upregulation is seen only for interstitial neutrophils [193]. Animal models also suggest that circulating neutrophils upregulate β1 integrins during sepsis, and this upregulation may influence neutrophil crawling and adhesion to endothelial cells [194,195].
Proteases can modulate endothelial cell regulation of inflammation through the ligation of protease-activated receptors (see Section 5), but further studies are needed to clarify whether these ligand/receptor interactions are involved in the pathogenesis of sepsis-associated organ failure.

Selectins and Selectin Receptors in Sepsis
Neutrophil L-selectin and its ligands (CD44, PSGL1 and ESEL-1) are important for neutrophil adhesion to endothelial cells [196,197], but the possible modulation of selectins/selectin ligands during sepsis has not been characterized in detail [197].

Soluble Adhesion Molecules as Modulators of Endothelial-Neutrophil Interactions in Sepsis?
Several adhesion molecules exist in biologically active soluble forms, and this is true both for selectins and ICAM-1 [198]. Membrane-bound forms of these adhesion molecules are expressed by endothelial cells and/or neutrophils and are important for extravasation and migration (see Section 2.2). The systemic levels of these soluble adhesion molecules can be altered in patients with bacterial infections/sepsis [198,199]. The serum levels of L-, P-, and E-selectins as well as ICAM-1 are then increased in immunocompetent patients. In contrast, patients with chemotherapy-induced neutropenia show only increased serum levels of L-selectin and ICAM-1 whereas P-and E-selectin levels are decreased during bacterial infections [198,199]. The possible influence of these infection-induced alterations on in vivo neutrophil migration has not been investigated, but the biologically active soluble forms will possibly compete with the corresponding membrane-expressed forms and thereby influence neutrophil migration [198].
Several other cell membrane molecules can also be released by neutrophils during sepsis [200]. First, CD14 is a part of the TLR4 receptor complex, and it is upregulated and released from the lipopolysaccharide-binding protein in sepsis. Second, the triggering receptor expressed on the myeloid cells 1 (TREM1) is also upregulated and released by neutrophils after exposure to bacteria. Third, azurocidin is released when neutrophils bind to selectins, and this protein can alter the endothelial cytoskeleton and thereby enhance neutrophil migration through the endothelial layer. However, additional studies are needed to clarify whether these soluble forms will influence neutrophil migration and/or can be used as a biomarker of sepsis in routine clinical practice.
The serum levels of soluble VCAM-1 and ICAM-1 correlate with organ dysfunction in sepsis [186,[201][202][203], and these clinical observations support the hypothesis that soluble adhesion molecules are involved in the pathogenesis of sepsis.

Multifactorial Modulation of Neutrophil Chemotaxis in Sepsis
The regulation of neutrophil chemotaxis shows complex alteration in patients with sepsis , and Table 2 summarizes how various neutrophil receptors/mediators are involved in the sepsis-associated modulation of neutrophils.

FPR LTB 4 receptor
The chemotactic responses to these mediators seem to be reduced in sepsis patients [170].

TLR9
This intracellular receptor is important for local recruitment of various immunocompetent cells including neutrophils [196]. TLR9 activation in neutrophils seems to impair chemotaxis and thereby reduce survival for patients with sepsis [79,244].
IL10 receptor This receptor mediates the inhibition of neutrophil chemotaxis as well as neutrophil cytokine release [91,245], but it can also reduce TLR-induced effects in neutrophils [246,247].

PAFR
Platelet-activating factor (PAF) receptor ligation seems to contribute to the failure of neutrophil migration, and receptor blocking increases survival in animal sepsis models [227,228].

