Human neutrophils ≠ murine neutrophils: does it matter?

Summary

Human and murine neutrophils differ with respect to representation in blood, receptors, nuclear morphology, signaling pathways, granule proteins, NADPH oxidase regulation, magnitude of oxidant and hypochlorous acid production, and their repertoire of secreted molecules. These differences often matter and can undermine extrapolations from murine studies to clinical care, as illustrated by several failed therapeutic interventions based on mouse models. Likewise, coevolution of host and pathogen undercuts fidelity of murine models of neutrophil-predominant human infections.

However, murine systems that accurately model the human condition can yield insights into human biology difficult to obtain otherwise. The challenge for investigators who employ murine systems is to distinguish models from pretenders and to know when the mouse provides biologically accurate insights. Testing with human neutrophils observations made in murine systems would provide a safeguard but is not always possible. At a minimum, studies that use exclusively murine neutrophils should have accurate titles supported by data and restrict conclusions to murine neutrophils and not encompass all neutrophils.

For now, the integration of evidence from studies of neutrophil biology performed using valid murine models coupled with testing in vitro of human neutrophils combines the best of both approaches to elucidate the mysteries of human neutrophil biology.

Perspective and purpose

Our understanding of many of the biological processes related to human health and disease derives in large part from two investigative approaches, namely clinical observations of patients afflicted with disease, either acquired or inherited, and animal experimental systems intended to model human disorders. Astute clinicians in search of alleviating the suffering of their patients drive basic and translational studies to identify underlying physiology, both normal and aberrant. Particularly powerful because of extensive background genetic information and well-characterized inbred strains, murine experimental systems have served to model a wide spectrum of human diseases, including infections, autoimmune disorders, inflammatory syndromes, and malignancies. Investigators often use murine experimental systems to assess the role of neutrophils in a specific disease, syndrome, or malignancy, either as their exclusive experimental tool or as an adjunct to their studies of human neutrophils in vitro. Enthusiasm for the use of murine systems as surrogates for human neutrophils is occasionally excessive and can distort perspective. For example, in a detailed review of the approaches available to study the contribution of neutrophils in murine model systems ¹, the authors enthusiastically note in their concluding comments that “the use of mouse models for the study of neutrophils in vivo represents a milestone in the understanding of the biology of these cells and has enabled the deep interrogation of a number of pathways leading to diverse pathologies” ¹. As evidence of the success of this approach, the authors cite the use of DNase I as therapy for lung disease in cystic fibrosis “after the discovery that NETs [neutrophil extracellular traps] play a pivotal role in the disease”. In fact, the Food and Drug Administration (FDA) approved the use of recombinant human DNase I for patients with cystic fibrosis in 1993, more than a decade before the publication of the first report of NETs ². Multiple studies of the role of the highly viscous, DNA-rich secretions created by the large number of spent and necrotic cells in the airways of patients with cystic fibrosis prompted the FDA’s decision.

There may be many effective ways to manipulate neutrophils in murine experimental systems, but the intention in utilizing such systems is to have a tractable model of neutrophil-dependent biology that both mirrors the appearance of the original (i.e. the human) and recapitulates the same underlying mechanism(s). That is, to be an accurate and informative model, the mouse neutrophil must reproduce the behavior, biology, and biochemistry of human neutrophils in the process under study. In that way, investigators can probe the model to identify key regulatory molecules, critical pathways, and events that fuel pathophysiology caused by human neutrophils. A murine system that phenotypically mirrors the human situation but not the underlying mechanism is not a bona fide model and may prompt misleading conclusions.

One might ask what adverse consequences could arise from the use of experimental systems anchored only on the assumption that the same mechanisms underly a given phenotype for both mouse and man. At least two come to mind. From the perspective of the pursuit of knowledge, yet undiscovered aspects of human biology not represented in mice might be missed when the observations lack data on mechanism. Furthermore, extrapolation of observations from murine studies might fuel clinical trials in humans that prove unsuccessful because of the rationale for a particular intervention is flawed ^3-7. Hence, the incentive to know the fidelity of murine (or any animal) models should be great.

This review is neutrophil-centric, addressing only those settings where neutrophils represent critical contributors to the biological process under study. Consequently, my purposes are two-fold: first, to summarize the recognized differences in structure, composition, and biochemistry between murine and human neutrophils. The second goal is to present situations where such differences are such that the murine experimental system fails to represent human biology faithfully and is unlikely to yield sound and clinically relevant insights.

The structure, composition, and biology of human neutrophils

Current understanding of the functional repertoire of human neutrophils, albeit incomplete, includes a wide variety of activities that contribute to host defense against threats, both infectious and inanimate, and, when excessive, can culminate in tissue damage ^8-11. Neutrophils are the predominant nucleated cell in human blood (50-70% of leukocytes) and serve as first responders to the invasion by microbes, the presence of foreign bodies, or the sequalae of tissue trauma.

