Developmental Immunology and Role of Host Defenses in Fetal and Neonatal Susceptibility to Infection

Classical and Mannose-Binding Lectin Pathways

Activation of the classical pathway is initiated when antibodies capable of engaging C1q to their Fc portion (immunoglobulin M [IgM], IgG1, IgG2, and IgG3 in humans) complex with microbial (or other) antigens. After C1q binds to the immune complex, there is sequential binding of C1r and C1s to C1q. C1s can then cleave C4, followed by C2, and the larger fragments of these bind covalently to the surface of the microbe or particle, forming the classical pathway C3 convertase (C2aC4b). C3 convertase cleaves C3, thereby liberating C3b, which binds to the microbe or particle, and C3a, which is released into the fluid phase. This pathway can also be activated directly by CRP. When CRP binds to the surface of a microbe, its conformation is altered such that it can bind C1q and activate the classical pathway. Similarly, when MBL or ficolins engage the surface of a microbe, their confirmation is altered, creating a binding site for mannose-binding lectin–associated serine protease (MASP)1 and MASP2. MASP2, in turn, cleaves C4 and C2, leading to the formation of the C3 convertase. Independently of their activation of downstream complement components, Clq and MBL can also bind apoptotic cells and initiate their uptake into mononuclear phagocytes by macropinocytosis.

Alternative Pathway

The alternative pathway is activated constitutively by the continuous low-level hydrolysis of C3 in solution, creating a binding site for factor B. This complex is in turn cleaved by factor D, generating C3b and Bb. If C3b and Bb bind to a microorganism, they form a more efficient system, which binds and activates additional C3 molecules, depositing C3b on the microbe and liberating C3a into the fluid phase. This interaction is facilitated by factor P (properdin) and inhibited by alternative pathway factors H and I. The classical pathway, by creating particle-bound C3b, also can activate the alternative pathway, thereby amplifying complement activation. This amplification step may be particularly important in the presence of small amounts of antibody. Bacteria vary in their capacity to activate the alternative pathway, which is determined by their ability to bind C3b and to protect the complex of C3b and Bb from the inhibitory effects of factors H and I. Sialic acid, a component of many bacterial polysaccharide capsules, including those of GBS and E. coli K1, favors factor H binding. Thus many bacterial pathogens are protected from the alternative pathway by their capsules and/or by proteins that bind to factor H. Antibody is needed for efficient opsonization of such organisms.

Terminal Components, Membrane Attack Complex, and Biologic Consequences of Complement Activation

Binding of C3b on the microbial surface facilitates microbial killing or removal through the interaction of C3b with CR1 receptors on phagocytes. C3b also is cleaved to C3bi, which binds to the CR3 receptor (Mac-1, Cd11b-CD18) and CR4 receptor (CD11c-CD18). C3bi receptors are β ₂ integrins, which are present on neutrophils, mononuclear phagocytes, and certain other cell types and also play a role in leukocyte adhesion. Along with IgG antibody, which binds to Fcγ receptors on phagocytes, C3b and C3bi promote phagocytosis and killing of bacteria and fungi. Bound C3b, together with C4b and C2a or together with Bb, form C5 convertases, which cleave C5. The smaller fragment, C5a, is released into solution. The larger fragment, C5b, triggers the recruitment of the terminal components, C6 to C9, which together form the membrane attack complex. This complex is assembled in lipid-containing cell membranes, which include the outer membrane of gram-negative bacteria and the plasma membrane of infected host cells. Once assembled in the membrane, this complex can lyse the cell. Such lysis appears to be a central defense mechanism against meningococcal and systemic gonococcal infection. Certain gram-negative organisms have mechanisms to impede complement-mediated lysis, and gram-positive bacteria are intrinsically resistant to complement-mediated lysis because they do not have an outer membrane.

The soluble fragment of C5, C5a, and, to a more limited degree, C3a and C4a cause vasodilatation and increase vascular permeability. C5a also is a potent chemotactic factor for phagocytes. In addition to these roles for complement in innate immunity, complement facilitates B-cell responses to T-cell–dependent antigens as discussed in the section B-Cell Activation and Immune Selection .

Complement in the Fetus and Neonate

Complement components are synthesized by hepatocytes and, for some components, also by macrophages. Little, if any, maternal complement is transferred to the fetus. Fetal synthesis of complement components can be detected in tissues as early as 6 to 14 weeks of gestation, depending on the specific complement component and tissue examined.

Table 4-3 summarizes published reports on classical pathway complement activity (CH ₅₀ ) and alternative pathway complement activity (AP ₅₀ ) and individual complement components in neonates. Substantial individual variability is seen, and in many term neonates, values of individual complement components or of CH ₅₀ or AP ₅₀ are within the adult range. Alternative pathway activity and components are more consistently decreased than are classical pathway activity and components. The most marked deficiency is in the terminal complement component C9, which correlates with poor killing of gram-negative bacteria by serum from neonates. The C9 deficiency in neonatal serum appears to be a more important factor in the inefficient killing of E. coli K1 than the deficiency in antigen-specific IgG antibodies. Preterm infants demonstrate a greater and more consistent decrease in both classical and alternative pathway complement activity and components. Mature but small-for-gestational-age infants have values similar to those for healthy term infants. The concentration of most complement proteins increases postnatally and reaches adult values by 6 to 18 months of age.

Opsonization is the process whereby soluble factors present in serum or other body fluids bind to the surface of microbes (or other particles) and thereby enhance their phagocytosis and killing. Some organisms are effective activators of the alternative pathway, whereas others require antibody to activate complement. Thus, depending on the organism, opsonic activity reflects antibody, MBL, ficolin, CRP, classical or alternative complement pathway activity, or combinations of these. Accordingly, it is not surprising that the efficiency with which neonatal sera opsonize organisms is quite variable. For example, although opsonization of S. aureus was normal in neonatal sera in all studies, opsonization of GBS, S. pneumoniae, E. coli, and other gram-negative rods was decreased against some strains and in some studies but not in others.

Neonatal sera generally are less able to opsonize organisms in the absence of antibody. This difference is compatible with deficits in the function of the alternative and MBL pathways and with the moderate reduction in alternative pathway components. This difference is not due to a reduced ability of neonatal sera to initiate complement activation through the alternative pathway. Neonatal sera also are less able to opsonize some strains of GBS in a classical pathway–dependent but antibody-independent manner. The deficit in antibody-independent opsonization is accentuated in sera from premature neonates, in whom there is significantly reduced levels of circulating MBL, MASP, and all three ficolins and may be further impaired by the depletion of complement components in septic neonates.

Sera from term neonates generate less chemotactic activity than adult sera. This diminished activity reflects a defect in complement activation rather than lack of antibody. These observations notwithstanding, preterm and term neonates do generate substantial amounts of activated complement products in response to infection in vivo.

Toll-like Receptors

Toll-like receptors are a family of structurally related proteins and are the most extensively characterized set of innate immune pattern recognition receptors. Ten different TLRs have been defined in humans. Their distinct ligand specificities, subcellular localization, and patterns of expression by specific cell types are shown in Table 4-4 .

TLR4 forms a functional LPS receptor with myeloid differentiation (MD-2) protein, a soluble protein required for surface expression of the TLR4/MD-2 receptor complex. CD14, which is expressed abundantly on the surface of monocytes and also exists in a soluble form in the plasma, facilitates recognition by the TLR4/MD-2 complex and is essential for recognition of smooth LPS present on pathogenic gram-negative bacteria. TLR2, which forms a heterodimer with TLR1 or TLR6, recognizes bacterial lipopeptides, lipoteichoic acid, and peptidoglycan, and this recognition is facilitated by CD14. TLR2 has a central role in the recognition of gram-positive bacteria and also contributes to recognition of fungi, including Candida species. TLR5 recognizes bacterial flagellin. Consistent with their role in recognition of microbial cell surface structures, these TLRs are displayed on the cell surface.

By contrast, TLRs 3, 7, 8, and 9 recognize nucleic acids: TLR3 binds double-stranded RNA, TLR7 and TLR8 bind single-stranded RNA, and TLR9 binds nonmethylated cytosine-guanine dinucleotide (CpG)-containing DNA. These TLRs appear primarily to function in antiviral recognition and defense, and TLR9 also contributes to defense against bacteria and fungi. These TLRs preferentially recognize features of nucleic acids that are more common in microbes than mammals, but their specificity may be as much based on location as nucleic sequence: TLRs 3, 7, 8, and 9 detect nucleic acids in a location where they should not be found—acidified late endolysosomes. Individuals lacking TLR3; TRIF (TIR-domain–containing adaptor-inducing interferon-β), which is required for intracellular signaling by TLR3; or a protein UNC93B, required for proper localization of TLRs 3, 7, 8, and 9 to endosomes, are unduly susceptible to infection with herpesviruses, particularly primary herpes simplex encephalitis.

A conserved cytoplasmic Toll/IL-1 receptor (TIR) domain links TLRs to downstream signaling pathways by interacting with adaptor proteins, including myeloid differentiation primary response protein 88 (MyD88) and TRIF. MyD88 is involved in signaling downstream of all TLRs with the exception of TLR3, which signals solely through TRIF. TLR signaling via MyD88 leads to the activation and translocation of the transcription factor nuclear factor kappa B (NF-κB) to the nucleus and to the induction or activation by extracellular signal-regulated kinase (ERK)/p38/c-Jun N -terminal kinase (JNK)/mitogen-activated protein kinases (MAPK) of other transcription factors, resulting in the production of the proinflammatory cytokines, including TNF-α, IL-1, and IL-6. Production of the antiinflammatory and immunomodulatory cytokine IL-10 is dependent on signal transducer and activator of transcription 3 (STAT3) activation in addition to NF-κB and MAPK activation.

In addition to these transcription factors, activation of interferon regulatory factor 3 (IRF3) and/or IRF7 is required for the induction of type I interferons (IFNs), which are key mediators of antiviral innate immunity. Activation of IRF3 and the production of type I IFNs downstream of TLR4 are dependent on TRIF. TRIF is also essential for the activation of IRF3 and for the production of type I IFNs and other cytokines via TLR3. Conversely, TLRs 7, 8, and 9 use the adaptor MyD88 to activate IRF7 and to induce the production of type I IFNs and other cytokines. Consistent with their role in detection of bacterial but not viral structures, TLR2 and TLR5 do not induce type I IFNs.

The production of IL-12 and IL-27 (but not of the structurally related cytokine IL-23) is, like the production of type I IFNs, dependent on IRF3 and IRF7. Consequently, signals via TLR3, TLR4, and TLR7/8, but not via TLR2 and TLR5, can induce the production of these two cytokines; by contrast, each of these TLRs, except TLR3, which signals exclusively via TRIF, can induce the production of IL-23. IL-23 promotes the production of IL-17 and IL-22, which contribute to host defenses to extracellular bacterial and fungal pathogens, whereas IL-12 and IL-27 stimulate interferon-gamma (IFN-γ) production by NK cells and thereby facilitate defense against viruses and other intracellular pathogens. Thus, through the concerted regulation of IL-12, IL-27, and type I IFNs, IRF3 and IRF7 link TLR recognition to host defenses against intracellular pathogens.