PPARγ
Ligation of the nuclear PPARγ inhibits neutrophil chemotaxis and downregulates the expression of several proinflammatory transcription factors (including NFκB and STAT6); its neutrophil expression is increased in sepsis [90,250,251].
General inhibition of neutrophil chemotaxis and associations with prognosis. Studies in animal models have described inadequate neutrophil migration towards infection foci during severe sepsis even though high chemokine levels are seen at the infection sites [74,204]. These abnormalities are probably caused by reduced responsiveness for all four main chemotactic mechanisms (see Figure 1). The migratory responses to formylated peptides, LTB 4 and CXCL8 are reduced, and neutrophil C5aR expression reaches an early peak during sepsis before a gradual decline [154]. Furthermore, patients with the largest reduction of neutrophil migration towards formylated peptides or LTB 4 have decreased survival [205]. Decreased neutrophil migration towards CXCL8 is seen especially in patients with severe sepsis; this chemotactic CXCL8 effect seems to be a part of a more complex dysregulated phenotype also including defects in NETosis and delayed apoptosis [206]. Finally, animal studies suggest that C5aR dysfunction alters neutrophil migration and thereby contributes to the development of organ failure [207][208][209].
The role of G-protein-coupled receptor (GPCR) kinases. Repeated exposure to agonists will lead to the internalization of many G-protein-coupled receptors, and this process seems to depend on G-protein-coupled receptor kinases (GPCR kinases) [210]. These kinases phosphorylate intracellular parts of activated receptors and facilitate the recruitment of arrestins that decouple the receptor from the G-protein and thereby lead to receptor internalization [211]: • TLR2/4/9 ligation induces the upregulation of GPKR kinase 2 and thereby internalization of CXCR2 [74,79,93,212,213]. In contrast, IL33 prevents TLR-induced upregulation of GPCR kinase 2 and can thereby stimulate neutrophil chemotaxis [93,214]; • The TLR induction of CXCR2 internalization seems to depend on TNFα [235][236][237][238][239][240][241]. This TNFα effect is possibly mediated by an autocrine loop because TNFα can be released by neutrophils (Table 1), but TNFα release by neighboring immunocompetent cells (see Section 6) may also contribute to local effects and to the increased systemic TNFα levels in sepsis patients [215]; • Both TLR and TNF-dependent pathways upregulate NO synthase that in turn induces GPCR kinase 2 and finally reduces neutrophil expression of CXCR2 [216]. Taken together these observations suggest a signaling cascade involving TLR→TNFα→NO synthase→GPCR kinase 2→CXCR2 downregulation; • Finally, PI3Kγ-initiated signaling can also cause GPCR kinase induction and thereby reduced CXCR2 expression [217,218].
Thus, there seems to be three different intracellular signaling events/pathways (i.e., TLR, TNFα, and PI3Kγ) that contribute to the internalization and thereby the reduction of CXCR2 cell surface expression in neutrophils in sepsis. CXCR1 versus CXCR2 expression. As explained above, CXCR2 is internalized in circulating neutrophils during sepsis [219][220][221]. Decreased neutrophil CXCR2 expression in sepsis patients correlates with the Apache II score, a parameter for disease severity [219]. In contrast, neutrophil CXCR1 levels are not altered in sepsis patients [219].
The role of the CCR2 chemokine receptor. CCR2 is induced on the neutrophil surface during sepsis in a TLR2/4 dependent manner; this promiscuous chemokine receptor is probably not involved in neutrophil recruitment to the infectious site [223], but it seems to be involved in neutrophil recruitment to vital organs and the pathogenesis of organ failure [223,224]. Finally, non-surviving septic shock patients have higher neutrophil CCR2 expression than survivors [224].
The role of Toll-like receptors. Neutrophil expression of TLR2 and TLR4 seems to be downregulated in patients with septic shock, and gene expression studies show an additional downregulation of several inflammatory response gene clusters, including TLR-dependent genes [225,226]. However, despite this TLR downregulation, these receptors mediate important inhibitory effects on CXCR3 expression (see above).
Platelet-activating factor (PAF). The PAF receptor activates MAP kinases and induces the production of proinflammatory cytokines including CXCL8 [227,228]. However, despite the increased cytokine releases the final effect of receptor ligation is reduced neutrophil migration, and receptor blocking increases survival in animal sepsis models [227,228].
Hydrogensulfide-synthesis and its influence on endothelial adhesion. The activity of the hydrogen sulfide-synthesizing enzyme cystathionine g-lyase is increased in sepsis; its increased activity seems to be associated with increased neutrophil adhesion to endothelium, increased migration/chemotaxis, and maintained/increased CXCR2 expression in murine sepsis [229,230].
The possible importance of LOX-1, IL17, and Peroxisome proliferator-activated receptor γ (PPARγ). Both LOX-1 and the acute phase protein α-1 acid protein contribute to the failure of neutrophil migration [89,231,232]; this is possibly mediated by the LOX-1 receptor recognition of several inflammatory products including CRP, apoptotic cells, various bacterial products, and activated platelets [89,233]. Finally, IL17 seems to facilitate the recruitment of neutrophils to the infection site [234], whereas the results from the studies of PPAR effects on neutrophil migration in sepsis are conflicting [204]. It is well-accepted that immunological dysregulation is essential in the development and progression of sepsis [204]. Neutrophils then have effector mechanisms that are essential for the removal of the infecting agent, including degranulation, phagocytosis, and the formation of extracellular traps, but they are also important for local immunoregulation at the infectious site and in the pathogenesis of distant organ dysfunctions [204].
Neutrophils are among the first immunocompetent cells that are recruited to an infectious site; they become important through several mechanisms that amplify inflammation during the progression of sepsis. Reversely migrating neutrophils still seem to have a proinflammatory phenotype characterized by their increased capacity of superoxide release, high ICAM-1 membrane expression, likely increased levels of inducible nitric oxide synthase, increased capacity of extracellular trap formation, and increased lifespan due to antiapoptotic signaling [252][253][254][255][256]. Experimental studies have also shown an association between the level of reverse-migrated neutrophils in the circulation and organ failure [253,257,258]. Taken together these observations are consistent with the hypothesis that reversed migration contributes to the systemic dissemination of the inflammatory response and to organ failure distant to the infection site. Experimental animal models of sepsis also suggest that LTB 4 promotes reverse neutrophil migration in sepsis and thereby contributes to the pathogenesis of organ failure [257,258]. Thus, even though available data on reversed neutrophil migration in human sepsis are limited, experimental evidence suggests that reversed migration is involved in the pathogenesis of sepsis-associated organ failure.