In response to engagement of receptors on their surface, neutrophils adhere to activated endothelial cells, transmigrate into tissue, and crawl towards the origin of the stimulating agent. Neutrophils are phagocytic and respond to particulate stimuli, with or without opsonization with complement, antibody, or both, by ingesting the target or, when too large to eat, by adhering to its surface. Phagocytosis prompts a coordinated response whereby intracellular granules fuse with nascent phagosomes and cytosolic components of the NADPH oxidase assemble on the extraluminal face of the phagosomal membrane. With fusion, granules release their contents into the phagosome, delivering an array of enzymes, antimicrobial peptides, and other active biomolecules (Fig. 1). Activation of the NADPH oxidase promptly follows its assembly and transfers electrons from cytosolic NADPH to molecular oxygen in the phagosome, thereby generating first superoxide ion, which is rapidly converted to hydrogen peroxide that in turn interacts with the azurophilic granule protein myeloperoxidase (MPO) to produce the potently bactericidal agent, hypochlorous acid (HOCl) ^12-16. HOCl reacts with the luminal contents of the phagosome, including both the ingested target (e.g. a microbe) and the granule proteins, thereby generating a host of modified proteins that are themselves microbicidal (Fig. 2) ^17-19. The synergistic activities among the individual granule proteins and between granule proteins and oxidants create an environment highly toxic to most microbes ^20-22.

Figure 1. Microbicidal action in human neutrophil phagosomes.

Antimicrobial events in the phagosomes of human neutrophils reflect the synergistic actions of granule proteins and oxidants. Granule proteins synthesized during granulocyte development in the bone marrow are packaged into one of several granule compartments (e.g. azurophilic and specific granules) and released into phagosomes by fusion with the nascent phagosome. Granule proteins include enzymes and proteins with direct bactericidal activity. Coincident with phagocytosis, the multicomponent NADPH oxidase assembles on the nascent phagosome and shuttles electrons from cytoplasmic NADPH to molecular O₂ in the lumen of the phagosome, thereby generating superoxide anion (${\mathrm{O}}_2^{-}$). Dismutation of${\mathrm{O}}_2^{-}$ produces H₂O₂ that in turn reacts with MPO to form HOCl. Most of the oxygen consumed by stimulated human neutrophils can be recovered as HOCl. Interactions among the granule proteins and between granule proteins and oxidants creates an environment toxic, and often lethal, for a wide variety of microbes [Figure from ⁸].

Figure 2. MPO-related events in phagosomes of human neutrophils.

At the onset of phagocytosis, granules and secretory vesicles fuse with nascent phagosomes, delivery granule proteins into the phagosomal lumen and inserting functionally important membrane proteins, including the voltage-gated proton channel Hv1 and chloride channels ClC3, KCC3, and cystic fibrosis transmembrane conductance regular (CFTR), into the membrane of the phagosomes. Granule fusion delivers MPO, and CFTR supports transport of Cl⁻ from cytoplasm into the phagosome. H₂O₂ generated by the NADPH oxidase, reacts with MPO to form Compound I (Cpd I), which mediates the two-electron oxidation of Cl⁻ to Cl⁺ and the production of HOCl. HOCl reacts with a wide variety of biomolecules, including proteins, lipids, carbohydrates, histones, and nucleic acids ^162-165. With respect to proteins, HOCl reacts most readily with sulfur-containing residues, notably methionine and cysteine, and generates monochloramines (NH₄Cl), protein chloramines (PNHCl), aldehydes, and other products, some of which exert sustained antimicrobial action [reviewed in ¹⁶⁶]. It is important to recognize that HOCl does not react preferentially with microbial proteins rather than host targets; all susceptible substrates are attacked regardless of their origin. Given that host proteins predominate in the phagosomes, most of the chlorinated and oxidized proteins are of host origin. [Figure from ⁹⁶]

The contribution of neutrophils to human innate immunity extends beyond the ingestion and killing of microbes ^9-11. Activated neutrophils release a host of chemokines and cytokines that recruit additional cells to the site and condition the milieu of the local inflammatory response ¹⁰. Furthermore, under optimal conditions, activation prompts neutrophils to engage transcriptional pathways that trigger apoptosis and initiate events that culminate in termination of the inflammatory response ^23-25. Sometimes, as with interactions with particular bacteria, such as Staphylococcus aureus, or when stimulation is excessive, neutrophils undergo a necrotic cell death ^24,26 and thereby release cytoplasmic contents that provide danger signals to drive additional inflammation. In some settings, neutrophils, as well as other cells, release extracellular traps (aka ETs), whose DNA may entrap microbes, contribute to autoimmune disease, or promote further inflammation ^27,28.

In summary, human neutrophils play a prominent role in every stage of the inflammatory response, as first responders to many different types of threats, as sources of some of the secreted immunomodulators released at the site, and as active participants in the termination of inflammation and the tissue repair needed to restore homeostasis ²⁹.

Human vs Murine Neutrophils

Morphology and distribution

It may be news but not a surprise that differences in cellular organization, contents, and activities exist between species ^30,31, largely reflecting customization to the lifestyles and needs of the individual species during evolution. Overall, however, human and murine neutrophils share many structural features

As noted earlier, neutrophils are the predominant nucleated cell in circulating human blood, at a concentration of 3-11 X 10³/μL, and account for 57-75% of the leukocytes. The neutrophil predominance is typical for many, but not all, mammals ³². Mice, like ruminants, have a lymphocyte-dominant hemogram, with lymphocytes at 75 to 90% of circulating leukocytes ³³. The maturation of neutrophils in humans and mice differs significantly, as CXCR4 binding retains human cells in bone marrow until they mature ^9,34, whereas murine neutrophils appear to complete maturation outside of the bone marrow ³⁵. Whether these variables would differentially alter the kinetics of normal neutrophil turnover or influence the emergency granulopoiesis that operates during acute inflammation in humans ⁹ is unknown.