Recent studies have also found that engagement of TLR7 and/or TLR8 can induce the production of high levels of IL-1β in a caspase-1–dependent manner in monocytes and DCs. As discussed immediately below, IL-1β secretion in response to many non-TLR stimuli involves the activation of caspase-1 by the inflammasome and the caspase-1–dependent cleavage of pro–IL-1β, but it is unclear if caspase-1 activation by the TLR7/8 pathway involves the inflammasome or a novel mechanism.

Nucleotide-Binding Domain– and Leucine-Rich Repeat–Containing Receptors (NLRs)

NLRs are a family of 22 proteins in humans, most of which play a role in innate immunity activation in response to microbial products or endogenously derived danger signals. These include nucleotide oligomerization domain 1 (NOD1), NOD2, and NLRP3 (also known as NALP3), which recognize components of bacterial peptidoglycan; NLRP3 is also involved in responses to components of gram-positive bacteria, including bacterial RNA and DNA, products of injured host cells such as uric acid, and noninfectious foreign substances, including asbestos and aluminum salts, which are widely used as the adjuvant alum ; NOD-like receptor C4 (NLRC4) (also known as IL-1-converting enzyme protease activating factor, or IPAF) is involved in responses to Salmonella and, like TLR5, the PAMP recognized is flagellin. NOD1 and NOD2 can activate MAP kinase and NF-κB pathways and proinflammatory cytokines in synergy with TLRs. By contrast, NLRP3 interacts with the mitochondrial antiviral signaling (MAVS) adaptor protein to activate the inflammasome, a protein platform that activates caspase-1, which, in turn, cleaves pro–IL-1 and pro–IL-18, allowing the secretion of the mature forms of these inflammatory cytokines. NLRC4 directly binds to caspase-1 and induces its activation. In addition to leading to IL-1 and IL-18 secretion, caspase-1 activation results in pores in the cell membrane and, as a consequence, a form of cell death known as pyroptosis.

Three NLR members that are not associated with inflammasomes—CTIIA, NLRC5, and NLRP10—have distinct roles in augmenting antigen presentation rather than in regulating innate immunity responses: CTIIA (class II activator) is a transcriptional coactivator that is essential for the upregulation of major histocompatibility complex (MHC) class II expression by IFN-γ ; NLRC5, also known as CITA, plays an analogous role for the IFN-γ–dependent enhancement of the MHC class I antigen presentation pathway by augmenting transcription of the MHC class I heavy and light chain genes and transporter associated with antigen processing (TAP)-1; NLRP10 appears to play an essential role in the migration of DCs from inflamed tissues via afferent lymphatics to the TD areas of draining lymph nodes.

Retinoic Acid–Inducible Gene-I–Like Receptors

There are three members of the retinoic acid–inducible gene-I (RIG)-I–like receptor (RLR) family: RIG-I, melanoma-differentiation–associated protein (MDA-5), and Laboratory of Genetics and Physiology 2 (LGP2). RLRs are present in the cytoplasm of nearly all mammalian cells, where they provide rapid, cell-intrinsic, antiviral surveillance. RIG-I is important for host resistance to a wide variety of RNA viruses, including influenza, parainfluenza, and hepatitis C virus, whereas MDA-5 is important for resistance to picornaviruses. RIG-I and MDA-5 interact with a common signaling adaptor (MAVS or IFN-β–promoter stimulator 1 [IPS-1]), which, like TRIF in the TLR3/4 pathway, induces the phosphorylation of IRF3 to stimulate type I IFN production.

C-Type Lectin Receptors

C-type lectin receptors (CLRs) are a family of surface proteins, which include DC-SIGN (dendritic cell–specific ICAM-3–grabbing nonintegrin), a receptor on DCs that is involved in their interaction with human immunodeficiency virus (HIV); the macrophage mannose receptor; dectin-1 and dectin-2, which are expressed by DCs and macrophages; and CLEC9A, which facilitates the uptake of necrotic material by DCs. Dectin-1 and dectin-2 bind to fungi, such as Candida, and activate cytokine production by a signaling pathway that involves the spleen tyrosine kinase (Syk) and caspase activation and recruitment domain 9 (CARD9), an adaptor protein that links Syk activity to the induction of NF-κB and activator protein-1 (AP-1), resulting in the production of cytokines. The recognition by dectins of the hyphal forms of fungi, which is indicative of invasion rather than colonization of a tissue, results in the preferential induction of cytokines, for instance, IL-1β, IL-6, and IL-23, that favor naïve CD4 T cells recognizing fungal antigens differentiating into Th17 cells. The secretion of IL-1β after dectin-1 engagement involves the processing of pro–IL-1β via a novel caspase-8 inflammasome rather than a canonical caspase-1–containing inflammasome. The importance of this pathway for fungal host defense is illustrated by the patients prone to opportunistic fungal infection who have genetic deficiencies of dectin, CARD9, or IL-17.

Cytoplasmic DNA Receptors

Cytoplasmic receptors for double-stranded DNA are found in leukocytes and other cell types that, upon binding of DNA, lead to the production of type I IFNs. The best characterized of these receptors involves cyclic guanosine monophosphate–adenosine monophosphate (GMP-AMP) synthase (cGAS) coupled with the STING (stimulator of interferon genes) adaptor protein. Upon DNA binding, cGAS synthesizes a novel cyclic dinucleotide, cyclic guanosine monophosphate–adenosine monophosphate (cGAMP), that binds to STING, which, in turn, activates kinase pathways that result in IRF3 phosphorylation and its translocation to the nucleus where type I IFN transcription is induced. The source of the DNA can be viral or bacterial and, unlike TLR9 DNA recognition, there is no preference for a particular nucleotide sequence, and the cytosine methylation status of CpGs also does not influence receptor activity. STING may also be coupled with other mammalian cytoplasmic DNA receptors, such as DDX41, but their role in host defense versus that of cGAS remains unclear. STING and innate immune responses can also be directly triggered by cyclic dinucleotides that are generated by bacteria, for instance, during intracellular infection with Listeria monocytogenes, but are biochemically distinct from the cyclic dinucleotides generated by cGAS.

Decoding the Nature of the Threat Through Combinatorial Receptor Engagement

The differing molecular components of specific microbes result in the engagement of different combinations of innate immune recognition receptors. The innate immune system uses combinatorial receptor recognition patterns to decode the nature of the microbe and then tailors the ensuing early innate response and the subsequent antigen-specific response to combat that specific type of infection. Extracellular bacteria engage TLRs 2, 4, and/or 5 on the cell surface and also activate NLRs, providing a molecular signature of this type of pathogen. This leads to the production of proinflammatory cytokines and IL-23 to recruit neutrophils and support the development of a T-helper (Th)17-type T-cell response (see “Differentiation of Activated Naïve T Cells into Effector and Memory Cells” ). Fungal products engage TLR2 and dectins, leading to a similar response. Conversely, virus recognition via TLRs 3, 7, 8, 9, and RLRs (for RNA viruses) and cGAS/STING (for DNA viruses) stimulates the production of type I IFNs and IFN-induced chemokines (e.g., CXCL10), which, in turn, induce and recruit CD8 and Th1 CD4 T cells. Nonviral intracellular bacterial pathogens also induce type I IFNs, such as via the TLR9 and cGAS/STING pathways, which collaborate with signals from cell surface TLRs and NLRs to induce the production of IL-12, resulting in Th1-type responses. The importance of these innate sensing mechanisms is underscored by strategies that pathogenic microbes have evolved to evade them and the mediators they induce.

Production

Polymorphonuclear leukocytes or granulocytes, including neutrophils, eosinophils, and basophils, are derived from CFU-GM. Neutrophils are the principal cells of interest in relation to defense against pyogenic pathogens. The first identifiable committed neutrophil precursor is the myeloblast, which sequentially matures into myelocytes, metamyelocytes, bands, and mature neutrophils. Myelocytes and more mature neutrophilic granulocytes cannot replicate and constitute the postmitotic neutrophil storage pool. The postmitotic neutrophil storage pool is an important reserve because these cells can be rapidly released into the circulation in response to inflammation. Mature neutrophils enter the circulation, where they remain for approximately 8 to 10 hours and are distributed equally and dynamically between circulating cells and those cells adherent to the vascular endothelium. After leaving the circulation, neutrophils do not recirculate and die after approximately 24 hours. Release of neutrophils from the marrow may be enhanced in part by cytokines, including IL-1, IL-17, and TNF-α, in response to infection or inflammation.

Neutrophil precursors are detected at the end of the first trimester, appearing somewhat later than macrophage precursors. Mature neutrophils are first detected by 14 to 16 weeks of gestation, but at midgestation, the numbers of postmitotic neutrophils in the fetal liver and bone marrow remain markedly lower than in term newborns and adults. By term, the numbers of circulating neutrophil precursors are 10- to 20-fold higher in the fetus and neonate than in the adult, and neonatal bone marrow also contains an abundance of neutrophil precursors. However, the rate of proliferation of neutrophil precursors in the human neonate appears to be near maximal, suggesting that the capacity to increase numbers in response to infection may be limited.

At birth, neutrophil counts are lower in preterm than term neonates and in neonates born by cesarean section without labor. Within hours of birth, the numbers of circulating neutrophils increase sharply. The number of neutrophils normally peaks shortly thereafter, whereas the fraction of neutrophils that are immature (bands and less mature forms) remains constant at about 15%. Peak counts occur at approximately 8 hours in infants born at greater than 28 weeks gestation and at approximately 24 hours in those born at less than 28 weeks gestation, then decline to a stable level by approximately 72 hours in those born without complications. Thereafter, the lower limit of normal for term and preterm neonates is approximately 2500 and 1000 per μL, respectively, and the upper limit of normal is approximately 7000 per μL for both term and preterm neonates.

Values may be influenced by a number of additional factors. Most important is the response to sepsis. Septic infants may have normal or increased neutrophil counts. Sepsis and other perinatal complications, including maternal hypertension, periventricular hemorrhage, and severe asphyxia, can cause neutropenia, however, and severe or fatal sepsis often is associated with persistent neutropenia, particularly in preterm neonates. Neutropenia may be associated with increased margination of circulating neutrophils, which occurs early in response to infection. However, neutropenia that is sustained often reflects depletion of the neonate’s limited postmitotic neutrophil storage pool. Septic neutropenic neonates in whom the neutrophil storage pool is depleted are more likely to die than are those with normal neutrophil storage pools. Leukemoid reactions also are observed at a frequency of approximately 1% in term neonates in the absence of an identifiable cause. Such reactions appear to reflect increased neutrophil production.

Circulating G-CSF levels in healthy infants are highest in the first hours after birth, and levels in premature neonates are generally higher than in term neonates. Levels decline rapidly in the neonatal period and more slowly thereafter. Plasma G-CSF levels tend to be elevated in infected neonates. Mononuclear cells and monocytes from midgestation fetuses and premature neonates generally produce less G-CSF and GM-CSF after stimulation in vitro than comparable adult cell types, whereas cells from term neonates produce amounts that are similar to or modestly less than those of adults.

