The neutrophil collective

Summary

More than a century after their discovery, neutrophils continue to puzzle immunologists. Their remarkable migratory, cytotoxic, phagocytic, and degranulating capacities raised the traditional perception that they are dedicated microbe hunters. Yet neutrophils possess an equally exceptional ability to acquire new traits across different environments, and when considered as a lineage collective, they are long-lived, reprogrammable, and retain memory of past insults. Here, we focus on the concept of the collective to make sense of both traditional properties and those that challenge existing dogmas. We model the structure of the collective as the combination of two biologically distinct compartments and discuss the unique properties that emerge beyond the sum of the individual cells. We hope that our review will stimulate discussion and spark new ideas about how neutrophils contribute to and can be exploited to promote health.

Introduction

Few cell types are as poorly understood as neutrophils. This is despite their early description as “microphages” by Metchnikoff and other pioneering immunologists more than a century ago, and their conspicuous abundance in blood and at sites of infection or inflammation, reflecting their antimicrobial prominence and avid migratory capacity. So, what is it about these cells that has made them so challenging to define? We argue that at the core of these challenges lies their inability to proliferate and short lifespan, which have prevented effective genetic marking and manipulation in culture systems beyond a few hours or days. In sharp contrast, research on other immune lineages, such as T lymphocytes, which do not have these limitations, has thrived, allowing for precise characterization of their origin and the discovery of a wealth of phenotypic, transcriptional, and functional subsets. This has ultimately enabled the immunotherapy revolution that has transformed cancer treatment.

In this review, we describe how our understanding of neutrophils has evolved over the years and how recent technical advances are rapidly reshaping the field. We focus on the collective rather than on the individual cells and model the basic structure of this neutrophil collective by integrating a wealth of work over the past two decades. Our goal is to remark and discuss the numerous unknowns surrounding neutrophils and to define the key hurdles hindering the enormous potential of neutrophils for a new revolution in immunotherapy.

What is a neutrophil?

Neutrophils are traditionally defined as myeloid leukocytes produced in the bone marrow that are abundant in blood (~10% in mice, >50% in humans) and rapidly accumulate at sites of injury or infection, where they engulf and destroy pathogens (reviewed in ¹). This definition, however, is rapidly evolving from a focus on migratory and antimicrobial properties to one that emphasizes their phenotypic and transcriptional plasticity and presence in non-injured tissues, as well as their prominent contributions to sterile inflammation, such as myocardial infarction, autoimmune diseases ^2,3 and cancer ⁴.

Browsing PubMed for articles published citing the term “neutrophil” offers insights into how the research focus on these cells has shifted over time, and the remarkable spike in interest in these cells in the past decade. For example, while characterization of their maturation in the bone marrow, infiltration in infected or inflamed tissues, and granule synthesis dominated until the 1970s, migration and mobilization from the marrow became the main focus over the next three decades. This shift in interest was driven in part by the discovery of monogenic mutations in patients that profoundly impact neutrophil migration and function and continue to provide insights into their biology ⁵. It wasn’t until the 2000s that the dynamism of gene expression in neutrophils began to be appreciated ⁶, coinciding with new paradigms of their migration within vessels and tissues by intravital microscopy and the identification of a “neutrostat” phenomenon that showed that neutrophil abundance, like many other key homeostatic parameters, is a physiological variable controlled by feedback mechanisms involving their uptake by tissue-resident phagocytes ⁷. However, the most influential discovery of this period was the demonstration that activated neutrophils can expel decondensed DNA coupled with histones and granule-derived proteins, producing structures fittingly called neutrophil extracellular traps or NETs ⁸, that are endowed with potent antimicrobial and pro-thrombotic properties ^9,10. About a decade later, elegant in vivo imaging studies reported swarming migration patterns of neutrophils to focal wounds that were mediated by relay of the leukotriene LTB4 between the migrating cells ¹¹. These studies highlighted the exciting concept that neutrophils rarely work alone and utilize exquisitely orchestrated group behaviors akin to other biological collectives, such as insects ¹². Finally, the past decade has seen studies revealing that neutrophils inhabit almost every tissue, even in conditions of health, and that they acquire phenotypic and functional properties in response to each environment ^13–15. These recent studies have uncovered principles of transcriptional regulation and environmental adaptation of neutrophils that were previously considered exclusive of long-lived cells, such as tissue-resident macrophages, in turn sparking a broader interest in other fields, including cancer ⁴, tissue repair ¹⁶ or neurodegeneration ¹⁷.

These evolving insights on neutrophil biology raise the question of what defines these cells. The presence of neutrophilic granules, surface expression of defined receptors (such as Ly6G in mice, or CD15/CD16/CD66b in humans), avid migration to specific chemokines or chemoattractants (e.g., CXCL1 and fMLP), or the production of NETs provide, in combination, a practical definition of what neutrophils are. However, this functional and phenotypic definition lacks precision as these proteins and activities vary depending on maturational stage or location, as illustrated by the observation that not all neutrophils in a given preparation, for example, produce NETs or migrate towards CXCL1. Hence, traditional definitions fail to recognize the diversity of neutrophils in the same way that the use of T cell as a brand name does not provide the biological precision needed in most contexts. Attempts to address this limitation have led to coining names for neutrophils “subsets”, but these tend to be restricted to individual studies ¹⁸ or polarized states (N1 vs. N2) in specific contexts, such as cancer ¹⁹. Ultimately, the lack of a precise definition or consensus regarding nomenclature is rooted in the absence of unambiguous genetic tracing approaches to distinguish real subsets. This generates confusion in the field; however, recent efforts to correct this should improve communication and favor progress ²⁰.