Mature human neutrophils possess lobed nuclei, typically with two to five lobes and 50% with three lobes. Certain drugs and some vitamin deficiencies (e.g. folic acid, vitamin B₁₂) can cause hypersegmentation of the nuclei of human neutrophils. The presence of multi-lobed nuclei is considered by some as an attribute that allows neutrophils to traverse tight spaces, as occurs during transmigration through vascular endothelium, because of the flexibility of the nucleus ^36,37. Apparently, high deformability allows human neutrophils to migrate rapidly and to accommodate the variety of vessel sizes encountered. Hypolobulated neutrophils exhibit reduced migratory behavior ³⁸. The multiple indentations in the nuclei of murine neutrophils can give an appearance of lobulation, although many cells possess nuclei with a ring form. The half-life of circulating neutrophils differs between humans and mice as well, with that for human neutrophils ~ 6.5 hours ³⁹ and for murine cells 11.4 hours ⁴⁰.

Receptors

In general, receptors on the cell surface engage ligands, soluble and particle-bound, to transmit signals intracellularly and initiate specific cellular responses. Consequently, the repertoire of receptors on a cell dictates the agonists that will be detected and the responsive pathways that will be activated. The interactions and crosstalk among signaling pathways sculpt the phenotype observed.

Although murine and human neutrophils share many receptors, significant differences have been identified in specific settings. In the case of the colony-stimulating factor receptor family (CSF1R/IL34/CSF1), studies of specific receptor knock-out mice demonstrate that CSF1R/CSF1 and CSF3R/CSF3 are required for normal development of monocytes and neutrophils, but their loss does not result in absolute neutropenia [reviewed in ⁴¹]. In fact, loss of CSF1 results in increased numbers of circulating murine neutrophils ^42-44. In contrast, mutations in the human gene encoding CSF3 and CSF3R result in severe congenital neutropenia, with affected children experiencing severe and life-threatening infections due to the absence of circulating neutrophils ^45-47.

The balance between pro- and anti-inflammatory activities in neutrophils is maintained, in part, by the interplay between different types of immunoreceptors, which are defined as “transmembrane structure(s) containing an extracellular immunoglobulin-like domains and intracellular signaling via conserved immunoreceptor tyrosine-based activations (ITAMs) or immunoreceptor tyrosine-based inhibitory motifs (ITIMS)”[reviewed in ⁴⁸. The interplay between these receptors provides neutrophils with the functional plasticity necessary to adapt to the wide variety of situations they encounter. Notable differences in the array of immunoreceptors in human and murine neutrophils have been identified and were comprehensively reviewed by van Rees et al. ⁴⁸ (Figure 3), although a few merit brief mention here.

Figure 3. Immunoreceptors on human and murine neutrophils.

Expression of members of the immunoreceptor family on human and murine neutrophils, including those with ITAM (green), ITIM (red), or other (gray) signaling pathways. Table originally published in ⁴⁸

Neutrophils in both species express receptors for Fc domains of aggregated IgG (FcγR) and participate in the ingestion of IgG-coated particles and microbes. Human neutrophils have three distinct Fcγ receptors, FcγRI (aka CD64), FcγRIIa (CD32a), and FcγRIIIb (CD16b), each with different affinities and behavior, and a receptor for IgA (FcαR). Both FcγRI and FcγRIIa have activating cytoplasmic motifs and signal via Syk, whereas FcγRIIIb is glycophosphatidylinositol-linked and lacks signaling motifs. Murine neutrophils lack FcRα, despite possessing IgA, and express an inhibitory FcγRIIb and activating FcγRIII and FcγRIV, the first signaling through SHIP and the latter two via Syk. Such differences in ligand recognition and signaling attributes (activating vs inhibitory) extend to other immunoreceptor types, including CD300 receptors, Siglec receptors, SIRP receptors and CEACAM receptors [see ⁴⁸ for detailed discussion].

The differential expression of CEACAM receptors has significant implications relevant to human disease. The CEACAM receptors have a critical role in host defense against the exclusively human pathogen, Neisseria gonorrhea (Ngc). Clinically, a robust neutrophilic inflammatory response characterizes infections with Ngc ⁴⁹. Human neutrophils express three CEACAMs (CEACAM 1, 3, and 6) that bind colony opacity-associated (opa) outer membrane proteins on Ngc and mediate direct binding and uptake of Ngc by neutrophils ^50-52. Murine neutrophils lack the opa-binding CEACAMs and do not bind Ngc. Genetic manipulation to express the necessary human proteins in murine neutrophils corrects this specific short-coming ⁵³.

In addition to the immunoreceptors, others differ between man and mouse, most notably the formyl peptide receptors (FPRs). N-formylated proteins from bacteria and mitochondria are the proteins first recognized as agonists for FPRs, but the list of ligands has grown significantly over time ⁵⁴. FPRs are G protein-coupled transmembrane proteins whose engagement by suitable ligands stimulates adhesion, degranulation, NADPH oxidase activation, and chemotaxis in neutrophils The human genome includes three genes that encode FPRs (FPR1, FPR2, FPR3), and human neutrophils express FPR1 and FPR2 on their plasma membrane. There are eight members of the murine FPR gene family, with the orthologue for human FPR1, mFPR1, sharing many of the structural and functional features of the human receptor. One receptor, mFpr-rs1, is expressed on murine neutrophils and initially thought to share functional properties with human FPR2. However, studies in transfected cell lines demonstrate that the response profile of mFPR-rs1 does not parallel that of human FPR2 ⁵⁵. Other studies have identified differences in the FPR-dependent, agonist-induced activation of the NADPH oxidase in human and murine neutrophils [e.g. ⁵⁶]. The variety of specific ligands for the individual FPRs and, in the case of mouse, tissue expression in non-leukocytes likely reflect differential expansion of the FPR gene cluster between mouse and man during evolution ^55,57,58 and may undermine the fidelity of murine experimental systems that aim to mirror FPR-dependent human disease. Such differences have prompted an effort to identify FPR1- and FPR2-specific ligands that operate in both species [e.g. ⁵⁹] and thus provide informative experimental tools