Exosomes in Sepsis: A Possible Mechanism for Neutrophil Communication with Neighboring Cells and for Their Influence on Distant Organs
Exosomes are extracellular 40-150 nm endosome-derived vesicles; they carry proteins and nucleotides/noncoding RNA as well as metabolites/lipids that can be transferred between cells [259]. Animal models have shown that neutrophil-derived exosomes are increased in the blood during sepsis, and these exosomes carry high levels of proinflammatory mediators, including cytokines and DAMP molecules. The exosomes probably represent an important form of communication between neutrophils and neighboring as well as distant cells/organs, and they possibly have a role in the development of distant organ dysfunctions. Studies in hepatocytes have shown that CXCR1 and CXCR2 regulate exosome release [260], but it is not known whether similar effects are seen in neutrophils during chemotaxis. The effect of microparticles on neutrophil migration depends on the cell surface molecules of the particles. Some microparticles seem to enhance neutrophil chemotaxis to CXCL8, and this is mediated by L-selectin and P-selectin glycoprotein ligand-1 present on the surface of these particles [261]. In contrast, particles may also mediate anti-inflammatory effects and inhibit neutrophil chemotaxis; this effect is associated with the surface expression of annexin-1 [262].

Several Molecular Markers of Neutrophil Heterogeneity Are Involved in the Regulation of Neutrophil Migration: The Possible Importance of Various Subsets in Sepsis
The modulation of neutrophil migration occurs in a complex context of a more general sepsis-induced immunosuppression [263,264]. The complexity is further increased by the identification of several neutrophil subsets that seem to differ with regard to regulation of migration .

Neutrophil Heterogeneity
The circulating neutrophils were previously regarded as fully differentiated phagocytic defender cells, but they are now regarded as a heterogeneous population including different effector and immunoregulatory subsets [270,284]. Several of the molecular markers used to identify various neutrophil subsets are important regulators of fundamental neutrophil functions, including neutrophil migration ( Table 3). The following neutrophil subsets were identified:

•
Mature or classic neutrophils have the phenotype CD16 + (Fc receptor), CD177 + (cell surface glycoprotein/CD32 binding), CTL-2 + (von Willebrand's factor receptor), CD11b/CD18 + , and CD11a/CD18 + (two integrins) [281]. An alternative strategy to identify neutrophil subsets is based on differences in cell density [284][285][286], and the low-density cells can then be further subclassified based on their expression of CD11b, CD16, and CD86 [284]; • Animal studies suggest that aged neutrophils have a different phenotype [287], but the survival time seems to differ between neutrophil subsets and cells with the HLA-DR + CD49 + CD80 + phenotype have an extended survival of up to 72 h [288]; • CD177 is a cell surface glycoprotein; most humans have both CD177 negative and CD177 positive neutrophils [289]. One study investigated 535 healthy individuals: 65% of them had >60% CD177 + circulating neutrophils, 25% had intermediate levels (20-60% CD177 + neutrophils), and a small minority of 14 patients had only CD177neutrophils. Proteinase 3 is a neutrophil intracellular protease; the fraction of circulating protease 3 positive neutrophils in healthy individuals varies between 0 and 100% [282] and there is a cellular covariation between CD177 and protease 3 [283]; • The minority of mature CD66b + CD10 + neutrophils differ in their regulation of T cell activation compared with CD66b + CD10neutrophils [281,290], a subset of neutrophils seem to have a specialized B cell helper function [291], and neutrophil subsets differ even in their capacity to support cancer cell proliferation [284].
Thus, several neutrophil subsets were identified, and animal studies suggest that the balance between neutrophil subsets can be important for the capacity to resist certain bacterial infections [281,292]. Furthermore, studies in human inflammatory diseases suggest that the neutrophil heterogeneity is even more extensive, including CD63 + or PD-L1 + subsets identified in patients with cystic fibrosis [293,294], CD64 + neutrophils as a marker of sepsis (see below) [270], subsets expressing the receptor activator of nuclear factor κB ligand (RANKL) identified in chronic obstructive pulmonary disease [295], CD49d + neutrophils detected in the nasal fluid during viral infections [296], and HLA-DR + subsets in patients with Leishmaniosis [297]. It is difficult to judge whether these subsets represent true differentiation or whether they are markers of activation, and it is not known whether these subsets can be detected in patients with sepsis. Table 3. Individual molecular markers used to characterize neutrophil subsets and their involvement in neutrophil migration.

Neutrophil Expression of the Fcγ Receptor CD64 in Patients with Sepsis: Increased Levels That Associate with Prognosis
The CD64 FcγR1 is expressed only at low levels on resting neutrophils; this expression is increased during sepsis, and the level of circulating CD64 + neutrophils is regarded as a possible diagnostic or prognostic biomarker in sepsis [272,273]. CD64 expression is increased early during sepsis, and the levels are generally higher for patients with septic shock. Furthermore, the CD64 level correlates with severity scores such as APACHE II and total SOFA scores and predicts survival/mortality for intensive-care sepsis patients. A recent study of sepsis patients showed that immature CD64 + neutrophils could be further subclassified; two new neutrophil subsets were then identified based on the expression of CD123 (the IL3 receptor) and PD-L1 [298]. Thus, CD64 + neutrophils are heterogeneous, but it is not known whether the diagnostic/prognostic impact of circulating CD64 + neutrophil levels is caused by all or only a subset of the CD64 + neutrophils.

