Editorial: Nyet to NETs? A pause for healthy skepticism

Abstract

Discussion of Parker et al. that challenges the evidence for the antimicrobial capacity of neutrophil extracellular traps (NETs) in vitro.

“Most institutions demand unqualified faith; but the institution of science makes skepticism a virtue.”

It is in the spirit of Merton's highlighting of the important role for skepticism in the scientific process that the article by Parker et al. [1] in the current issue of JLB provides incentive for reflection on the nature and relevance of NETs. A brief introduction to NETs will provide the context in which to appreciate the insights from Parker et al. [1].

In 2004, Zychlinsky and colleagues [2] described a previously unappreciated behavior of human neutrophils, namely, their capacity under very specific conditions to extrude DNA, histones, and some granule contents to form in vitro structures that were visualized beautifully using immunofluorescence microscopy and designated NETs. Subsequent studies extended the phenomenon to neutrophils from other species, including mice, fish, and chickens [3–5], and inspired an array of Joycean neologisms (NETosis, ETosis, NET-like structures, netting neutrophils). In pursuit of the potential molecular mechanisms for NETs formation, studies have demonstrated requirements for a functional NADPH oxidase and granule proteins MPO and human neutrophil elastase, and linked histone citrullination to their generation [6–9]. Most relevant to the article by Parker et al. [1], it has been clearly demonstrated that microbes can be trapped in preformed NETs by virtue of electrostatic interactions between the immobilized DNA and the microbial surface [10], prompting speculation that NETs generation might serve as “one of the first lines of defense against pathogens” [11]. Neither enthusiasm about NETs formation in vitro nor hyperbole about their importance in vivo has waned since their first description.

The recipe for triggering NETs formation in vitro is straightforward. In brief, expose isolated human neutrophils suspended in a low protein medium (2% HSA in RPMI) on uncoated glass coverslips to 100 nM PMA for up to 240 min at 37°C in 5% CO₂, then fix and prepare for immunolabeling or SEM [12]. Although certain biologically relevant stimuli can elicit the same phenomena, PMA remains the most potent agonist for NETs formation [12]. Studies about the in vitro consequences of NETs frequently highlight the capacity of preformed NETs to trap and kill microbes and speculate on their potential contribution to host defense. One recurring theme that has emerged as evidence supporting such a role for NETs is the observation that DNase or endonuclease expression by target bacteria nullifies NETs-dependent killing observed in vitro [13, 14]. When subsequently tested in animal models of infection, the same mutants that express DNase/endonuclease are more virulent. Without rigorously examining other possible explanations for the altered virulence in the presence of DNase, various authors consider this correlation as proof that the greater virulence of DNase-expressing mutants reflects their capacity to inhibit NETs formation. Although this extrapolation could prove to be valid when scrutinized more critically, there are at present limited, if any, data directly or causally linking NETs formation to host defense. In fact, unambiguous demonstration of NETs formation in vivo is lacking. The identification of NETs in vivo rests on colocalization of extracellular DNA, histones, and neutrophil granule proteins (e.g., MPO, elastase, cathelin-related antimicrobial peptide) or on visualization by SEM. However, the validity of the latter method has been undermined recently by a report that fibrin and NETs cannot be distinguished on the morphological criteria currently in use for NETs [15], thus leaving the in vitro demonstration of NETs-mediated antimicrobial action as the most compelling evidence linking NETs formation and host defense.

In this issue of JLB, the report of Parker et al. [1] delivers two important take-home messages. The first message challenges the notion that NETs formed in vitro kill entrapped microbes. As done previously by others, Parker et al. [1] presented PMA-preformed NETs with Staphylococcus aureus, an organism whose fate in humans is largely determined by interactions with neutrophils. Whereas other investigators determined the viability of organisms in NETs by directly enumerating CFU using live–dead staining or measuring CFU in the absence or presence of cytochalasin D (ostensibly, to distinguish extracellular from intraphagosomal killing), the New Zealand scientists released the staphylococci from NETs before assessing their viability. As the released staphylococci were viable, it was concluded that there was no direct killing of staphylococci by NETs alone. These findings, if reproduced by other equally skeptical investigators, suggest that although NETs entrapment may limit the movement of microbes and possibly arrest bacterial growth, NETs-associated organisms are not dead.