Signaling

Downstream of receptors are the biochemical pathways that relay signals to the intracellular machinery that elicits cellular responses. The Rho family GTPase Rac2 is a component required for activation of the phagocyte NADPH oxidase ^60,61 and represents the predominant isoform in human neutrophils, with 80-95% Rac2 and the remainder Rac1 ^62,63. Murine neutrophils, on the other hand, express Rac1 and Rac2 in equal amounts. Mice deficient in Rac2 exhibit markedly depressed formyl peptide-stimulated F-actin polymerization, chemotaxis, and NADPH oxidase activity despite the presence of Rac1, which shares ~92% amino acid identity with Rac2 ⁶⁴. The Bokoch lab exploited these differences in Rac isoform expression between murine and human neutrophils to identify differential regulation of Rac1 and Rac2 over the gradient of fMLF concentration encountered as neutrophils crawl toward the source of chemoattractant ⁶⁵. Such studies illustrate how an appreciation of the differences between murine and human neutrophils can enable the dissection of complex human physiology.

Exposure to specific cytokines, chemokines, and other proinflammatory molecules converts neutrophils from their resting state to one in which they have a heightened responsiveness to stimuli, a process referred to as priming. The primed phenotype can include enhanced adhesion, chemotaxis, phagocytosis, degranulation, apoptosis, and NADPH oxidase activity, depending on the priming agent and the context [reviewed in ⁶⁶]. Typically, the priming agent does not itself activate neutrophils but instead prompts neutrophils to have an augmented sensitivity or response to subsequent exposure to a known agonist. Approximately 10% of circulating human and murine neutrophils exhibit a phenotype with increased expression of adhesion and activation markers and are “primed for recruitment” to a nascent inflammatory site ⁶⁷.

Priming of the NADPH oxidase in human neutrophils depends on selective phosphorylation of a subset of serine residues whose phosphorylation controls oxidase assembly and activation [reviewed in ⁶⁸]. Critical to priming by GM-CSF, TNFα, and LPS, for formyl peptide-activation of the NADPH oxidase is MAPK phosphorylation of SER345 in p47^{phox 69}. Following its phosphorylation, proline-isomerase Pin1 binds to PRO346 and promotes conformational changes in p47^phox that render other serine residues accessible for PKC-dependent phosphorylation ^70-72], which then exposes motifs in p47^phox that mediate assembly of a complete and functional oxidase. Thus, the phosphorylation of SER345 and binding of Pin1 serve as essential early steps in the priming activity of GM-CSF, TNFα, and LPS on formyl peptide activation of NADPH oxidase in human neutrophils ^71-73.

Exposure of murine neutrophils to LPS, TNFα, or cytochalasin B primes NADPH oxidase stimulation by FPR agonists ^56,74, although it is unknown if the underlying mechanism is the same as that for the human phagocyte NADPH oxidase. In general, the phosphorylation sites in murine p47^phox differ from those in the human protein (although rat p47^phox is identical to human), and SER345 is not conserved in murine p47^phox. Although murine p47^phox has SER346, the required adjacent residue is not proline, making it unlikely that its phosphorylation by MAPK would mediate priming in a Pin1-dependent fashion ^71,75. However, THR356 in murine p47^phox is phosphorylated by p38MAPK and has an adjacent proline residue, making it possible that the same mechanism, albeit with different residues, supports murine and human neutrophil priming [personal communication, Jamel El Benna, Ph.D., Research Director, Centre de Recherche sur l’Inflammation (CRI), Laboratoire d’Excellence Inflamex, Unversité de Paris, INSERM-U11499, Paris, France].

As part of the signaling pathway in TNFα-primed formyl peptide activation of the NADPH oxidase, there is sequential activation of different classes of phosphoinositide-3-kinases (PI3Ks), with the first phase dependent on PI3Kγ and the second on PI3Kδ ⁷⁶. TNFα primes murine neutrophils as well but less robustly than human neutrophils and with dependence on only PI3Kγ ⁷⁶. Although the authors offered no explanation for the differences in the observed PI3K isoform-dependent TNFα priming between human and murine neutrophils, they recommended caution when using murine experimental systems to predict the efficacy of PI3K inhibitors on responses of human neutrophils.

Secreted proteins

Granule proteins

Proteomic analyses of human ^77,78 and murine ^79,80 neutrophlls reveal many similarities, as expected, but some important differences [reviewed in ¹⁰]. Neutrophils possess intracellular compartments, granules, in which antimicrobial proteins, enzymes, and other bioactive molecules are stored and ready for release during degranulation into nascent phagosomes or the extracellular space [reviewed ⁸¹]. The azurophilic granules of human neutrophils possess three proteins (at least) that are absent from the granules of murine neutrophils: α-defensins, azurocidin (aka CAP37, aka heparin-binding protein, HBP), and bactericidal permeability increasing protein (BPI aka CAP57). Prominent contributors to host defense against infection and inflammation, these cationic antimicrobial peptides both damage or kill microbes as well as neutralize pathogen-associated factors that promote or exacerbate sepsis ⁸².