Migration to Sites of Infection or Injury

After release from the bone marrow into the blood, neutrophils circulate until they are called upon to enter infected or injured tissues. Neutrophils adhere selectively to endothelium in such tissues but not in normal tissues. The adhesion and subsequent migration of neutrophils through blood vessels into tissues and to the site of infection result from a multistep process, which is governed by the pattern of expression on their surface of adhesion molecules and receptors for chemotactic factors and by the local patterns and gradients of adhesion molecule and chemotactic factors in the tissues.

The adhesion molecules involved in neutrophil migration from the blood into tissues include selectins, integrins, and the molecules to which they adhere. The selectins are named by the cell types in which they are primarily expressed: L-selectin by leukocytes, E-selectin by endothelial cells, and P-selectin by platelets and endothelial cells. L-selectin is constitutively expressed on leukocytes and appears to bind to tissue- or inflammation-specific carbohydrate-containing ligands on endothelial cells. E-selectin and P-selectin are expressed on activated not resting endothelial cells or platelets. E- and P-selectin bind to sialylated glycoproteins on the surface of leukocytes, including P-selectin glycoprotein ligand-1. L-selectin binds to glycoproteins and glycolipids, which are expressed on vascular endothelial cells in specific tissues. The integrins are a large family of heterodimeric proteins composed of an α and a β chain. The β ₂ integrins LFA-1 (CD11a-CD18) and Mac-1 (CD11b-CD18) play a critical role in neutrophil function because neutrophils do not express other integrins in substantial amounts. The β ₂ integrins are constitutively expressed on neutrophils, but their abundance and avidity for their endothelial ligands are increased after activation of neutrophils in response to chemotactic factors. Their endothelial ligands include intercellular adhesion molecule 1 (ICAM-1) and ICAM-2. Both are constitutively expressed on endothelium, but ICAM-1 expression is increased markedly by exposure to inflammatory mediators, including IL-1, TNF-α, and LPS.

Chemotactic factors may be derived directly from bacterial components, such as n -formylated-Met-Leu-Phe (fMLP) peptide; from activated complement, including C5a; and from host cell lipids, including leukotriene B ₄ (LTB ₄ ). In addition, a large family of chemotactic cytokines (chemokines) is synthesized by macrophages and many other cell types (see Tables 4-1 and 4-2 ). Chemokines constitute a cytokine superfamily, with more than 50 members known at present, most of which are secreted and of relatively low molecular weight. Chemokines attract various leukocyte populations, which bear the appropriate G protein–linked chemokine receptors. They can be divided into four families according to their pattern of amino-terminal cysteine residues: CC, CXC, C, and CX3C (X represents a noncysteine amino acid between the cysteines). A nomenclature for the chemokines and their receptors has been adopted, in which the family is first denoted (e.g., CC), followed by L for ligand (the chemokine itself) and a number, or followed by R (for receptor) and a number. Functionally, chemokines also can be defined by their principal function—homeostatic or inflammatory cell migration—and by the subsets of cells on which they act. Neutrophils are attracted by the subset of CXC chemokines that contain a glutamine-leucine-arginine motif, including the prototypical neutrophil chemokine CXLC8, also known as IL-8.

These adhesion molecules and chemotactic factors act in a coordinated fashion to allow neutrophil recruitment. In response to injury or inflammatory cytokines, E-selectin and P-selectin are expressed on the endothelium of capillaries or postcapillary venules. Neutrophils in the blood adhere to these selectins in a low-avidity fashion, allowing them to roll along the vessel walls. This step is transient and reversible unless a second, high-avidity interaction is triggered. If, at the time of the low-avidity binding, neutrophils also encounter chemotactic factors released from the tissues or from the endothelium itself, they rapidly upregulate the avidity and abundance of LFA-1 and Mac-1 on the neutrophil cell surface. This process results in high-avidity binding of neutrophils to endothelial cells, which, in the presence of a gradient of chemotactic factors from the tissue to the blood vessel, induces neutrophils to migrate across the endothelium and into the tissues by diapedesis.

The profound importance of integrin- and selectin-mediated leukocyte adhesion is illustrated by the genetic leukocyte adhesion deficiency (LAD) syndromes. In LAD 1, deficiency of the common β ₂ integrin chain (CD18) results in inability of leukocytes to exit the bloodstream and reach sites of infection and injury in the tissues. Affected patients are profoundly susceptible to infections with pathogenic and nonpathogenic bacteria and may present in early infancy with delayed separation of the umbilical cord, omphalitis, and severe bacterial infection without pus formation. Two related syndromes, LAD 2 and LAD 3, are due, respectively, to a defect in synthesis (fucosylation) of the carbohydrate selectin ligands and mutations in the gene encoding kindlin-3, a protein required for integrin activation.

Migration of Neonatal Neutrophils

The ability of neonatal neutrophils to migrate from the blood into sites of infection and inflammation is reduced or delayed, and the transition from a neutrophilic to mononuclear cell inflammatory response is delayed. This diminished delivery of neutrophils may result in part from defects in adhesion and chemotaxis.

Adhesion of neonatal neutrophils under resting conditions is normal or at most modestly impaired, whereas adhesion of activated cells is deficient. Adhesion and rolling of neonatal neutrophils to activated endothelium under conditions of flow similar to those found in capillaries or postcapillary venules is variable but on average approximately 50% of that observed with adult neutrophils. This decreased adhesion appears to reflect, at least in part, decreased abundance and shedding of L-selectin and decreased binding of neonatal neutrophils to P-selectin. Resting neonatal and adult neutrophils have similar amounts of Mac-1 and LFA-1 on their plasma membrane, but neonatal neutrophils have a reduced ability to upregulate expression of these integrins after exposure to chemotactic agents. Reduced integrin upregulation is associated with a parallel decrease in adhesion to activated endothelium or ICAM-1 suggesting that a deficit in adhesion underlies the diminished ability of neonatal neutrophils to migrate through endothelium into tissues, particularly in preterm neonates.

In nearly all studies in which neutrophil migration has been examined in vitro, chemotaxis of neonatal neutrophils was less than that of adult neutrophils. The response of neonatal neutrophils to a variety of chemotactic factors, including fMLP, LTB ₄ , and neutrophil-specific chemokines, including IL-8, is reduced. Chemotactic factor binding and dose response patterns of neonatal neutrophils appear similar to adult neutrophils, whereas downstream processes, including expression of Ras-related C3 botulinum toxin substrate 2 (Rac2), the increases in the free (nonprotein bound) intracellular calcium concentration [Ca ²⁺ ] _i and inositol phospholipid generation and the change in cell membrane potential are impaired. An additional factor may be the reduced deformability of neonatal neutrophils, which may limit their ability to enter the tissues after binding to the vascular endothelium. Decreased generation of chemotactic factors in neonatal serum may compound the intrinsic chemotactic deficits of neonatal neutrophils. However, the generation of other chemotactic agents, such as LTB ₄ , by neonatal neutrophils appears to be normal.

Phagocytosis

Having reached the site of infection, neutrophils must bind, phagocytose, and kill the pathogen. Opsonization greatly facilitates this process. Neutrophils express on their surface receptors for multiple opsonins, including receptors for the Fc portion of the IgG molecule (Fcγ receptors): FcγRI (CD64), FcγRIIA (CD32), and FcγRIIIB (CD16). Neutrophils also express receptors for activated complement components C3b and C3bi, which are bound by CR1, CR3 (CD11b-CD18), and CR4 (CD11c-CD18), respectively. Opsonized bacteria bind and cross link Fcγ and C3b-C3bi receptors. This cross-linkage transmits a signal for ingestion and for the activation of the cell’s microbicidal mechanisms.

Under optimal in vitro conditions, neutrophils from healthy neonates bind and ingest gram-positive and gram-negative bacteria as well as or only slightly less efficiently than an adult’s neutrophils. However, the concentrations of opsonins are reduced in serum from neonates, in particular, preterm neonates, and when concentrations of opsonins are limited, neutrophils from neonates ingest bacteria less efficiently than those from adults. Consistent with this finding, phagocytosis of bacteria by neutrophils from preterm, but not term, neonates is reduced compared with adult neutrophils when assayed in whole blood. Why neonatal neutrophils have impaired phagocytosis when concentrations of opsonins are limiting is incompletely understood. Basal expression of receptors for opsonized bacteria is not greatly different. Neutrophils from neonates, particularly preterm neonates, express greater amounts of the high-affinity FcγRI, lesser amounts of FcγRIII, and similar or slightly reduced amounts of FcγRIIA at birth, compared with adults; values in preterm neonates approach those of term neonates by 1 month of age. Expression of complement receptors on neutrophils from term neonates and adults is similar, but reduced expression of CR3 on neutrophils from preterm neonates has been reported in some studies. Neutrophils from preterm neonates also are less able to upregulate CR3 in response to LPS and chemotactic factors. Lower expression of proteins involved in the engulfment process, including Rac2 as noted above, may also contribute to impaired phagocytosis when concentrations of opsonins are limited.

Killing

After ingestion, neutrophils kill ingested microbes through oxygen-dependent and oxygen-independent mechanisms. Oxygen-dependent microbicidal mechanisms are of central importance, as illustrated by the severe compromise in defenses against a wide range of pyogenic pathogens (with the exception of catalase-negative bacteria) observed in individuals with a genetic defect in this system. Children with this disorder have a defect in one of several proteins that constitute the phagocyte oxidase, which is activated during receptor-mediated phagocytosis. The assembly of the oxidase in the plasma membrane results in the generation and delivery of reactive oxygen metabolites, including superoxide anion, hydrogen peroxide, and hydroxyl radicals. These oxygen radicals, along with the granule protein myeloperoxidase, are discharged into the phagocytic vacuole, where they collaborate in killing ingested microbes. In addition to this oxygen-dependent pathway, neutrophils contain other granule proteins with potent microbicidal activity, including the defensins HNPs 1 to 4, LL-37, elastase, cathepsin G, and a protein that binds selectively to and helps to kill gram-negative bacteria, bactericidal permeability-increasing (BPI) protein.

Oxygen-dependent and oxygen-independent microbicidal mechanisms of neutrophils from neonates and adults do not differ greatly. Generation of superoxide anion and hydrogen peroxide by neutrophils from term neonates is generally similar to or greater than by cells from adults in response to soluble stimuli. Although a modest reduction in the generation of reactive oxygen metabolites by neutrophils from preterm compared with term neonates was seen in response to some strains of coagulase-negative staphylococci, this was not observed with other strains nor with a strain of GBS and is of uncertain significance. By contrast, LPS primes adult neutrophils for increased production of reactive oxygen metabolites, but priming is much reduced with neonatal neutrophils, which could limit their efficacy in response to infection in vivo. The few studies of oxygen-independent microbicidal mechanisms suggest that neonatal neutrophils contain and release reduced amounts of BPI (approximately two to threefold) and lactoferrin (approximately twofold) compared with adult neutrophils but contain or release comparable amounts of myeloperoxidase, defensins, and lysozyme.

Consistent with these findings, killing of ingested gram-positive and gram-negative bacteria and Candida organisms by neutrophils from neonates and adults is generally similar. However, variable and usually mildly decreased bactericidal activity has been noted against Pseudomonas aeruginosa, S. aureus, and certain strains of GBS. Deficits in killing of engulfed microbes by neonatal neutrophils are more apparent at high ratios of bacteria to neutrophils, as is killing by neutrophils from sick or stressed neonates (i.e., those born prematurely or who have sepsis, respiratory impairment, hyperbilirubinemia, premature rupture of membranes, or hypoglycemia).