Neutrophils from Patients with Sepsis Show Increased CD177 (a Cell-Surface Glycoprotein) and Decreased CD10 Expression, and the Levels of the Two Markers Show Inverse Correlation
A study of neutrophils derived from 45 patients with septic shock identified 364 upregulated and 328 downregulated genes. CD177 mRNA had the highest fold change alteration when patients were compared with healthy individuals, and this difference was also confirmed at the protein level [299]. However, the sepsis patients were also characterized by a constant CD177 negative neutrophil subset, CD177 was coexpressed with protease 3, and circulating neutrophils in addition showed decreased CD10 levels with a significant inverse correlation between CD177 and CD10 protein levels. The CD10 metalloprotease is a negative regulator of chemotaxis [279], and the increased CD177/decreased CD10 combination will thus facilitate the extravasation/migration of this subset (Table 4).

The Fraction of Olfactomedin 4 (OLFM4) Positive Neutrophils Is Increased in Patients with Sepsis, and High Levels Are Associated with Adverse Prognosis Even in Adjusted Analyses
OLFM4 is a neutrophil-specific granule protein secreted through exosomes [278]. The percentage of OLFM4-expressing neutrophils in healthy individuals varies between 8 and 57% (Table 4), and the levels of positive cells increase in sepsis. However, the levels can also be increased especially in other bacterial infections, but also in viral and parasitic infections, sterile inflammatory disorders, and in cancer patients [274,278,300].
A recent study described the increased mortality of sepsis patients with high levels of circulating OLFM4 + neutrophils and the same was true for non-septic patients with infections and Systemic Inflammatory Response Syndrome (SIRS). In total, 56% of sepsis patients with elevated OLFM4 + neutrophils died compared to 18% of patients with OLFM4 + neutrophils below 37.6% [301]. These associations between the OLFM4 + neutrophil subset and mortality were significant when adjusting for age, sex, absolute neutrophil count, comorbidities, and the total SOFA score. Furthermore, similar observations were made in a prospective pediatric study including patients with septic shock; patients with at least two organ failures at day 7 of septic shock or dying before day 28 had a higher percentage of OLFM4 + neutrophils compared to patients with less severe disease [302]. These associations were also significant after adjusted analyses. Thus, high levels in the circulation of the OLFM4 + neutrophil subset are associated with severe/multiple organ failures, and this cell subset may therefore have a role in the pathogenesis of organ failure [302][303][304].
OLFM4 can be released by neutrophils, and the serum level of OLFM4 is increased in sepsis patients, especially patients with severe disease [303]. Finally, OLFM4 inhibits cathepsin C which activates cathepsin G [305], and cathepsin G is a regulator of neutrophil chemotaxis [195,[276][277][278]306]. Thus, the modulation of neutrophil migration may contribute to the adverse prognostic impact of the ILFM4 + neutrophil subset and/or soluble OLFM4.

Myeloid-Derived Suppressor Cells in Sepsis; Further Neutrophil Development to an Immunosuppressive Neutrophil Subset Associated with Adverse Prognosis in Sepsis Patients
The two main myeloid-derived suppressor cell (MDSC) subsets are monocytic and granulocyte (G)-MDSC; the latter has the phenotype CD33 + CD11b + CD14 + HLA-DR low [307]. Thus, G-MDSCs represent a subset or a further neutrophil differentiation/development. Several neutrophil-secreted cytokines can function as expanders of MDSCs, and the TLRtarget NFκB can function as an MDSC activator [307].
There is an increased expression of the PD-1 ligand by MDSCs in sepsis patients, and ligation of the PD-1 checkpoint on T cells leads to T cell apoptosis that can contribute to the observed lymphopenia during sepsis [307]. Furthermore, high G-MDSC levels are associated with adverse prognosis in sepsis patients, whereas a similar association is not observed for monocytic MDSCs [306,[308][309][310]. Furthermore, typical mature G-MDSCs as well as a CD14 low mature G-MDSC subset show increased levels especially in grampositive sepsis [311]. Finally, low CD10 (CD10 dim ) and/or CD16 (CD16 dim ) expression by G-MDSC were associated with an adverse prognosis in sepsis [312]. Thus, G-MDSCs subsets are included among neutrophils and increased MDSCs seem to contribute to the immunocompromised state and an adverse prognosis in sepsis.

Important Bacterial Products That Interfere with Neutrophil Functions
Sepsis can be caused by a wide range of bacteria; gram-positive infections have increased over time and are now almost as common as gram-negative infections [313,314]. This increase is probably due to the increased use of invasive procedures. Various Staphylococcus species are the most common among gram-positive infections (30-35% of cases), although Enterococcus infections (10%) are also common. A wide range of gram-negative bacteria can cause sepsis (55-60%), whereas fungal infections constitute a smaller subset (15%) with Candida species being the most common. The most common sites of infection are respiratory (40% of cases), genitourinal (15-20%), and abdominal (5-10%), but 6-8% of patients have an unknown site [84,85,[314][315][316]. In this section, we will describe how common sepsis-causing bacteria interact with neutrophils and thereby modulate their migration [84,85,214,243,269,317]. Most neutrophil studies referred to in this section and the following Sections 12-16 are based on circulating neutrophils; it is stated in the text when cells were derived from other compartments.