The α-defensins comprise nearly 50% of the protein in human azurophilic granules ⁸³, exert broad antimicrobial action against viruses, bacteria (Gram-positive and Gram-negative), and fungi ^84-86, and are chemotactic for T cell subtypes and immature dendritic cells ⁸⁷. Concentrations of α-defensins in plasma increase from ~ 42 ng/ml at rest to 900-170,000 ng/ml in sepsis, and α-defensins appear in pulmonary secretions during inflammatory lung disease ⁸⁸. The cationic antimicrobial peptide azurocidin meets infectious threats by directly damaging ingested microbes and by summoning additional host defenses, thereby serving both as a chemoattractant and as an activator of monocytes and macrophages ⁸⁹. Furthermore, azurocidin acts on endothelial cells and fibroblasts, thereby contributing to tissue damage in some settings ⁸⁹. BPI, the third of the azurophilic granule proteins present in human but absent from murine neutrophils, binds avidly to the surface of Gram-negative bacteria, thereby compromising microbial integrity ⁹⁰, and to endotoxin, thus neutralizing its proinflammatory action ^91,92 and its contribution to the pathophysiology of many infectious and inflammatory diseases ⁹³. Taken together, the absence of these three antimicrobial proteins from murine neutrophils represents a substantial difference in the relative abilities of human and murine neutrophils to damage microbes, recruit additional host cells to the site of infection, and neutralize proinflammatory microbial components such as LPS.

But in addition to the absence from murine neutrophils of agents of host defense, there are differences in the relative amounts of some antimicrobial granule proteins, as is the case for MPO. Depending on the strain of mouse, murine neutrophils express 20 to 50% of the MPO present in human neutrophils ^85,94, and the NADPH oxidase activity of murine neutrophils is 2- to 4-fold less than that of human neutrophils in response to conventional stimuli such as opsonized zymosan or phorbol myristate acetate (personal communication, Dr. Mary Dinauer, Department of Pediatrics and Department of Pathology and Immunology, Washington University School of Medicine, St. Louis, MO, USA). Thus, the capacity of murine neutrophils to generate HOCl and its reaction products (vide infra) is much less than that of human neutrophils. Because the MPO-H₂O₂-chloride system provides a potent and perhaps the most efficient oxygen-dependent antimicrobial activity in human neutrophil phagosomes ^{13,18,19,21,95,96}, its absence would compromise intraphagosomal killing.

A comparison of how human and murine neutrophils interact with Candida albicans illustrates the biological consequences of generating too little HOCl. Human neutrophils kill C.albicans, a common human commensal and one of the predominant fungal pathogens for humans ⁹⁷, and prevent fungal filamentation in an MPO-dependent fashion ⁹⁸. Neutrophils from individuals with an inherited deficiency of MPO fail to kill C.albicans in vitro or to damage hyphal elements in C.albicans and Aspergillus fumigatus ^99-104. The defective candidacidal activity in vitro due to insufficient HOCl production has clinical manifestations, as four of the first six MPO-deficient humans had severe infection with disseminated or visceral candidiasis ^101-103,105. In many ways, the interaction of murine neutrophils with C.albicans resembles that of MPO-deficient human neutrophils. Murine neutrophils kill C.albicans less effectively than do human neutrophils and fail to prevent Candida germ tube formation ⁹⁸. In fact, hyphal growth of ingested C.albicans results in the death of the murine neutrophil. Ineffective fungal killing occurs in murine blood as well as in isolated neutrophils, with significantly greater fungal survival and filamentation in murine blood than in human blood ¹⁰⁶. Although the defective candidacidal action of murine neutrophils likely reflects both the absence of α-defensins and the lesser amounts of MPO, the depressed activity of whole blood suggests that that soluble factors that act against C.albicans in mice are relatively ineffective as well. Together these inadequacies in cellular and soluble factor defenses constitute a serious short-coming in the use of murine experimental systems to model human immune response to C.albicans.

Cytokines

Secreted biomolecules that are immunoregulatory and serve to tune the immune response, both at the site of acute inflammation and systemically, cytokines were once not considered part of the repertoire of human neutrophil activity. However, since the early 1990s when investigators demonstrated that human neutrophils produced IL-8 and TNFα in response to phagocytosis ^107,108, subsequent studies have identified additional cytokines produced by human neutrophils in response to a variety of agonists and in diverse experimental settings [reviewed in^{10,109,110 111}. Neutrophil-derived cytokines include both those stored intracellularly [e.g. TNFα-related apoptosis-inducing ligand (TRAIL) ^112,113] as well as those synthesized de novo and secreted in response to in vitro stimulation. The list includes both pro- and anti-inflammatory cytokines, chemokines, colony stimulating factors, TNFα family members, and factors that promote angiogenesis and fibrogenesis ¹¹¹. Although neutrophils express only ~10 to 20% the RNA of other leukocytes ¹¹⁴ and therefore produce and secrete relatively less cytokine per cell, the enormous number of neutrophils at an inflammatory site enables neutrophil-derived agents to influence significantly local events. Secreting such bioactive agents, stimulated neutrophils communicate with other cells, both near and far, to modulate the inflammatory response ⁹.