Neutrophils also function in trapping and killing extracellular bacteria in neutrophil extracellular traps (NETs), which consist of a complex of DNA, histones, granule enzymes, and antimicrobial proteins that are extruded from the cell. Neonatal neutrophils from both term and preterm infants are unable to form NETs and therefore have impaired extracellular bacterial killing in vitro.

Activation by Innate Immune Receptors

Although neutrophils were formally viewed by many as cells highly specialized for the uptake and killing of bacteria and fungi, human neutrophils also express a variety of innate immune receptors that are functional, including all TLRs except TLR3 and TLR7, NLRs (NOD2 and NLRP3), and RLRs (RIG-I and MDA-5). Moreover, neutrophils respond to TLR agonists or NOD2 ligands with production of IL-8 and other chemokines, superoxide generation, and shedding of L-selectin, and they also produce IL-1β after NLRP3 activation. In mice, neutrophil-derived IL-1β is sufficient for the formation of abscesses in S. aureus infection, with cytokine production involving both TLRs and NLRs. This suggests that neutrophils may be an important in vivo source of proinflammatory cytokines, at least in certain contexts. There is little information on neonatal neutrophil activation by innate immune receptors. One study of infants with severe RSV found that neutrophils of the blood and lung had reduced expression of TLR4. As TLR4 is a receptor for RSV G protein, it is plausible that this decreased expression could contribute to the disease severity.

Neutrophil Clearance and Resolution of Neutrophilic Inflammation

One to 2 days after egress from the bone marrow, neutrophils undergo apoptosis and are efficiently cleared by tissue macrophages without producing inflammation or injury. In the context of infection or sterile inflammation, their survival is prolonged by CSFs and other inflammatory mediators, allowing them to aid in microbial clearance, while at the same time augmenting or perpetuating tissue injury. Studies from several groups have shown that spontaneous and anti-Fas–induced apoptosis of isolated neonatal neutrophils is reduced when these cells are cultured in vitro. The greater survival of neonatal than adult neutrophils was associated with reduced expression of the apoptosis-inducing Fas receptor and pro-apoptotic members of the Bcl-2 family. The increased survival of neonatal neutrophils has led some to speculate that this may help to compensate for the neonate’s limited neutrophil storage pool in protection against infection, but this might also contribute to persistent untoward inflammation and tissue injury.

Effects of Immunomodulators

After systemic treatment with G-CSF and GM-CSF, the numbers of neutrophils increase in neonates, as does expression of CR3 on these cells. The increased numbers likely reflect increased production and survival. GM-CSF and IFN-γ enhance the chemotactic response of neonatal neutrophils, although, at high concentrations, GM-CSF inhibits chemotaxis while augmenting oxygen radical production. Of potential concern, indomethacin, which is used clinically to facilitate ductal closure in premature neonates, impairs chemotaxis of cells from term and preterm neonates.

Production and Differentiation of Monocytes and Resident Tissue Macrophages

Together, monocytes and tissue macrophages are referred to as mononuclear phagocytes. Under steady-state conditions, monocytes are released from the bone marrow within 24 hours and circulate in the blood for 1 to 3 days before moving to the tissues, where they differentiate into tissue macrophages. All monocytes express CD14, which serves as a co-receptor for recognition of LPS by TLR4/MD-2. CD14 is commonly used as a lineage marker for these cells because it is the only cell type that expresses it in high amounts. Additional commonly used mononuclear phagocyte markers include CD68, a glycoprotein that binds to low density lipoprotein, and CD163, a scavenger receptor for hemoglobin-haptoglobin complexes. Monocytes also express histocompatibility leukocyte antigen (HLA)-DR (MHC class II) and can present antigens to CD4 T cells, which is discussed in detail in the section “Antigen Presentation by Classical Major Histocompatibility Complex Molecules,” although the amounts expressed and efficiency of antigen presentation are less than by DCs. Monocytes are heterogeneous. In the healthy adult circulation, CD14 ⁺ monocytes can be divided into three subsets based on their levels of expression of CD14 and CD16 (a Fc receptor for IgG)—CD14 ⁺⁺ CD16 ⁻ , CD14 ⁺ CD16 ⁺ , and CD14 ^dim CD16 ⁺⁺ , which comprise approximately 85%, 6%, and 9% of total blood monocytes, respectively. CD14 ⁺⁺ CD16 ⁻ monocytes have relatively low levels of surface expression of CD11c, CD80/86, CD163, and HLA-DR compared with CD14 ⁺ CD16 ⁺ monocytes. Both CD11c, a β ₂ integrin molecule, and L-selectin (CD62-L), are involved in leukocyte adhesion, and CD80 and CD86 are important for co-stimulation to T cells. The CD14 ^dim CD16 ⁺⁺ cell subset, which was only recently described, is highly responsive to viral nucleic acids but not bacterial products, such as LPS, although this point is controversial. Studies in humanized mice suggest that CD14 ^dim CD16 ⁺⁺ monocytes “patrol” the endothelium in an L-selectin–dependent manner.

Macrophages are resident in tissues throughout the body and include distinct populations of alveolar macrophages (lung), histiocytes (interstitial connective tissue), Kupffer cells (liver), microglia (brain), and osteoclasts (bone). Macrophages have multiple functions, including the clearance of dead host cells (efferocytosis), phagocytosis and killing of microbes, secretion of inflammatory mediators, and presentation of antigen to T cells. They can also act to limit or suppress inflammation in certain contexts and promote tissue repair. With the exception of inflammatory macrophages, which are similar to neutrophils in being short-lived, most macrophage populations are relatively long-lived cells, but their life span may vary from days to possibly years, depending on the particular macrophage population.

Macrophages are detectable as early as 4 weeks of fetal life in the yolk sac and are found shortly thereafter in the liver and then in the bone marrow. Recent studies in mice indicate that the major tissue subsets, including liver Kupffer cells and lung alveolar, splenic, and peritoneal macrophages, are established before birth (21 days of gestation) and are subsequently maintained by proliferation in situ rather than by replenishment from circulating monocytes. These findings, along with others documenting tissue macrophage proliferation in the mouse in diverse inflammatory responses, for instance, tissue nematode infection and atherosclerosis, suggest that tissue macrophage proliferation could be an important part of human mononuclear cell homeostasis and regulation by inflammation.

The capacity of the fetus and the neonate to produce monocytes is at least as great as that of adults. The numbers of monocytes per volume of blood in neonates are equal to or greater than those in adults. Cord blood has a similar fraction of monocytes that are CD14 ⁺⁺ CD16 ⁻ and CD14 ⁺ CD16 ⁺ as adult peripheral blood. Cord blood monocytes express approximately 50% as much HLA-DR as adult monocytes, and a larger fraction of neonatal monocytes lack detectable HLA-DR ; this reduced expression in cord blood applies to both the CD14 ⁺⁺ CD16 ⁻ and CD14 ⁺ CD16 ⁺ subsets. In contrast, other surface markers are expressed similarly, with the CD14 ⁺ CD16 ⁺ subset of both cord blood and adult peripheral blood having significantly higher levels of CD11c, CD80/86, and CD163 than CD14 ⁺⁺ CD16 ⁻ monocytes. To the best of our knowledge, there have been no studies reporting the frequency and phenotype of the recently described CD14 ^dim CD16 ⁺⁺ monocyte subset in human neonates.

The numbers of tissue macrophages in human neonates are not well characterized. Limited data in humans, which are consistent with data in nonhuman primates and other mammals, suggest that the lung contains few macrophages until shortly before term. Postnatally, the numbers of lung macrophages increase to adult levels by 24 to 48 hours in healthy monkeys. A similar increase occurs in humans, although the data are less complete and by necessity derived from individuals with clinical problems necessitating tracheobronchial lavage. The blood of premature neonates contains increased numbers of pitted erythrocytes or erythrocytes containing Howell-Jolly bodies, suggesting that the ability of splenic and liver macrophages to clear these effete cells, and perhaps microbial cells, may be reduced in the fetus and premature infant.

Migration to Sites of Infection and Delayed Hypersensitivity Responses

Like neutrophils, mononuclear phagocytes express the adhesion molecules L-selectin and β ₂ integrins. These cells also express substantial amounts of the α ₄ β ₁ integrin very late antigen-4 (VLA-4), allowing them, unlike neutrophils, to adhere efficiently to endothelium expressing vascular cell adhesion molecule 1 (VCAM-1), the ligand for VLA-4. Interaction of VLA-4 with VCAM-1 allows monocytes to enter tissues in states in which there is little or no neutrophilic inflammation. Chemokines that are chemotactic for neutrophils are not generally chemotactic for monocytes, and vice versa. Monocytes respond to a range of CC chemokines, such as CCL2 (monocyte chemotactic protein-1 [MCP-1]).

The acute inflammatory response is characterized by an initial infiltration of neutrophils that is followed within 6 to 12 hours by influx of mononuclear phagocytes. Some inflammatory responses, including delayed-type hypersensitivity (DTH) reactions induced by the injection of antigens, for instance, purified protein derivative (PPD), to which the individual is immune (i.e., has developed an antigen-specific T-cell response) are characterized by the influx of mononuclear phagocytes and lymphocytes with very minimal or no initial neutrophilic phase.

The influx of monocytes into sites of inflammation, including DTH responses, is delayed and attenuated in neonates compared with adults. This is true even when antigen-specific T-cell responses are evident in vitro, suggesting that decreased migration of monocytes and lymphocytes into the tissues is predominantly responsible for the poor response in neonates. Whether this delay results from impaired chemotaxis of neonatal monocytes or impaired generation of chemotactic factors or both is unresolved

Antimicrobial Properties of Monocytes and Macrophages

Although neutrophils ingest and kill pyogenic bacteria more efficiently, resident macrophages are the initial line of phagocyte defense against microbial invasion in the tissues. When the microbial insult is modest, these cells are sufficient. If not, they produce cytokines and other inflammatory mediators to direct the recruitment of circulating neutrophils and monocytes from the blood. Monocytes and macrophages express receptors that allow them to bind to microbes. These receptors include FcγRI, II, and III, which bind IgG-coated microbes ; FcαR, which binds IgA-coated microbes ; and the CR1 and CR3 receptors, which bind microbes coated with C3b and C3bi, respectively. Microbes bound through these receptors are efficiently engulfed by macrophages and once ingested can be killed by microbicidal mechanisms using many methods similar to those used by neutrophils. Mononuclear phagocytes generate reactive oxygen metabolites but in lesser amounts than neutrophils. Circulating monocytes, but not tissue macrophages, contain myeloperoxidase, which facilitates the microbicidal activity of hydrogen peroxide. The expression of microbicidal granule proteins differs somewhat in mononuclear phagocytes and neutrophils; for instance, human mononuclear phagocytes express β-defensins but not α-defensins, whereas neutrophils express both.