Modulation of Neutrophil Migration by Staphylococcus aureus
The effects of Staphylococcus aureus on neutrophils, including neutrophil migration, have been extensively studied, and several bacterial molecules can modulate neutrophil migration [107,318]:

•
The chemotaxis inhibitory protein of Staphylococcus aureus (CHIP) can bind and thereby block the extracellular parts of the chemotactic C5a and formylated peptide receptors but without initiation of intracellular signaling [319][320][321][322][323][324]; • Agonistic ligation of the neutrophil formyl peptide receptor-like-1 stimulates neutrophil migration [38]. Staphylococcus aureus releases additional proteins that bind to and block this receptor and thereby inhibit neutrophil chemotaxis [317,324,325]. Some of these antagonists also block Fcγ receptors [107,318]; • The cysteine protease Staphopain A cleaves the N-terminal part of the CXCR2 receptor and thereby inhibits its binding of chemokine ligands [326]; • Selectins are important for the initial steps of neutrophil extravasation, and neutrophils express the P-selectin glycoprotein ligand 1 [22,107,318]. Staphylococcus aureus secretes two proteins that bind to this selectin ligand and thereby inhibit neutrophil rolling on endothelial cells [327,328]. One of these proteins can also bind to the N-terminal end of other G-protein-coupled receptors and thereby inhibit their binding with other ligands; this mechanism inhibits effects of complement factors and CXCL chemokines on neutrophil migration [329]; • TLR2 can recognize several staphylococcal lipoproteins that have a TLR2 antagonizing effect [330,331]; • Staphylococcal superantigen-like proteins 1 and 5 are broad protease inhibitors; they inhibit the neutrophil proteases MMP8 and MMP9 and thereby the proteolytic cleavage and potentiation of CXCL8 [332].
These observations clearly illustrate that bacterial molecules can modulate/inhibit neutrophil migration through various molecular mechanisms. These mechanisms may also contribute to the increased levels of certain MDSC subsets especially in gram-positive infections [311].

Streptococcal Modulation of Neutrophil Migration
Several cell wall components derived from Streptococcus pneumoniae, including lipoteichoic acid and other lipoproteins, are recognized by TLR2, and downregulation of chemokine receptors would therefore be expected [242]. Furthermore, TLR4 can bind the exotoxin pneumolysin, and pneumococcal DNA can bind to TLR9 [242]. One would expect ligation of these TLRs to cause downregulation of chemokine receptors and thereby inhibition of neutrophil migration (see Section 9.7).

Gram-Negative Bacteria; TLR4 Binding and Binding of Microorganisms to Lipid Rafts
Lipopolysaccharide is a ligand for TLR4 [243,333], and gram-negative bacteria would therefore be expected to inhibit neutrophil migration through TLR4-mediated downregulation of chemokine receptors (see Section 9.7).
Lipid rafts are membrane domains rich in sphingolipids/cholesterol, and they contain glycosylphosphatidylinositol-anchored proteins as well as other signaling molecules also anchored via lipids (see Section 7) [334]. The lipid rafts are important for chemotactic signaling in neutrophils, but certain components can also function as receptors for pathogen-associated molecular patterns [334,335]. This is true for lactocylceramides that are able to bind directly to pathogen components including molecules derived from mycobacteria and candida species, such as Pneumocystis jeroveci beta-glycan and Escherichia coli proteins [335]. Other glycosphingolipids are also able to bind microbiological molecules including Escherichia coli, Pseudomonas aeruginosa, and shigella enterotoxins. Thus, the glycosphingolipids in lipid rafts are able to function as pathogen-associated molecular pattern receptors, and at least some of these (e.g., soluble beta-glucan from Candida albicans) can induce neutrophil migration [334,335].
Caveola are plasma membrane invaginations with a diameter of 60-80 nm; they are a subset of lipid rafts enriched in cholesterol and sphingolipids (see Sections 2.3 and 7) [34]. Caveolin is a major constituent of the caveola; it exists in the isoforms caveolin-1-3 and is involved in cholesterol transport [336,337]. However, several G-protein-coupled receptors are sequestered within the caveola through interactions with caveolin-1, a protein that is expressed by many different cells including neutrophils [338]. Caveolins can also bind various microorganisms [34]. Furthermore, caveolin-1-deficient mice show increased mortality during infections with Salmonella and Pseudomonas aeruginosa, but at the same time, these mice have increased neutrophil infiltration at the infection site [34]. Thus, caveola are involved both in microbial binding as well as in G-protein-coupled receptor signaling, and experimental models suggest that caveolin is important for neutrophil migration [338]. Future studies have to clarify whether caveola-mediated effects on neutrophil migration are caused by binding to neutrophil caveola or whether indirect effects mediated by binding to endothelial caveola, for example, also contribute.