Human and murine neutrophils differ in the cytokines expressed and secreted, summarized in ¹¹¹. (Figure 4). Human neutrophils do not produce IFNβ, IL-10, or IL-17 ^{115 116}, for example. Considering the far-reaching and multiple immunomodulatory properties of cytokines, one anticipates that the inflammatory milieu in many contexts would differ in mice and humans and, more importantly, that it would be difficult to predict the ways in which events in the two species would differ. It is possible that the cytokines and granule proteins act in concert to recruit and activate immune cells in an inflammatory process ¹⁰, thus adding to both the complexity of the biology and to the uncertainty in predicting species’ differences and their consequences.

Figure 4. Chemokines and cytokines.

Chemokines and cytokines potentially expressed or produced by human (panel A) and murine (panel B) neutrophils. The evidence for expression came from assessing gene expression, immunochemical staining, protein quantitation by enzyme-linked immunoabsorbant assays as well as some biological assays. The * indicates that the evidence was limited to mRNA detection and ? indicates data that are controversial at the time of this publication (2014) Table originally published in ¹¹¹

Experimental studies using mice

The significant differences in the genetic composition of mice and humans comes as no surprise, given the divergence of the two species during evolution 65 to 75 million years ago. Although the mouse genome contains more genes than does that of humans (20,210 vs 19,042 ⁶), a large majority of genes (75% of murine, 80% of human) are closely related functionally [reviewed in ⁶]. However, mice and humans differ markedly in their genomic responses to proinflammatory stimuli such as trauma, burns, and endotoxemia ¹¹⁷. In part, these differences may reflect the contrasting strategies that mice and humans have developed to optimize their odds of survival in an environment replete with infectious threats.

In general, multicellular organisms can adopt one of two distinct strategies to encounters with infectious threats: resistance or tolerance ^6,118-120. Resistance mechanisms include those targeted to kill (or at least damage) and eliminate invading microbes. Alternatively, tolerance allows the host to maintain function despite the presence of large numbers of potential pathogens. Although neither mice nor man utilizes one strategy at the expense of the other, resistance mechanisms predominate in human response to infection, particularly with respect to the innate immune system, whereas mice rely more on tolerance to endure infectious threats. Detailed discussions of these alternative approaches to infection have been presented more extensively elsewhere ^119,120, but the contrasting sensitivities of murine and human innate immune responses to lipopolysaccharide (LPS) illustrate the differences between resistance and tolerance. Simply put, mice tolerate endotoxemia and are highly resistant to LPS, with a lethal dose ~ 10 mg/kg. In contrast, humans mount a robust immune response to LPS and are exquisitely sensitive, with ~15 μg/kg triggering shock ^6,7. These diametrically opposite responses to endotoxin reflect differences in the determinants of sentinel and reactive immune cells and in the soluble factors that contribute critically to detection of LPS ¹²¹. Such genetic and phenotypic differences in the innate immune responses to a prototypical microbial proinflammatory agonist such as LPS may explain, at least in part, the legacy of failed therapies to target endotoxemia and sepsis: the murine experimental systems used to mimic the human response to endotoxin were imposters, not models. Consequently, the extrapolations from mouse to man fell short of expectations ^5-7.

The differences in genomic responses of mice and humans include proinflammatory stimuli in addition to those to endotoxin ¹¹⁷. Many investigators working with murine experimental systems to explore physiologic or pathophysiologic mechanisms that relate to human neutrophil biology or innate immune responses have identified some of the limitations of such systems that approximate phenomenology but do not mirror the human condition with respect to underlying mechanism. Several examples illustrate this important point.

• Neutrophils are critical contributors to the pathophysiology of many autoimmune diseases and targets for therapy with intravenous immunoglobulin G (IVIG), which is frequently used clinically in many inflammatory and autoimmune diseases ¹²². IVIG contains antibodies to the death receptor Fas (aka CD95) and to Siglec-9, thereby prompting cytokine-dependent neutrophil death when infused into humans and mediating a beneficial clinical response ^123-126. In contrast to the effects on human neutrophils, IVIG, F(ab’)₂ or Fc fragments of IgG, pooled murine IgG, or agonistic monoclonal antibodies to Fas or Siglec-9 have no effect on murine neutrophils. These contrasting results prompted the authors to underscore “the need to establish experimental systems that take into account the divergent species-related effects of IVIG on neutrophils” ¹²⁷.

• The α-defensins released from activated human neutrophils promotes damage in acute lung injury by disrupting capillary-epithelial cell barriers, and the absence of α-defensins from murine neutrophils disqualifies unmodified murine systems as potential models of acute human lung injury. The creation of transgenic mice expressing α-defensins in their neutrophils was required to provide a better approximation of acute lung injury in humans for study ⁸⁸.

• Although both human and murine neutrophils generate extracellular traps in response to larval Strongyloides stercoralis, the extracellular traps from human, but not murine, neutrophils are required for killing of larvae ¹²⁸.

• Whereas binding of Shiga toxin to murine neutrophils depends on the presence of globotriaosylceramide 3 (Gb₃) on the plasma membrane, its association with human neutrophils does not. Consequently, human serum amyloid P protects mice but not humans from the deleterious effects of Shiga toxin 2, an important virulence factor in hemolytic uremic syndrome caused by enteropathogenic E.coli O157:H7 ¹²⁹.