The microbicidal activity of resident tissue macrophages is relatively modest, especially compared with the robust activity of neutrophils. This limited activity may be important in allowing macrophages to remove dead or damaged host cells and small numbers of microbes without excessively damaging host tissues. However, in response to infection, macrophage microbicidal and proinflammatory functions are enhanced in a process referred to as classical (M1) macrophage activation. Macrophage activation results from the integration of signals from TLRs and other innate immune pattern recognition receptors and receptors for activated complement components, immune complexes, cytokines, and ligands produced by other immune cells, including IFN-γ, CD40 ligand, TNF-α, and GM-CSF.

The increased antimicrobial activity of M1-activated macrophages results in part from increased expression of FcγRI, enhanced phagocytic activity, and increased production of reactive oxygen metabolites. Other antimicrobial mechanisms induced by activation of these cells include the catabolism of tryptophan by indoleamine 2,3-dioxygenase, scavenging of iron, and production of nitric oxide and its metabolites by the inducible nitric oxide synthase (iNOS). The last is a major mechanism by which activated murine macrophages inhibit or kill a variety of intracellular pathogens. However, the role of nitric oxide in the antimicrobial activity of human macrophages is controversial, although polymorphisms of the iNOS gene locus in humans have been associated with the host susceptibility to Mycobacteria tuberculosis. Activated mononuclear phagocytes also secrete a number of noncytokine products that are potentially important in host defense mechanisms. These include complement components, fibronectin, and lysozyme.

Classical (M1) activation of macrophages plays a critical role in defense against infection with intracellular bacterial and protozoan pathogens that replicate within phagocytic vacuoles. Support for this notion comes from studies in humans and mice with genetic deficiencies that impair the activation of macrophages by IFN-γ. Humans with genetic defects involving IL-12, which induces IFN-γ production by NK and T cells, the IL-12 receptor, the IFN-γ receptor or the transcription factor STAT1, which is activated via this receptor, suffer unduly from infections with mycobacteria and Salmonella. Treatment of humans with antagonists of TNF-α also impair antimycobacterial defenses. Patients with the hyper-IgM syndrome, which is due to a defect in CD40 ligand, or with autosomal recessive hyper-IgM due to a mutation in CD40, the receptor for CD40 ligand, are predisposed to disease caused by Pneumocystis jirovecii and Cryptosporidium parvum, in addition to the problems they experience from defects in antibody production (see section CD4 T-Cell Help for Antibody Production ). These findings are consistent with the notion that IFN-γ, TNF-α, and CD40 ligand–mediated M1 macrophage activation is important in host defense against these pathogens and that these molecules activate macrophages, at least in part, in a nonredundant manner.

By contrast to this canonical pathway of macrophage activation, macrophages exposed to cytokines produced by Th2 cells, such as IL-4, which are induced by infection with parasitic helminths and as part of allergic responses, are activated in an alternative manner. These alternatively activated macrophages, also termed M2 cells, dampen acute inflammation, impede the generation of reactive nitrogen products, limit proinflammatory T-cell responses, and foster fibrosis through the production of arginase and other mediators. Although best characterized in mice, this alternative pathway appears to be relevant in humans as well, and a recent study suggests that there is substantial additional diversity in the function of human macrophage subsets beyond the M1 and M2 paradigms.

Mononuclear Phagocytes Produce Cytokines and Other Mediators That Regulate Inflammation and Immunity

Monocytes and macrophages produce cytokines, chemokines, colony-stimulating factors and other mediators in response to ligand binding by TLRs and other pattern recognition receptors expressed by these cells (described in the next section), cytokines produced by other cell types, activated complement components and other mediators, and engagement of CD40 on their surface by CD40 ligand expressed on activated CD4 T cells. These include the cytokines IL-1α, IL-1β, TNF-α, and IL-6, which induce the production of prostaglandin E2, which, in turn, induces fever, accounting for the antipyretic effect of drugs that inhibit prostaglandin synthesis. Fever may have a beneficial role in host resistance to infection by inhibiting the growth of certain microorganisms and by enhancing host immune responses. TNF-α, IL-1, and IL-6 also act on the liver to induce the acute-phase response, which is associated with decreased albumin synthesis and increased synthesis of certain complement components, fibrinogen, CRP, and MBL. G-CSF, GM-CSF, and macrophage-specific colony-stimulating factor (M-CSF) enhance the production of their respective target cell populations, increasing the numbers of phagocytes available. At the sites of infection or injury, TNF-α and IL-1 increase endothelial cell expression of adhesion molecules, including E-selectin and P-selectin, ICAM-1, and VCAM-1, increase endothelial cell procoagulant activity and enhance neutrophil adhesiveness by upregulating β ₂ integrin expression. IL-6 may help to terminate neutrophil recruitment into tissues and to facilitate a switch from an inflammatory infiltrate rich in neutrophils to one dominated by monocytes and lymphocytes. IL-8 and other Glu-Leu-Arg (ELR)-containing CXC chemokines enhance the avidity of neutrophil β ₂ integrins for ICAM-1 and attract neutrophils into the inflammatory-infectious focus; CC chemokines play a similar role in attracting mononuclear phagocytes and lymphocytes. These and additional factors contribute to edema, redness, and leukocyte infiltration, which characterize inflammation.

In addition to secreting cytokines that regulate the acute inflammatory response and play a crucial role in host defense to extracellular bacterial and fungal pathogens, monocytes and macrophages (and DCs; see later) produce cytokines that mediate and regulate defense against intracellular viral, bacterial, and protozoan pathogens. Type I IFNs directly inhibit viral replication in host cells, as do IFN-γ and TNF-α. IL-12, IL-23, and IL-27 are members of a family of heterodimeric cytokines that help to regulate T-cell and NK-cell differentiation and function. IL-12 is composed of IL-12/23 p40 and p35; IL-23 is composed of IL-12/23 p40 and p19; and IL-27 is composed of Epstein-Barr virus–induced gene 3 protein (EBI-3) and p28. IL-12, in concert with IL-15 and IL-18, enhance NK-cell lytic function and production of IFN-γ and facilitate the development of type 1 CD4 T helper (Th1) and CD8 effector T cells, which are discussed more fully in the section “Differentiation of Activated Naïve T Cells into Effector and Memory Cells” and which play a critical role in control of infection with intracellular bacterial, protozoal, and viral pathogens. IFN-γ activates macrophages, allowing them to control infection with intracellular pathogens, and enhances their capacity to produce IL-12 and TNF-α, which, in turn, amplifies IFN-γ production by NK cells and causes T cells to differentiate into IFN-γ–producing Th1 T cells. IL-27 also facilitates IFN-γ production, while at the same time inducing the expression of IL-10, which dampens inflammatory and Th1 T-cell responses to limit tissue injury. By contrast, IL-23 favors IL-17–producing Th17 T-cell responses, in which IL-17 promotes neutrophil production, acute inflammation, and defense to extracellular pathogens.

The production of cytokines by mononuclear phagocytes normally is restricted temporally and anatomically to cells in contact with microbial products, antigen-stimulated T cells, or other agonists. When excess production of proinflammatory cytokines occurs systemically, septic shock and disseminated intravascular coagulation may ensue, underscoring the importance of closely regulated and anatomically restricted production of proinflammatory mediators. Tight control of inflammation normally is achieved by a combination of positive and negative feedback regulation. For example, TNF-α, IL-1, and microbial products that induce their production also cause macrophages to produce cytokines that attenuate inflammation and dampen immunity, including IL-10 and the IL-1 receptor antagonist. Inflammation is also attenuated by the production of antiinflammatory lipid mediators (resolvins), including the lipoxins, protectins, and maresins.

Cytokine Production, Toll-like Receptors, and Regulation of Innate Immunity and Inflammation by Neonatal Monocytes and Macrophages

Much of the older literature suggested that blood mononuclear cells (BMCs) obtained from cord blood or neonatal peripheral blood were less efficient in general than adult BMC in the production of cytokines in response to LPS, other TLR ligands, or whole bacteria. BMCs consist of monocytes, DCs, and B, T, and NK lymphocytes. Because of the greater than 10-fold abundance of monocytes compared with DCs in blood and BMCs, and the relatively limited production by lymphocytes of cytokines in response to TLR agonists, monocytes were likely to be the predominant source for most of these cytokines. The recent development of routine flow cytometric analysis of 6- to 12-color monoclonal antibody staining for intracellular cytokines and surface markers applied to BMC now permits an assessment of the levels of production of multiple cytokines per cell for each cell type. Also, because of simplicity and a desire to minimize manipulations that might activate or alter blood leukocyte function, many recent studies have been done with whole blood to which TLR ligands are added directly ex vivo, with cytokine production evaluated by enzyme-linked immunosorbent assay (ELISA) or bead-based fluorescent assays or, in some cases, by multiparameter flow cytometry. The findings for recent flow cytometric studies using whole blood or BMC will be emphasized here, although secreted cytokine levels will also be included where the likely cell source can be inferred. As TLR3 and TLR9 are expressed by certain DC population but not monocytes, and monocytes, in general, produce type I IFN poorly, the results for type I IFN production and other cytokines in response to TLR3 or TLR9 ligands in whole-blood assays will be discussed separately in the section “Dendritic Cells: The Link Between Innate and Adaptive Immunity.”

The preponderance of the currently available data does not support the notion of a general inability of neonatal monocytes (and, as discussed later, DCs) to produce cytokines, but rather suggests a difference in the nature of their response.

There is a clear, substantial, and with rare exception, consistent deficit in the production of IL-12p70 and IFN-γ in response to TLR agonists that act on monocytes (TLR2, TLR4, TLR7/8). In assays using whole blood or BMC, the IFN-γ produced is probably mainly derived from NK cells responding to IL-12p70 rather than from T cells. These deficits of cord blood leukocytes are likely cell intrinsic because they are consistently observed with either whole-blood assays or assays of BMC cultured with heterologous adult serum or with fetal calf serum. As discussed in the section “Dendritic Cells: The Link Between Innate and Adaptive Immunity,” there are also marked reductions in type I IFN production by DCs in cord blood. Because type I IFNs synergize with IL-12p70 and IFN-γ in promoting Th1 development from naïve CD4 T cells, this cytokine profile combined with T-cell intrinsic mechanisms, which are discussed later, would be expected to limit the generation of Th1 effectors in neonates, as has been observed in neonatal HSV infection. As LPS-induced IL-12p70 production by monocytes is itself markedly increased by IFN-γ, it is noteworthy that IL-12p70 secretion and IL-12p35 mRNA expression in cord blood after combined LPS and IFN-γ stimulation is decreased, and this decrease persisted until at least 1 month of age. Whole-blood assays indicate that adult levels of production of IL-12p70 in response to LPS are achieved by 6 months of age. A large body of evidence, including from human genetic immunodeficiencies, has clearly demonstrated that IL-12p70 and IFN-γ are essential for Th1-type immune control of pathogens, especially mycobacteria. Thus it is plausible that these antigen-presenting cell (APC) limitations in Th1 cytokine production are an important mechanism contributing to the well-described vulnerability of the neonate and young infant to Mycobacterium tuberculosis infection.

A similar deficit in TNF-α production is evident in response to stimulation with TLR2, TLR3, TLR4, and TLR5 agonists, particularly when whole blood is used or BMC or monocytes are cultured in high (≥50%) concentrations of neonatal serum. In contrast, TNF-α production in response to TLR8 agonists or whole gram-positive or gram-negative bacteria is similar. Further, in flow cytometric analysis of BMC cultured in adult serum, production of TNF-α by cord blood and adult peripheral blood monocytes was similar. As discussed later, these findings are explained at least in part by the presence of an inhibitory factor in cord blood plasma and suggest that any cell-intrinsic limitations in TNF-α production by cord blood monocytes may be relatively subtle.