Candida Species and Neutrophil Migration
Candida species and especially Candida albicans show several molecular interactions that can modulate neutrophil migration and the proinflammatory neutrophil phenotype. First, various chemotactic formyl peptide receptors bind fungal molecules [339]. Second, the αMβ2 integrin has two distinct ligand-binding sites that cause different cellular responses; one of these sites is highly promiscuous and can bind both endogenous and microbial ligands [340]. Dual ligation of αMβ2 by the extracellular matrix protein fibronectin and candida β-glucan will then enhance neutrophil chemotaxis, and at the same time, there is a shift in fibronectin binding from α5β1 to α3β1 integrins. Thus, these effects of candida on neutrophils are caused by other molecular mechanisms aside from the bacterial effects.
Recent studies have described that the β-1,6-long glucosyl side-chain-branched betaglucan derived from Candida albicans causes the dose-dependent induction of neutrophil migration by binding to lactosylceramide followed by activation of the Src family kinase/PI-3K/G-protein signaling pathway [341].
Furthermore, resolvins are a class of anti-inflammatory lipids derived from omega-3 polyunsaturated fatty acids, and Candida albicans can synthesize resolvin E1 that reduces neutrophil chemotaxis in response to CXCL8 but enhances the phagocytosis of Candida [342]. Finally, secretory aspartyl proteinases (Saps) of Candida albicans increase in vitro chemotaxis of neutrophils. Sap2 and Sap6 then induce neutrophil migration and the release of chemotactic chemokines [343][344][345][346].

Effects of Antimicrobial Drugs on Neutrophil Migration
Several studies have described the effects of antimicrobial agents on neutrophil migration, including agents used in the treatment of sepsis:
We emphasize that there are relatively few studies, hence the effects have mainly been observed in experimental models, the molecular mechanisms have not been characterized, and the possible importance of these effects in sepsis has not been investigated.

Elderly Patients
Neutrophils from elderly individuals were investigated in several studies (Table 4) . Functional studies have confirmed that both endothelial adhesion and chemotaxis are altered in the elderly [357][358][359][368][369][370][371]. Elderly individuals show preserved chemokinesis (i.e., random migration) whereas chemotaxis is reduced, and this chemotactic defect can be detected from the sixth decade of life [359]. A possible overall effect is therefore less accurate neutrophil migration towards inflammatory sites [359,369,371]. Increased basal PI3K activity seems to be at least partly responsible for the reduced chemotaxis, and inhibition of PI3Kγ/PI3Kδ was suggested as a possible strategy to restore neutrophil migration [359]. Table 4. Important aging-associated direct and indirect effects on neutrophil migration.

Modulation of cell surface molecules, release of soluble mediators
CD11a and CD11b integrin expression are not altered in elderly, but CD16 expression is significantly reduced [357,358].
Plasma levels of neutrophil elastase are increased, suggesting a preactivated basal state of neutrophils [359] Altered effects of receptor ligation or altered downstream signaling Fc receptors and formylated peptide receptors show altered downstream signaling through MAP kinases, the Janus kinase (JAK)/Signal transducer and activation of transcription (STAT) and PI3K-Akt pathways [360,361]. These effects are due to altered membrane structure with altered receptor recruitment (including TLRs) to lipid rafts [362][363][364].

Functional effects
Neutrophils from elderly do not have general defects in endothelial adherence or transendothelial migration [367,368]. Neutrophil chemotaxis in the elderly is impaired due to increased PI3K signaling [359,369]. This impairment is worsened in patients with pneumonia and the worsening is associated with disease severity [359]. Chemotaxis is decreased by surgery; patients recover partly within six weeks and completely within six months [370]. Chemokinesis (random movements) is increased in the elderly whereas chemotaxis is impaired; the final effect is possibly inaccurate neutrophil migration in the direction of infections [359,369,371]. Murine models of aging have shown high reversed transendothelial migration by neutrophils in inflamed tissues; this is due to the desensitization of CXCR1 by CXCL1 releasing cells localized at endothelial cell junctions [372]. These neutrophils then re-enter the circulation and disseminate into the lungs where they cause organ failure [372].

In vivo observations
Aged individuals show increased neutrophil levels in airways/bronchoalveolar lavage fluid [373][374][375]. Elderly patients with pneumococcal pneumonia show increased pulmonary infiltration of neutrophils [376]. Some animal models show reduced neutrophil numbers at infection sites in aged mice (Staphylococcus aureus and Pseudomonas aeruginosa), even though local chemokine levels are high and circulating neutrophils express high CXCR2 levels [359,371,376].
However, other models of inflammation have given different results [377].