• As discussed earlier, the reduced capacity of murine neutrophils to generate HOCl undermines the accuracy, at the mechanistic level, of studies to elucidate the role of neutrophils in human infections with C.albicans. Similarly, murine neutrophils do not respond to Mycobacterium tuberculosis in the same MPO-dependent fashion as do human neutrophils ¹³⁰. Furthermore, MPO activity figures prominently in sterile settings of chronic inflammation, including atherosclerosis ¹³¹. Whereas MPO-dependent biochemistry has been implicated in the initiation and progression of human atherosclerosis, some murine systems used to study atherosclerosis lack evidence for a contribution from MPO ^132,133.

• In many settings, the behavior of tumors in mice fails to mirror critical aspects of human disease ¹³⁴. In part, the discrepancies likely reflect the absence of immunoediting, a fundamental determinant in the selection and development of human malignancies ¹³⁵. Many human tumors are extensively infiltrated with neutrophils¹³⁶, and the presence of tumor-associated neutrophils has been implicated in tumor angiogenesis ¹³⁷. Neutrophil infiltration of human tumors has been assigned prognostic significance, and some investigators in the field have advocated for neutrophil-targeted therapy as part of patient management ^135,137-139, which together underscore the clinical importance of having valid animal models.

Modeling human infections in which neutrophils play a central role

The examples mentioned above demonstrate that in some biological settings where neutrophils serve a defining role, the murine experimental system does not mirror the human condition with sufficient accuracy to provide informative insights into the mechanism(s) underlying human disease. However, many investigators have utilized murine models successfully to explore the pathogenesis of human clinical disorders, thus providing evidence that in the right setting, a murine model can serve as a powerful and informative experimental tool. Unfortunately, the identification of the right situation is a challenge often unrecognized by reviewers of grants or manuscripts who greet data derived from in vitro studies of human neutrophils with a request for the applicant or authors to test the hypothesis in question in a mouse model or knock-out. Unfortunately, such requests often prompt investigators to comply and thereby generate data that may be compelling but too often lack a connection with the human biology intended to be explicated.

To suggest using a mouse system to model human infections in which neutrophils figure prominently fails to consider the coevolution of humans and human-adapted microbes. Humans and their pathogens have coevolved to the extent that reciprocal adaptive changes have occurred in each and have thereby exerted selective pressures on each. This coevolution of host and pathogen serves as a potent agent in the biology of human infection ¹⁴⁰. Accordingly, the human immune system has specific strategies to target certain microbes, and, on the other hand, microbes that are successful pathogens employ adaptive responses to resist, evade, or counteract host responses. The coadaptation of microbes to humans and of human innate immunity to microbes has fostered the development of reciprocal constructs and responses that, in some cases, are so finely tuned that the introduction of a given microbe into a non-human species yields no valuable or valid insights, regardless of how similar the phenotype or entertaining the result of the murine infection. Some examples of the disconnect between mouse and man are obvious. Exclusively human pathogens such as Neisseria gonorrhoeae cannot be studied in a mouse model without significant genetic retooling of the mouse, as the organism is exquisitely and exclusively adapted to infect humans. For other microbes, it may be less readily apparent why a mouse model is in many ways inappropriate, as is the case for Staphylococcus aureus.

Staphylococcal infections in murine experimental systems

S.aureus is a human commensal that causes frequent and serious infections in humans, often complicated by local tissue necrosis, distant metastatic spread, and commonly death ¹⁴¹. Early host defense against S.aureus is neutrophil-mediated and oxidant-dependent ¹⁴², convincingly demonstrated by the morbidity and mortality of staphylococcal disease in individuals with chronic granulomatous disease, who lack a functional NADPH oxidase ¹⁴³ and are particularly susceptible to infection with S.aureus ¹⁴⁴. S.aureus has adapted successfully to man and become a prominent cause of infection in even the immunocompetent host.

Evidence from in vitro studies of normal human and of MPO-deficient neutrophils demonstrates that optimal killing of ingested S.aureus by human neutrophils is MPO-dependent ^95,145. However, neutrophils fail to kill all ingested staphylococci, and surviving staphylococci within neutrophils often trigger necrotic cell death of the host cell ¹⁴⁶, thereby releasing proinflammatory agonists that promote more inflammation and tissue destruction locally and metastatic infection to bone, heart, lungs and other vital organs. Adaptations during co-evolution with humans have brought S.aureus success as a human pathogen. In contrast to its relationship with humans, S.aureus is not an organism that naturally infects mice, although strains may be adapted to cause transmissible disease among mice ¹⁴⁷.

Strategies that S.aureus has employed to gain foothold in humans include those that block chemotaxis, ingestion, and killing by neutrophils [reviewed in ¹⁴⁸]. For example, the chemotaxis inhibitory protein of S.aureus (CHIPS) binds the C5a and formyl peptide receptors of human neutrophils better than murine receptors to inhibit human neutrophil chemotaxis ¹⁴⁹. Secreted staphylococcal proteins such as staphylococcal complement inhibitor (SCIN) and its homologues block steps in complement activation in the cascade that culminates in opsonization of the organism and phagocytosis by neutrophils ^150,151. S.aureus secretes a variety of leukotoxins, including Panton Valentine Leukocidin (PVL), which target specifically human cells to promote their lytic cell death ^152-155. Some have little or /no binding to murine leukocytes [reviewed ¹⁵³] (Figure 5). Even within the phagosome, S.aureus targets specifically human neutrophil attack by secreting SPIN (staphylococcal peroxidase inhibitor), a protein that binds to and inhibits the activity of human MPO [reviewed ¹⁵⁶]. S.aureus increase expression of SPIN within the phagosomes of human neutrophils, and SPIN inhibits MPO activity and the generation of HOCl ^157,158. SPIN does not bind or inhibit the MPO from rabbit, mouse, horse, or cow neutrophils, providing evidence of the remarkable human adaptation achieved by S.aureus ^156,157.