In any case, with the exception of TNF-α, production by neonatal cells of cytokines central to host defense against extracellular bacterial and fungal pathogens, acute inflammation, and Th17-type responses appear not to be greatly decreased and, in some cases, is more robust. IL-1 production by cells from term neonates and adults is similar or, at most, marginally reduced. In response to TLR agonists, neonatal monocytes produced equal or greater amounts of IL-6, IL-23, and IL-10, compared with adult cells, but were less able to produce multiple cytokines simultaneously. In one recent longitudinal study of the ontogeny of TLR-mediated cytokine responses in South African infants, this high level of production of IL-6 and IL-23, which was particularly robust in response to TLR2 agonists, gradually declined in the first year of life to adult peripheral blood levels. As IL-1, IL-6, and IL-23 production by APCs instructs the differentiation of naïve CD4 T cells into Th17-type effector cells, these findings are in agreement with apparently intact Th17 immunity in the neonate. Although some early studies suggested that the production of the immunoregulatory and antiinflammatory cytokine IL-10 by neonatal cells was reduced, most studies have found greater production of IL-10 by whole blood and equal or greater production by BMCs. These high levels of IL-10 in whole-blood assays, which are particularly prominent after TLR2 stimulation, gradually decline to adult levels in the first year of life, indicating that they are not merely a feature of cord blood leukocytes.

The basis for lower production of certain cytokines by neonatal monocytes and macrophages in response to microbial products that signal through TLRs is incompletely understood. Cell surface expression of TLR2, TLR4, and TLR8 by adult and neonatal monocytes is similar in most studies, and expression of CD14, which facilitates responses to LPS and TLR2 agonists, is similar or, at most, slightly reduced on neonatal monocytes. Monocytes from premature infants express less TLR2 and TLR4 at birth but increase their expression to amounts comparable to term infants by 2 weeks of age. By reverse-transcriptase polymerase chain reaction (RT-PCR) analysis, neonatal and adult monocytes contain similar amounts of TLRs 1 to 9, MD-2, CD14, MyD88, TIR domain-containing adaptor protein (TIRAP), and interleukin-1 receptor-associated kinase 4 (IRAK4) mRNA. One group reported that neonatal monocytes have reduced amounts of MyD88 protein but found no difference between adult and neonatal monocytes in LPS-induced activation of ERK1/2 and p38 kinases and phosphorylation and degradation of inhibitor of kappa B (IκB), events that are downstream of MyD88. Moreover, decreased expression of MyD88 is not sufficient to explain the diminished induction of HLA-DR and CD40 on neonatal monocytes in response to LPS, or of CD40 and CD80 on neonatal dendritic cells in response to LPS and polyinosinic:polycytidylic (poly [I:C]) acid, because these TLR ligands induce co-stimulatory molecules by TRIF- and type I IFN–dependent but MyD88-independent pathways.

Resolution of Mononuclear Phagocytic Inflammation

The resolution of mononuclear phagocytic inflammation involves the secretion of antiinflammatory cytokines, such as IL-10, IL-1 receptor antagonist, and resolvin lipid mediators. Phagocytosis-induced cell death (PICD) after ingestion of bacteria may also potentially be beneficial or harmful to the host, depending on whether cell death results in the death of the ingested microbe or its release into the extracellular space. As was observed for neonatal neutrophils, cord blood monocytes had reduced PICD compared with adult peripheral blood monocytes, which could contribute to persistent inflammation and sequelae, such as bronchopulmonary dysplasia and periventricular leukomalacia, in septic neonates.

Neonatal Conventional Dendritic Cells

Conventional dendritic cells overall constitute a similar fraction (0.5%-1.0%) of blood mononuclear cells in neonates, children, and adults, but cDCs only constitute about 25% of the total of circulating DCs in neonates, whereas cDCs constitute about 75% of the total in adults. The absolute numbers of cDCs remain constant from the neonatal period into adulthood, whereas the fraction and absolute numbers of pDCs decline with increasing postnatal age, reaching numbers similar to adults at greater than or equal to 5 years of age. To the best of our knowledge, the relative proportion of circulating cDCs in the newborn that are CD1c ⁺ and CD141 ⁺ has not yet been reported. It is likely that the CD1c ⁺ cDC subset predominates in the neonatal circulation as in the adult circulation (10:1 ratio with respect to CD141 ⁺ cells), so that the results of cDC (Lin ⁻ CD11c ⁺ ) studies using cord blood or infant peripheral blood mainly reflect the properties of the CD1c ⁺ subset.

Older flow cytometric studies reported that expression of MHC class II (HLA-DR), CD40, CD80, and CD86 on cord blood and adult peripheral blood cDCs were similar, although a more recent study found that the basal surface expression of HLA-DR and CD80 on cord blood cDCs was substantially decreased compared with circulating adult cDCs. Although cord blood cDCs can stimulate allogeneic cord blood T cells in vitro, it is unclear whether neonatal cDCs are as proficient as adult cDCs in processing and presenting foreign antigens to T cells. Limitations in cord blood cDC antigen presentation are plausible because cord blood cDCs do not upregulate CD40 and CD80 to the same degree as adult cDCs in response to agonists for TLR2/6 (macrophage-activated lipopeptide [MALP]), TLR3 (poly I:C), TLR4 (LPS), or TLR7, whereas TLR8 agonists induce equivalent increases in expression. Decreased maturation of cord blood cDCs was also observed in response to pertussis toxin.

Flow cytometric studies of cytokine expression by cord blood cDCs indicate that they were approximately 50% as efficient as adult cDCs at producing TNF-α in response to LPS, whereas TNF-α production in response to TLR8 agonists was similar. In contrast to TNF-α, IL-1α and IL-6 production by adult cDCs and cDCs from cord blood of term infants in response to LPS was similar, as was production of IL-12/IL-23p40 in response to most TLR agonists. However, cDCs in the cord blood of premature infants have markedly lower responses to these stimuli, which is paralleled by reduced secretion of IL-23 by cord blood mononuclear cell cultures. Given the importance of IL-23 production by cDCs in instructing naïve CD4 T cells for differentiation into Th17 effector cells, it is plausible that premature infants may have compromised Th17 immunity, which may increase their risk of extracellular bacterial and fungal infection.

However, these limitations in cDC function present at birth may be quite transient. A recent longitudinal study of a cohort of South African infants that evaluated cDC cytokine accumulation of TNF-α, IL-6 and IL-12/23p40 by flow cytometry after whole-blood stimulation with TLR agonists found that the cytokine level per cell was higher at 2 weeks of age compared with adults. This enhanced cytokine accumulation was consistent in that it was observed for multiple cytokines (TNF-α, IL-6, and IL-12/23p40) and after stimulation with TLR2/1, TLR4, or TLR7/8 agonists. Of interest, in most cases, the cytokine level per cell gradually declined to adult levels by 1 year of age. This apparently rapid postnatal maturation cDC function may not be limited to cytokine production capacity because there are also marked increases in the cDC expression of HLA-DR and CD80 by 3 months of age compared with at birth.

Earlier studies using stimulation of whole blood or unfractionated blood mononuclear cells also indicated that cord blood cDC responses to TLR3 stimulation were substantially reduced compared with those of adults. TLR3-induced cDC function can be inferred after poly I:C stimulation of whole blood or blood mononuclear cells because poly I:C activates blood cells primarily via TLR3, and cDCs are the only cell type in blood that expresses substantial amounts of TLR3. Therefore the observation that poly I:C–stimulated cord blood and cord blood mononuclear cells had modestly diminished production of type I IFNs and markedly decreased production of IL-12p70 suggested that cord blood cDCs had decreased TLR3-dependent function. The production of IFN-γ by poly I:C–stimulated cord blood and cord blood mononuclear cells also likely reflected diminished TLR3-mediated IL-12p70 and type I IFN production by cDCs, which, in turn, resulted in diminished production of IFN-γ by cord blood NK cells (see Fetal and Neonatal Natural Killer Cell-Mediated Cytotoxicity and Cytokine Production ).

Several studies suggest that the reduced TLR3-mediated cDC function increases during the first month of life. For example, a study in The Gambia found that poly I:C stimulation of whole blood induced significantly higher levels of TNF-α and IFN-γ at 1 month of age compared with at birth (cord blood). Similarly, a study of Belgian infants showed a significant increase in IL-12p70 secretion after poly I:C whole blood stimulation at 1 month of age compared with at birth. In a longitudinal study of South African infants peripheral blood samples from 2-week-old infants had substantially higher levels of IL-12p70, IL-23, IFN-γ, and chemokines that are upregulated by IFN-γ (MCP-1 and IFN-γ–induced protein-10[IP-10]) than did adult blood samples, and these high levels declined to those of adults by approximately 1 year of age. In contrast to these studies, which indicated rapid postnatal acquisition of adult levels of competency for TLR3-induced cDC function, one study of Canadian infants found that poly I:C treatment of cord blood mononuclear cells induced levels of IL-12/23p40, IL-12p70, IFN-γ, and type I IFN that were substantially lower than those of adult peripheral blood mononuclear cells, and they remained so at 1 and 2 years of age. These different results, at least those in the first month of life, could reflect humoral factors, such as adenosine or maternally derived hormones, acting to reduce the cDC-derived production of cytokines in whole blood. However, the precise basis for these findings is unclear, and there is precedent for substantially different levels of TLR-induced cytokine production by cDCs when they are assayed as purified cells versus in whole-blood assays.

A recent study used transcriptional profiling to compare the response of DCs (a mixture of cDCs and pDCs) from cord blood and adult peripheral blood to incubation with RSV. Of interest, cord blood DCs had a transcriptional profile indicating increased transforming growth factor (TGF)-β– and TGF-β–dependent gene expression, and secreted increased levels of TGF-β compared with adult DCs. When RSV-infected DCs were co-cultured with autologous circulating T cells, the adult co-cultured cells secreted relatively high levels of IL-12p70, TNF-α, IL-2, IL-10, IL-13, and IFN-γ, whereas the cord blood co-cultured cells secreted low levels of these cytokines and, compared to the adult cell co-cultures, high levels of IL-1β, IL-4, IL-6, and IL-17. Blockade of TGF-β signaling using a chemical inhibitor in the cord blood co-cultures markedly increased IL-12p70 secretion and more modestly enhanced the levels of other cytokines, such as IL-1β and IL-6. This approach did not address to what extent the results are due to differences between adult and neonatal T-cell populations, that is, a substantial fraction of effector memory T-cells were already committed to the Th1 lineage in adult preparations, whereas the cord blood T-cells were predominantly uncommitted antigenically-naïve T-cells. The importance of cDCs versus pDCs for TGF-β production is also unclear. Despite these limitations, the findings suggest that cord blood cDCs in response to RSV secrete a cytokine milieu (high levels of IL-1β, IL-6, and TGF-β and low levels IL-12p70) that favor Th17 differentiation rather than Th1 differentiation, which could contribute to the immunopathogenesis of RSV infection in young infants.