Sepsis, Frailty, and Neutrophils
Frailty should be regarded as a manifestation of unhealthy aging and is defined as an aging-related syndrome of physiological decline with marked vulnerability to adverse outcomes [378]. Frail older individuals often present with an increased burden of symptoms (including weakness/fatigue), medical complexity, and reduced tolerance to medical and surgical interventions. Frailty can be measured based on several factors including comorbidity, use of medication, nutritional status, blood tests, or physical/emotional/cognitive function, and multiple tools for screening and assessment of frailty have been developed (e.g., The Clinical Frailty Scale, CFS) [379,380].
Frail elderly people are vulnerable to adverse health outcomes, and frailty has a prognostic accuracy higher than both SIRS and Quick SOFA (qSOFA) criteria among elderly patients admitted to hospital with suspected infection, implying that frailty as a prognostic factor can be used for risk-stratification [381]. Furthermore, preexisting clinical frailty is associated with worse outcomes, including increased mortality for elderly sepsis patients [382], and frailty together with age and disease severity are identified as predictors of long-term mortality in very old intensive care patients with sepsis [383]. Finally, a longer duration of mechanical ventilation in frail sepsis patients who underwent protocol-based weaning was also observed [384].
Frailty can be associated with systemic signs of inflammation, and aging characterized by additional frailty is associated with abnormalities in neutrophil levels and neutrophil migration [385,386]. Some studies describe that frail elderly people, compared to age-and sex-matched controls, show a stable and low-grade inflammation according to C-reactive protein levels (referred to as inflammageing), and this is associated with an expansion of circulating myeloid cells [387,388]. A positive correlation between the frailty score and neutrophil count was described for frail elderly women, whereas a negative correlation was seen for the frailty score and lymphocyte counts [387]. Other studies have also described associations between white blood cell counts and frailty, particularly neutrophil and mono-cyte counts [389]. Hence, both total neutrophil counts and neutrophil to lymphocyte ratio are increased in frail elderly people [390].
A previous study described that neutrophils from frail elderly individuals showed increased expression of integrin αM/CD11b and HLA-DR. The frequency of CD11b + neutrophils correlated with plasma TNFα levels whereas the frequency of HLA-DR + neutrophils correlated with circulating mitochondrial DNA [391]. Additional experimental studies suggested that these two soluble markers were responsible for the induction of this neutrophil phenotype. Furthermore, the aging-associated reduction in the accuracy of migration is worsened by frailty, and this inaccuracy seems to correlate both with physical and cognitive markers of frailty [392]. Additional regression analysis suggested that the frailty index together with age and Charlson's comorbidity index were able to predict neutrophil chemotaxis. Finally, the reduced neutrophil chemotaxis showed no association with systemic signs of inflammation (inflammageing), but it could be reversed by selective PI3K inhibitors [392].
Studies of the effects of physical activity/lifestyle on neutrophil migration and levels of circulating neutrophils have given conflicting results [393,394]. Taken together these studies suggest that lifestyle differences do not have any major impact on neutrophil migration and cannot explain the neutrophil dysfunctions in elderly/frail individual.

Neutrophil Migration in Patients with Myelodysplastic Syndrome
Myelodysplastic syndrome (MDS) is a hematological malignancy characterized by the abnormal differentiation of myeloid cells and usually one or more cytopenias in peripheral blood; the patients have dysplastic neutrophils and an increased risk of severe bacterial infections [395,396]. Various defects in neutrophil migration were described in MDS, especially for patients with advanced disease. First, MDS patients show decreased in vitro neutrophil chemotaxis in response to the formylated peptides CXCL1 and CXCL8 [397]. This reduction is associated with altered signaling through the ERK1/2 and PI-3K-Akt pathways [398]. Second, MDS neutrophils show decreased expression of the αMβ2 integrin that is important for transendothelial migration [399,400]. Third, MDS neutrophils show increased levels of miR-34a and miR-155 that reduce neutrophil migration towards formylated peptides and CXCL8 [401]. Finally, reduced neutrophil migration was also described for the chronic myelomonocytic leukemia variant [402]. Thus, reduced neutrophil migration seems to contribute to the increased risk of severe bacterial infections in these patients.

Neutrophil Migration after Stem Cell Transplantation
Autologous or allogeneic hematopoietic stem cell transplantation is used in the treatment of several hematological malignancies [403]. The increased risk of severe bacterial infections is associated with the duration and severity of the early posttransplant neutropenia [404]. However, reduced neutrophil chemotaxis [405,406] together with reduced CD16 expression [406] can be observed even after neutrophil reconstitution, whereas β2 integrin expression is normal [406]. Other neutrophil effector functions can also be altered in these patients [407][408][409], although the phagocytic capacity seems to be normal [405,408,409]. Finally, neutrophil functions are further modulated during sepsis in these patients [410]. To conclude, reduced neutrophil migration is observed after stem cell transplantation, but it should be emphasized that the available studies are few and the question of patient heterogeneity was not addressed.

Is There an Effect of Gender with Regard to the Severity and Mortality of Sepsis?
Previous studies suggest that sex hormones are important metabolic regulators in patients with sepsis [411], and several studies have even described gender differences with regard to the severity and/or mortality of sepsis patients [412][413][414][415]. Increased male mortality was described for patients with surgical septic shock [416], and male gender was also described as an independent risk factor for developing post-traumatic sepsis [417] as well as other major infections after surgery [418]. The molecular and cellular mechanisms behind this gender difference were not characterized in detail, but sex hormones are also involved in immunoregulation [419,420] and one hypothesis has therefore been that females develop stronger innate and adaptive immune responses than males. This hypothesis is supported by observations suggesting that estrogens, progesterone, testosterone, and other androgens are involved in the regulation of neutrophil activation and migration [419][420][421][422][423][424]. The targeting of sex hormone metabolism has even been suggested as a possible therapeutic strategy in sepsis [413,415].