The second take-home message from Parker et al. [1] is more a reminder than a cautionary note. NETs formation in human neutrophils is MPO-dependent, and MPO is present in NETs. Accordingly, Parker et al. [1] recovered in NETs 80% of the MPO released from stimulated neutrophils, and the NETs-associated MPO retained its capacity for enzymatic activity. However, MPO alone lacks antimicrobial activity but rather serves as the essential catalyst in the MPO-H₂O₂-chloride system that generates HOCl (or bleach) and is responsible for the efficient antimicrobial activity characteristic of human neutrophils (reviewed in ref. [16]). Within phagosomes, concomitant stimulation of the NADPH-dependent phagocyte oxidase and degranulation provides H₂O₂ and MPO, respectively, which react to mediate the two-electron oxidation of Cl^– to Cl⁺ and production of the potent microbicidal agent HOCl. Thus, the presence of enzymatically competent MPO in NETs provides the potential for efficient microbicidal action, but realization of that potential requires a source of H₂O₂. Of course, H₂O₂ from the neutrophils from which the NETs were generated would be long gone, given its rapid dissipation relative to the kinetics of NETs formation, but neighboring phagocytes or NOX proteins expressed in nonphagocytic cells might provide the required oxidants.

In the context of understanding the biology of NETs, the observations of Parker et al. [1] inspire a “wish list” of NETs-related questions, both old and new, that merit focused consideration (Table 1). First, the elucidation of the molecular mechanisms underlying NETs formation should include the consideration that distinct pathways may operate in different cells and species. Second, the capability of NETs to mediate microbial killing in vitro needs to be re-examined with attention toward the efficient release of organisms from NETs prior to assessing microbial viability. It is critical to distinguish NETs-mediated killing from the failure to recover viable bacteria as a result of their entrapment. Third, efforts to demonstrate directly, rather than by correlations, that NETs are formed in vivo and participate in biological processes, whether physiologic, pathologic, or both, should continue. Lastly, if NETs form in vivo, then experiments to test, rather than prove, the hypothesis that NETs production contributes to normal host defense against infection should be pursued. Demonstration of the latter poses a significant challenge, however, given that “neutrophils that were not prestimulated (i.e., with 10 nM PMA for 60–240 min) killed S. aureus efficiently exclusively by phagocytosis” [6]. As the bulk of neutrophil-mediated killing of staphylococci thus occurs in the phagosome, the relative contribution of NETs to overall host defense may be minimal in settings where intact neutrophils would also exist. Consequently, one would anticipate that the absence of NETs formation during an experimental infection with an organism such as S. aureus would yield a subtle phenotype. Lastly, substantial experimental evidence demonstrates that nearly coincident with their activation, neutrophils initiate transcriptional programs that promote apoptosis and generate biochemical signals that terminate the inflammatory response. Defining the mechanisms by which stimuli that promote NETs formation, interrupt these events would provide novel insights into fundamental aspects of signal transduction in neutrophils.

Table 1. Top Four NET Queries in My Queue.

What are the mechanisms for NETs formation? —What are the relationships among NETs formation and autophagy, mitochondrial DNA release, and histone citrullation? —Are all NETs the same? What mechanisms operate in cells that constitutively lack MPO (e.g., chicken heterophils)? What is the fate of microbes trapped in NETs in vitro? —Are organisms trapped and thereby immobilized for attack by other elements of the immune system—soluble and cellular? —Are organisms killed in situ by NETs alone? If so, how? If killing is MPO-dependent, what are the sources of H2O2? Are NETs generated in vivo during infection? —Do NETs appear late in infection after neutrophils are exhausted and dead? —Are NETs present in abscesses or areas where large numbers of dead neutrophils are generated? What is the contribution of NETs-mediated antimicrobial action to overall host defense? —In the presence of intact phagocytes, what fraction of microbial killing or bacteriostasis can be attributed NETs? —How do the signals that promote formation of NETs derail those that initiate events that promote apoptosis?

These comments are not intended to undermine the concept of NETs or their biological relevance, as both merit careful experimental examination, but rather to encourage healthy skepticism pending more compelling data, akin to Alexander Pope's admonition [17] three centuries ago:

“Be not the first by whom the New are try'd,

Nor yet the last to lay the Old aside.”