Neonatal Plasmacytoid Dendritic Cells

Cells with the histologic features of immature pDCs are found in fetal lymph nodes between 19 to 21 weeks of gestation. Using flow cytometry, pDC-lineage cells (Lin ⁻ CD11c ⁻ CD4 ⁺ CD45RA ⁺ CD123 ⁺ ) have been identified in fetal liver and bone marrow as early as 16 weeks of gestation; these can be further divided into CD34 ⁺ and CD34 ⁻ subsets that are likely to be pro-pDCs and immature pDCs, respectively. Of interest, both fetal liver pro-pDCs and immature pDCs were capable of high levels of type I IFN production in response to stimulation with irradiated HSV-1, indicating this function of pDCs is established early in ontogeny.

The absolute concentration of pDCs in the term neonate is approximately twofold higher than in the adult and gradually declines after birth, reaching numbers similar to adults at greater than or equal to 5 years of age. The biologic significance of the predominance of pDCs in the neonatal circulation is uncertain. In the prematurely born neonate, there is a trend for a greater proportion of cord blood pDCs with a pro-pDC (CD34 ⁺ ) surface phenotype than is observed in adult blood.

Compared with adult pDCs, multiple studies have found that cord blood pDCs produce less type I IFNs in response to unmethylated CpG oligonucleotide, a potent TLR9 ligand. or to TLR7 agonists. Because pDCs appear to be the main sources of type I IFN in response to CpG oligonucleotide stimulation of whole blood or blood mononuclear cells, type I IFN production from these cells can be attributed to pDCs. Thus diminished type I IFN expression in 4-day-old neonates suggests that decreased pDC function persists until at least this age, These reductions in type I IFN production in response to TLR7 and TLR9 agonists are more pronounced in the cord blood of prematurely born infants and are associated with greater immaturity of premature infant pDCs, based on their decreased expression of CD304 (BDCA-4) and “immature” morphology by electron microscopy. The defect in neonatal pDC type I IFN production in response to CpG stimulation appears to result in part from impaired activation and translocation of IRF7 to the nucleus. Consistent with these findings, TNF-α is more modestly reduced, and production of IL-6 is comparable to adult pDCs.

In addition to diminished type I IFN responses, cord blood pDCs also express lower amounts of HLA-DR, CD40, CD80, CD86, and CCR7 after stimulation with TLR7 or TLR9 agonists than adult pDCs. Cord blood pDCs may also have a tendency for reduced survival during culture with CpG oligonucleotides, which could contribute to their reduced responses to these stimuli.

In older studies, type I IFN production by blood cells from neonates in response to direct viral stimulation was significantly diminished, for instance, to HSV, cytomegalovirus (CMV), and parainfluenza virus. As for stimulation by TLR7 or TLR9 ligands, this reduced type I IFN response to herpesviruses was associated with decreased IRF7 cytoplasmic to nuclear translocation. However, a recent report found that type I IFN production in response to influenza A virus (live or heat-inactivated), HIV, or HSV was similar for cord blood and adult peripheral blood pDCs, regardless of whether whole blood, blood mononuclear cells, or purified pDCs were used. The production of TNF-α and the chemokines CCL3 and CCL4 in this study were also similar for cord blood and adult peripheral blood pDCs by using these assay conditions. The reasons for these different findings remain unclear, but they raise the possibility that production of such cytokines by neonatal pDCs may be as competent as those of adults in response to strong viral stimulation. Therefore additional studies to determine pDC function ex vivo in the setting of neonatal viral infection will be of interest.

In limited studies, the age at which responses to TLR ligands become comparable to adult cells has been assessed. CpG-induced upregulation of HLA-DR and CD80 on pDCs did not reach adult levels until 6 to 9 months of age. pDC-derived chemokine production (interferon gamma induced protein 10 [IP-10] and monokine induced by gamma interferon [MIG]) in whole blood was lower during the first year of life compared with adult pDCs. Of interest, pDC-derived IL-6, IL-8, IL-10, and IL-1β were significantly higher than in adult cells from 3 months of age onward, suggesting that neonatal pDCs have a unique cytokine profile that may inhibit Th1 responses (i.e., IL-10) and promote Th17 responses (IL-6 and IL-1β). Consistent with this finding, which suggested a maturation of pDC function by 9 months of age, one study found that pDC production of TNF-α and IL-6 in response to either TLR7/8 or TLR9 agonists at 1 year of age was similar to those of adult pDCs.

Overview and Development

NK cells are large granular lymphocytes with cytotoxic function, which, unlike T and B lymphocytes, lack antigen-specific T-cell receptor (TCR) or B-cell receptor (BCR) receptors characteristic of the adaptive immune system and instead express a diverse array of activating and inhibitory receptors. Although NK cells are generally considered to be a component of the innate rather than adaptive immune system, in mouse models they can be primed by infection, such as with herpesviruses, and retain certain features of memory/effector T cells, such as enhanced cytotoxicity and cytokine secretion for at least several months postinfection. In clinical practice, NK cells are usually defined as lymphoid cells that express CD16 and CD56 but not CD3, that is, have a CD16 ⁺ CD56 ⁺ CD3 ⁻ surface phenotype. Virtually all circulating NK cells from adults also express the NK-cell–specific NKp30 and NKp46 receptors, along with CD2 and CD161, and approximately 50% express CD57, but these molecules are found on other lymphocyte types as well.

The fetal liver produces NK cells as early as 6 weeks of gestation, but the bone marrow is the major site for NK-cell production from late gestation onward. NK cells are derived from bone marrow cells that lack surface molecules specific for other cell lineages (i.e., CD34 ⁺ Lin ⁻ cells) but express CD7 or CD38. NK-cell development is dependent on the IL-15/IL-15 receptor (which consists of the IL-15 receptor α chain, the IL-2 receptor β chain, and the common γ chain)/Janus tyrosine kinase 3 (JAK3) pathway. Based on the lack of NK-cell development in genetic immunodeficiencies, human NK-cell development also requires adenylate kinase 2 and the GATA binding protein 2 transcription factor. In vitro studies suggest a NK lineage cell developmental sequence in which CD161 is acquired early in NK development. At the next developmental stage, NKp30, NKp46, 2B4, and NKG2D are expressed on the cell surface, followed by members of the killer cell inhibitor receptor (KIR) family, CD94-NKG2A, CD2, and CD56; the function of these molecules is discussed in the sections that follow.

NK cells are functionally defined by their natural ability to lyse virally infected or tumor target cells in a non–HLA-restricted manner that does not require prior sensitization. NK cells preferentially recognize and kill cells expressing ligands for activating receptors that are not antigen-specific in conjunction with reduced or absent expression of self-HLA class I molecules, a property referred to as natural cytotoxicity. These ligands for activating receptors are characteristically increased in response to stresses, such as infection or malignant transformation. This is in contrast with cytotoxic CD8 T cells, which are triggered to lyse targets after the recognition of foreign antigenic peptides bound to self-HLA class I molecules or self-peptides bound to foreign HLA class I molecules. NK cells also have the ability to kill target cells that are coated with IgG antibodies, a process known as antibody-dependent cellular cytotoxicity (ADCC). ADCC requires the recognition of IgG bound to the target cell by the NK-cell FcγRIIIB receptor (CD16).

Mature NK cells can be subdivided into CD56 ^hi CD16 ^lo and CD56 ^lo CD16 ^hi populations. CD56 ^hi CD16 ^lo cells usually are only a minority of mature NK cells in the circulation but express CCR7 and L-selectin and predominate in lymph node tissue. CD56 ^hi CD16 ^lo cells have limited cytotoxic capacity but produce cytokines and chemokines efficiently, whereas the inverse is true for CD56 ^lo CD16 ^hi NK cells. These features suggest that the CD56 ^hi CD16 ^lo subset could regulate lymph node T cells and DCs through cytokine secretion. Developmental studies suggest that CD56 ^hi CD16 ^lo NK cells are less mature than CD56 ^lo CD16 ^hi NK cells, but the precise precursor-product relationship of these subsets under various conditions in vivo has not been firmly established.

NK cells are particularly important in the early containment of viral infections, especially with pathogens that may initially avoid control by adaptive immune mechanisms. Infection of host cells by the herpesvirus group, including HSV, CMV, and varicella-zoster virus (VZV), and some adenoviruses leads to decreased surface expression of HLA class I molecules. Viral protein–mediated decreases in expression of HLA class I may limit the ability of CD8 T cells to lyse virally infected cells and to clonally expand from naïve precursors. These virus-mediated effects may be particularly important during early infection, when CD8 T cells with appropriate antigen specificity are present at a low frequency. By contrast, decreased HLA class I expression in conjunction with the expression of ligands for activating receptors facilitates recognition and lysis by NK cells. The importance of NK cells in the initial control of human herpesvirus infections is suggested by the observation that persons with selective deficiency of NK-cell numbers or function are prone to severe infection with HSV, CMV, and VZV.

Natural Killer Cell Receptors

NK-cell cytotoxicity is regulated by a complex array of inhibitory and activating receptor–ligand interactions with target cells ( Fig. 4-3 ). The expression of multiple combinations of inhibitory and activating receptors leads to a large amount of NK-cell diversity. By some estimates, there are 6000 to 30,000 different NK-cell populations within an individual. Individual NK cells in the bone marrow appear to undergo a tuning process so that the threshold for their activation is appropriate for their particular repertoire of receptors. NK-cell activation is inhibited by recognition of HLA class I molecules expressed on nontransformed, uninfected cells; this recognition is presumed to provide a net inhibitory signal that predominates over activating signals. Infection or other perturbations of the host target cell, such as malignant transformation, can reduce the amount of HLA class I on the cell surface, thereby reducing inhibitory signaling, and upregulate other molecules that promote NK-cell activation, such as MHC class I–related chains A and B (MICA and MICB).

There are two major families of inhibitory NK-cell receptors that recognize HLA class I molecules in humans: KIR and the CD94-containing C-type lectin families. KIRs with a long cytoplasmic domain transmit signals that inhibit NK-cell activation; most, although not all, NK cells express one or more inhibitory KIRs on their surface. Most NK cells, including all those not expressing any inhibitory KIRs, also express inhibitory CD94-NKG2A receptors. KIRs bind to polymorphic HLA-B, HLA-C, or HLA-A molecules, whereas CD94-NKG2A binds to HLA-E, which is monomorphic (HLA molecules are discussed later in the sections “Antigen Presentation by Classic Major Histocompatibility Complex Molecules” and “Nonclassic Antigen-Presentation Molecules”). Because HLA-E reaches the cell surface only when its peptide binding groove is occupied by hydrophobic peptides derived from the leader sequences of HLA-A, HLA-B, and HLA-C molecules, the amount of HLA-E on the cell surface reflects the overall levels of HLA-A, HLA-B, and HLA-C molecules on that cell.

In addition to CD94-NKG2A, a third group of inhibitory receptors that broadly recognize HLA class I molecules are the leukocyte immunoglobulin-like receptors B1 and B2 (LILRB1 and LILRB2). LILRB1 and LILRB2, also referred to as LIR1/CD85J and LIR2/CD85d, bind to HLA-A, HLA-B, and HLA-C molecules, as well as the nonconventional class I molecules HLA-E, HLA-F, and HLA-G. HLA-G is the only HLA class I molecule constitutively expressed on the surface of fetal trophoblasts. Thus the interaction of LILRB1 and LILRB2 with HLA-G is thought to protect the placenta from injury by maternal NK cells.