Figure 5. Staphylococcal bicomponent leukotoxins.

Human-adapted S.aureus produce distinct leukocidins that target myeloid and erythroid receptors with species-specific differences in susceptibility as demonstrated by experimental data (*). § designates potential pairs of the S component with an F component from another leukocidin; II indicates that DARC makes erythrocytes susceptible to the actions of LukED and HlgAB; ¶ HlgAB engages CCR2 and DARC in both human and murine cells, but only human CXCR1 and CXCR2. C5aR1, C5a anaphylatoxin chemotactic receptor 1; CCR2, CC-chemokine receptor 2; CXCR1, CXC chemokine receptor 1; DARC, Duffy antigen receptor for chemokines; HlgA, γ-haemolysin A; LukA, leukocidin A; NK cells, natural killer cells; PVL, Panton–Valentine leukocidin. Table originally published in ¹⁵³

*Shown are the leukocytes for which there are experimental data in the literature. ‡On the basis of published susceptibility of tested primary cells. §Potential non-cognate pairing of the S component with an F component of another leukocidin. which results in functional mixed pores or inactive hybrid complexes, depending on the pair. ∥DARC renders erythrocytes susceptible to the haemolytic activity of LukED and HlgAB. ¶HlgAB targets both human and murine CCR2 and DARC, but only human CXCRl and CXCR2. CSaRl. C5a anaphylatoxin chemotactic receptor 1; CCR2, CC-chemokine receptor 2; CXCRl. CXC chemokine receptor 1; DARC. Duffy antigen receptor for chemokines: HlgA, y-haemolysin A: LukA. leukocidin A; NK cells, natural killer cells: PVL. Panton-Valentine leukocidin.

Thus, during its coevolution with man, human-adapted strains of S.aureus have acquired attributes and behaviors finely tuned to elements of the human innate immune system in ways that cannot be precisely mimicked in unmodified murine experimental systems.

Summary

I chose this topic with great trepidation, since I have previously written about the functionally relevant differences between human and murine neutrophils and their potential consequences with respect to the use of murine experimental systems to model human neutrophil-dependent biology ¹³³. Furthermore, others have contrasted human and murine immune systems in great detail ^{6,31,106,159,160}, including a recently published thought-provoking perspective ¹⁶¹. However, I decided that the exercise has merit, but neither as an indictment of murine models, since that would be unjustified and not my intention, nor as a tribute to in vitro studies of human neutrophils.

Of course, the in vitro study of isolated human neutrophils, the experimental approach that has anchored my scientific career, has many shortcomings. The process of isolating neutrophils from venous blood itself imposes stresses on cells that are non-physiologic (exposure to plastics, dilute buffers, anticoagulants, centrifugal forces, density gradient media etc) and likely perturb neutrophils. Are isolated neutrophils ever “resting”? Some investigators keep isolated neutrophils at 4°C (or colder in an ice bath) before starting experiments in order to avoid inadvertent stimulation. What changes in membrane behavior or neutrophil metabolism accompany such cold exposure? Most consequential to the behavior of isolated neutrophils however is the absence from such studies of soluble factors and other cells, circulating as well as tissue-bound, with which neutrophils communicate. In the absence of such trophic factors, neutrophils isolated from venous blood only approximate neutrophils in vivo. Animal models avoid all these shortcomings (and likely more).

How best to proceed? Without question, human and murine neutrophils differ in many ways: differences in relative abundance in circulation blood, in receptors, signaling pathways, granule proteins, NADPH oxidase regulation, magnitude of oxidant and HOCl production, and repertoire of cytokines and other secreted molecules they produce. Consequently, authors should use and editors should demand accurate titles for manuscripts. If all the data in a report derive from the study of murine neutrophils or a murine model exclusively, then the correct and scientifically accurate title of the manuscript should include mouse/murine to modify the word neutrophils. Sometimes, but not always, the differences matter greatly and can derail extrapolations from murine studies to clinical care, as illustrated by several failed clinical interventions based on presumed murine models, such as therapy for sepsis.

However, when bona fide models, murine experimental systems can provide insights into human biology difficult to obtain in any other way. Furthermore, when murine models may fall short of explicating mechanisms underlying certain human diseases, the novel insights into general immune mechanisms have been remarkable. The challenge for investigators who employ murine experimental systems, however, is to distinguish models from pretenders and to know when the mouse provides accurate insights into the human biology. Testing in humans those observations made in the murine systems would provide a safeguard, but such experiments are not always possible. The expense and the housing requirements can preclude the use of animal models that better mimic human neutrophils, such as rabbit or pig. For now, the integration of evidence from studies of neutrophil biology performed using sound murine models coupled with testing in vitro of human neutrophils will combine the best of both approaches to elucidate the mysteries of human neutrophil biology.

Abbreviations:

fMLF

formyl-methionyl-leucyl-phenylalanine

FPR

formyl peptide receptor

HOCl

hypochlorous acid

LPS

lipopolysaccharide

MPO

myeloperoxidase

NETs

neutrophil extracellular traps

TNFα

tumor necrosis factor α