The Metabolic Heterogeneity of Sepsis Patients: A Subset of Patients Has a Systemic Metabolic Profile Similar to SIRS Patients
The systemic metabolic profiles at the time of hospital admission for patients with bacterial infections and either SIRS or sepsis according to the 2016 criteria [1,2] have recently been characterized [425,426]. These two patient groups differed with regard to the amino acid, lysophosphatidylcholine, and sphingolipid metabolism, but a minor subset of exceptional sepsis patients had a metabolomic profile similar to SIRS patients. Sphingolipids (see Sections 2.2, 3.6, 11.3 and 11.4), lysophosphatidylcholines (see Section 3.6), and amino acids [427][428][429][430] are important regulators of neutrophil migration, and the metabolic heterogeneity of sepsis patients may therefore reflect heterogeneity with regard to the regulation of neutrophil migration.

Therapeutic Targeting of Neutrophils in Sepsis: Tried and Suggested Strategies
Several strategies for the inhibition of molecular mechanisms involved in the regulation of neutrophil migration were tried or considered as possible therapeutic approaches in sepsis. Targeting of neutrophil migration as a possible strategy is supported by the observation that reduced chemotaxis is associated with advanced disease and reduced survival in sepsis patients [205,206,431]. Indeed, decreased neutrophil CXCR2 expression in sepsis correlates with the Apache II severity score [219] and there is an association between high levels of circulating OLFM4 + neutrophils and mortality [301]. The strategies listed below were considered and/or tried in humans.
Hematopoietic growth factor therapy. In vivo G-CSF therapy (see Section 2.1 [13,171,172]) does not alter the expression of CD16, CD64, or L-selectin by circulating and bronchoalveolar lavage fluid neutrophils derived from patients with pneumonia [435]. G-CSF therapy of healthy stem cell donors decreases neutrophil chemotaxis and this effect may also contribute to the reduced chemotaxis in sepsis patients even though the G-CSF responsiveness of neutrophils seems to be decreased in sepsis patients compared with healthy individuals [405,436]. Both GM-CSF and the GM-CSF/IL3 fusion protein also decrease neutrophil chemotaxis [437,438]. G-CSF and GM-CSF therapy was tried in sepsis, but a recent met analysis concluded that none of these growth factors should be recommended for the routine treatment of patients with sepsis [439].
Activated protein C treatment. Neutrophils express receptors for this coagulation mediator and recombinant human-activated protein C was tried in the treatment of sepsis (see also Section 5); treatment with this agent decreases chemotaxis for neutrophils derived from the circulation and bronchoalveolar lavage fluid in a human endotoxin-induced pulmonary inflammation model [440].
Polymyxin B hemoperfusion. This procedure can reduce endotoxin levels but did not provide a benefit for a group of highly selected sepsis patients; the procedure reduced neutrophil CD11b expression but did not alter chemotaxis [441].
Inhibition of intracellular signaling. PI3K contributes to organ failure in sepsis, and inhibition of this pathway may be a possible strategy for targeting neutrophils and inhibiting neutrophil migration (Section 13.1) [217,218,317,442]. Furthermore, observations in an animal model of endotoxin-induced inflammation have shown that p38 inhibition suppresses pulmonary neutrophil infiltration [442]. These observations suggest that pathway inhibitors should be further investigated as a possible therapeutic strategy in sepsis.
Checkpoint inhibitors. MDSCs can be increased in sepsis, and these cells show increased expression of the PD-1 ligand (see Section 10.5) [307]. The use of T cell checkpoint inhibitors to restore T cell dysfunctions was investigated in two early clinical studies [443,444]; this strategy may have indirect effects on neutrophil migration and/or alter the effects of local neutrophil-induced recruitment of various T cell subsets to inflammatory sites (Section 6) [125][126][127].
None of these strategies should be regarded as an established treatment for the routine handling of patients with sepsis, but several of them show that such strategies have the capacity to alter neutrophil functionand several of them even show the capacity for neutrophil migration. Furthermore, some of these studies suggest that certain strategies are effective only for subsets of patients. In our opinion, future clinical studies of such therapeutic approaches should therefore have a focus on the biological and clinical heterogeneity of the patients.

Conclusions
Neutrophil migration is regulated by complex molecular mechanisms involving both endothelial cell adhesion/transmigration and extravascular chemotactic migration. These mechanisms are modulated in sepsis; common observations include reduced neutrophil migration toward infection sites and even reversed migration that possibly contributes to organ failures. However, sepsis patients should be regarded as heterogeneous with regard to neutrophil migration because the migration is possibly also influenced by the infecting agent, antimicrobial treatment, patient age/frailty, and other diseases predisposing to bacterial infections. Future clinical studies should clarify whether the treatment of sepsis patients needs to be individualized based on this heterogeneity.