Countering the effects of these inhibitory receptors are multiple types of activating receptors. NKG2D is found on NK cells as well as in certain T-cell populations. NKG2D recognizes MICA, MICB, and UL16-binding proteins (ULBPs 1-4). MICA and MICB are nonclassical HLA class I molecules that are expressed on stressed or infected cells. ULBPs are a group of HLA class I–like molecules expressed on many cell types that were first identified and named based on their ability to bind to the human CMV UL16 viral protein. In human CMV infection, UL16 probably limits NK-cell– and T-cell–mediated activation by binding to ULBPs internally and preventing their surface expression on the infected cell.

NK cells also express NKp30, NKp44, and NKp46. NKp44 and NKp46 can trigger NK-cell cytotoxicity through their recognition of influenza virus hemagglutinin and Sendai (parainfluenza family) virus hemagglutinin-neuraminidase. These receptors also recognize ligands on tumor cells and cells infected with herpesviruses, but with the exception of NKp30, which recognizes the tumor ligand B7-H6, the definitive nature of the ligands is not known.

The proteins 2B4 (CD244) and NTBA are members of the signaling lymphocytic-activation molecule (SLAM) protein family and are expressed on most NK cells. 2B4 binds to CD48, whereas the ligand for NTBA remains unclear. Both 2B4 and NTBA engagement triggers NK-cell activation through SLAM-associated protein (SAP), an intracellular adapter protein that is mutated in the X-linked lymphoproliferative syndrome.

Finally, NK cells may express KIRs with short cytoplasmic tails, which unlike their counterparts with long cytoplasmic tails, activate NK cells by a DNAX-activating protein 12 (DAP12) signaling mechanism. Also, unlike CD94-NKG2A, CD94-NKG2C is an activating receptor complex that also activates via DAP12 signaling. These activating KIRs and CD94-NKG2C and their respective inhibitory forms have identical or very similar ligand specificities. How NK cells integrate the effects on natural cytotoxicity of these multiple inhibitory and activating receptors, particularly those that recognize the same or similar ligands, remains unclear.

Natural Killer Cell Cytotoxicity

After binding via adhesion molecule interactions and activation, NK cells release perforin and granzymes from preformed cytotoxic granules into a synapse formed between the NK cell and its target, leading to death by apoptosis of the target cell. NK cell–mediated cytotoxicity also may be mediated by Fas ligand or TRAIL (TNF-related apoptosis-inducing ligand) expressed on the activated NK cell surface. Fas–Fas ligand interactions appear not to be essential for human NK cell control of viral infections because persons with dominant-negative mutations of the Fas or Fas ligand genes develop autoimmunity but do not experience an increased severity of virus infections. In contrast with natural cytotoxicity, in which perforin/granzyme-dependent mechanisms appear to be predominant, ADCC appears to use both perforin/granzyme- and Fas ligand–dependent cytotoxic mechanisms.

Natural Killer Cell Cytokine Responsiveness and Dependence

NK cell proliferation and cytotoxicity are enhanced in vitro by cytokines produced by T cells (IL-2, IFN-γ), APCs (IL-1β, IL-12p70, IL-18, and type I IFNs), and nonhematopoietic cells (IL-15, stem cell factor, Flt3 ligand, IFN-β). IL-15, which appears critical for the development of NK cells, also promotes the survival of mature NK cells and, like IL-12, increases the expression of perforin and granzymes. A subset of NK cells found in mucosal-associated lymphoid tissues responds to IL-23 by producing cytokines, including IL-22, that help to protect the gut from bacterial pathogens.

Natural Killer Cell Cytokine and Chemokine Production

NK cells are also important producers of IFN-γ and TNF-α in the early phase of the immune response to viruses, and IFN-γ may promote the development of CD4 T cells into Th1 effector cells (see “Differentiation of Activated Naïve T Cells into Effector and Memory Cells” ). NK-cell–mediated IFN-γ production may be induced by the ligation of surface β1 integrins on the NK-cell surface, as well as by the cytokines IL-1, IL-12, IL-15, and IL-18, which are produced by DCs and mononuclear phagocytes. The combination of IL-12 and IL-15 also potently induces NK cells to produce the CC chemokine macrophage inhibitory protein-1α (MIP-1α; CCL-3), which may help to attract other types of mononuclear cells to sites of infection, where NK-cell–mediated lysis takes place. NK cells from HIV-infected persons also are able to produce a variety of CC chemokines, including MIP-1α, MIP-1β (CCL-4), and RANTES (regulated on activation, normal T-cell expressed and secreted; CCL-5) in response to treatment with IL-2 alone; these chemokines may help prevent HIV infection of T cells and mononuclear phagocytes by acting as antagonists of the HIV co-receptor CCR5. NK cells also can be triggered to produce a similar array of cytokines during ADCC in vitro, but the role of such ADCC-derived cytokines in regulating immune responses in vivo is poorly defined. Some of the cytokine-dependent mechanisms by which NK cells, T cells, and APCs may influence each other’s function, such as in response to infection with viruses and other intracellular pathogens, are summarized in Figure 4-4 .

Natural Killer Cells of the Maternal Decidua and Human Leukocyte Antigen G

The maternal decidua contains a prominent population of NK cells, which may help contribute to the maintenance of pregnancy. NK cells belonging to the CD56 ^hi CD16 ^lo subset, which have a high capacity for cytokine production but low capacity for cytotoxicity, predominate. Murine studies suggest that maternal NK-cell–derived cytokines, such as IFN-γ, may help to remodel the spiral arteries of the placenta. Although the NK-cell populations of the decidua have a low capacity for cytotoxicity, their presence in a tissue lacking expression of HLA-A, HLA-B, and HLA-C molecules could potentially contribute to placental damage and fetal rejection. As noted earlier, the expression by human fetal trophoblast of HLA-G is thought to protect this tissue from attack by maternal NK cells through binding to the inhibitory receptors LILRB1 and LILRB2.

Natural Killer–Cell Numbers and Surface Phenotype in the Fetus and Neonate

Circulating NK cells become increasingly abundant during the second trimester, and at term, their numbers in the neonatal circulation (approximately 15% of total lymphocytes) are typically the same as or greater than in adults. Relatively high frequencies of NK cells are found in the fetal liver, lung, and spleen, whereas frequencies in the bone marrow and mesenteric lymph nodes are relatively low. The fraction of blood neonatal and adult NK cells that are CD56 ^hi CD16 ^−/lo (approximately 10%) and CD56 ^lo CD16 ⁺ NK cells (approximately 90%) is similar. Earlier studies of cell surface molecule expression on neonatal NK cells varied in their conclusions regarding the expression of molecules involved with adhesion, activation, inhibition, and cytotoxicity by neonatal NK cells. More recent studies using newer and more reliable methods suggest that neonatal NK cells have decreased expression of ICAM-1 but similar expression of other adhesion molecules, similar or greater expression of the inhibitory CD94-NKG2A complex but reduced expression of the inhibitory LILRB1 (LIR1) receptor, similar or greater expression of the activating NKp30 and NKp46 receptors, similar or slightly reduced expression of the activating NKG2D receptor, and similar to reduced surface expression of CD57. The abundance of the cytotoxic molecules perforin, granzyme B, Fas ligand, and TRAIL is as great or greater in neonatal NK cells as in adult NK cells. Thus these findings suggest that neonatal NK cells differ phenotypically from but are not simply immature versions of adult NK cells.

Congenital viral or Toxoplasma infection during the second trimester can increase the number of circulating NK cells, which have phenotypic features of activated cells.

Fetal and Neonatal Natural Killer Cell–Mediated Cytotoxicity and Cytokine Production

Recent studies of NK cells of the fetal lung and other tissues have found that KIR-expressing cells are hyporesponsive in terms of cytotoxicity compared with those of adult peripheral blood or lung tissue. Fetal tissue NK-cell function also appears to be substantially more sensitive to the suppressive effects of TGF-β than adult NK cells. The cytotoxic function of circulating NK cells increases progressively during fetal life to reach values approximately 50% (a range of 15%-60% in various studies) of those in adult cells at term, as determined in assays using tumor cell targets and either unpurified or NK cell–enriched preparations. Reduced cytotoxic activity by neonatal NK cells has been observed in studies using cord blood from vaginal or cesarean section deliveries or peripheral blood obtained 2 to 4 days after birth ; full function is not achieved until at least 9 to 12 months of age. Decreased cytotoxic activity by neonatal NK cells compared with adult cells also is consistently observed with HSV- and CMV-infected target cells. By contrast, both neonatal and adult NK cells had equivalent cytotoxic activity against HIV-1–infected cells. These results suggest that ligands on the target cell or the target cell’s intrinsic sensitivity to induction of apoptosis may influence fetal and neonatal NK-cell function. The mechanisms of these pathogen-related differences remain unclear but may contribute to the severity of neonatal HSV infection. Paralleling the reduction in natural cytotoxic activity of neonatal cells, ADCC of neonatal mononuclear cells is approximately 50% of that of adult mononuclear cells, including against HSV-infected targets.

The reduced cytotoxic activity of neonatal NK cells appears not to reflect decreased expression of cytotoxic molecules but, instead, may result from diminished adhesion to target cells, perhaps as a result of decreased expression of ICAM-1 or diminished recycling of cells to kill multiple targets. However, the mechanisms responsible for diminished neonatal NK-cell cytotoxicity have not been conclusively defined. Cytokines, including IL-2, IL-12, IL-15, IFN-α, IFN-β, and IFN-γ, can augment the cytotoxic activity of neonatal NK cells, as they do for adult NK cells, and with fetal tissue NK cells, the augmentation of cytotoxicity may exceed that for adult NK cells. Consistent with the ability of IL-2 and IFN-γ to augment their cytolytic activity, neonatal NK cells express on their surface receptors for IL-2/IL-15 and IFN-γ in numbers that are equal to or greater than those of adult NK cells. Treatment of neonatal NK cells with ionomycin and phorbol myristate acetate (PMA) also enhances natural cytotoxicity to levels present in adult NK cells. This increase is blocked by inhibitors of granule exocytosis, indicating that decreased release of granules containing perforin and granzyme may contribute to reduced neonatal NK cytotoxicity. Finally, decreased neonatal NK cytotoxicity is not determined at the level of the precursor cells of the NK-cell lineage: Donor-derived NK cells appear early after cord blood transplantation, with good cytotoxicity effected through the perforin/granzyme and Fas–Fas ligand cytotoxic pathways.

Neonatal NK cells produce IFN-γ as effectively as adult NK cells in response to exogenous IL-2, IL-12, IL-18, HSV, and polyclonal stimulation with ionomycin and PMA, but fewer neonatal NK cells express TNF-α than do adult NK cells after ionomycin and PMA stimulation. Neonatal NK cells produce chemokines that suppress the growth of HIV strains that use CCR5 as a co-receptor to infect CD4 T cells but not those strains that use CXCR4 as a receptor.