The way immunologists define neutrophils has been shaped by the model systems used to study them. While classical descriptions derive from human blood, mouse models have been crucial to dissect the core mechanisms of development, migration, and in vivo distribution. However, several molecular and functional traits differ between these species. Beyond the notable difference in blood abundance, the repertoire of adhesion receptors, granule proteins, cytokine networks, and oxidative burst pathways varies between the two species, leading to different activation thresholds and effector mechanisms. ²¹. Other model systems have also provided complementary insights. For example, studies in zebrafish confirmed dynamic behaviors such as swarming and reverse migration also found in mice ^22,23, and in vitro systems such as immortalized HoxB8 progenitors and human iPSC-derived neutrophils have facilitated genetic and molecular analyses^24,25. While here we focus primarily on mouse and human systems, where molecular resolution and translational relevance are greatest, it is likely that similar organizational principles of the neutrophil lineage extend across vertebrates.

The neutrophil collective

It is intriguing that neutrophils are not usually referred to as individual cells. Their short lifespans and incapacity to divide make it intuitive to infer that the biological impact of the individual neutrophil must be small. Instead, recognizing their organization as a “collective” highlights the importance of the group, both in scenarios of health and disease. This is not to undermine the importance of the individual, but to emphasize the emergence of new properties and patterns that are distinct from the sum of the individual cells, as broadly recognized, for example, in the context of collective cell migration ²⁶. Paradoxically, the concept of the neutrophil collective has received little attention as a distinct biological entity. Here, we argue that focusing on this feature provides new insights into the evolutionary “logic” of neutrophils and facilitates the interpretation of recently defined properties of these cells in vivo, including phenotypic and functional diversity, dynamics, or long-term reprogramming (i.e., biological memory; Figure 1).

Figure 1. Structure of the neutrophil collective.

We model the collective as the sum of two biologically and functionally distinct populations (compartments) of neutrophils. The Granulopoietic and Mature compartments differ in their capacity for long-term maintenance and lifespan, physiological functions, location, and regulatory mechanisms. Neutrophils in the Mature compartment eventually die through immunosilent (apoptosis and phagocytic uptake) or inflammagenic mechanisms (necrosis, pyroptosis, or NETosis). The model also proposes irreversible transitions of neutrophils between compartments. Pink and yellow cells represent distinct neutrophil populations arising from separate granulopoietic pipelines regulated by different factors or across marrow environments. Figure generated with BioRender.

Initial attempts to model this collective (or at least part of it) have emerged by integrating the transcriptome of individual neutrophils across different conditions and tissues ^18,27. Here, we extend this notion by proposing that the neutrophil collective is composed of two interconnected but separate biological elements, the Granulopoietic and the Mature compartments, and enumerate three defining properties: 1) that the Granulopoietic compartment is regulated by systemic cues and is long-lasting; 2) that local cues preferentially regulate the Mature compartment, which has a brief temporal impact; and 3) that the cells of this collective are connected in a non-reversible manner.

Functional compartments of the neutrophil collective

The Granulopoietic compartment is comprised of immature precursors that typically reside in the bone marrow and spleen and produces mature neutrophils. Common myeloid progenitors, considered the source of all myeloid cells, give rise to monocytic and dendritic cell progenitors (MDPs) and granulocytic progenitors (GMPs), which can also produce monocytes but are primarily committed to producing neutrophils. The earliest progenitor committed to the neutrophil lineage is the myelocyte, which progressively differentiates into non-proliferative metamyelocytes, banded and mature neutrophils ^28,29. This classification, based on traditional morphological inspection in hematoxylin-eosin preparations, has now been molecularly and transcriptionally characterized in mice and humans. In mice, CD34^LO Ly6C+ CD115^NEG pro-neutrophils are equivalent to myelocytes, and differentiate into CXCR4+ pre-neutrophils, Ly6G+ CD101^NEG immature, and finally CXCR2+ CD101+ mature neutrophils ^30,31. In humans, CD66b^NEG CD64^DIM CD115^NEG CD117+ CD71+ are the earliest committed progenitors that differentiate into CD66b+ CD15+ CD101+ CD10+ CD16+ mature neutrophils, as demonstrated in vitro and by cell transfer into immunocompromised mice. Importantly, detailed analyses have proposed that the different maturation stages align along a transcriptional continuum ^30–33.

Beyond its molecular profiling and classification of precursor cells, the Granulopoietic compartment exhibits critical features (Figure 1), including proliferative capacity, a long lifespan through replenishment from primitive hematopoietic progenitors, and localization inside the bone marrow and spleen. Under conditions of stress, however, this proliferative compartment can be mobilized to populate the liver, lungs, or wounds, as reported in the context of cancer, hematopoietic stress (e.g., anemia), chronic inflammation, treatment with mobilizing agents, or bacterial infections in mice and humans ^34–38. Immature precursors also appear in the circulation under these conditions (referred to as a left shift), although there is no evidence that maturation and expansion of these immature forms occur in the circulation. It is intriguing that only a fraction of precursors “escape” the marrow, suggesting that this mobilizable population may be endowed with distinct roles in the context of organismal stress. Granulocyte Colony-Stimulating Factor (G-CSF; encoded by CSF3) is the master cytokine regulating the proliferation and expansion of progenitors, as well as the survival of mature neutrophils, and their mobilization to the bloodstream ³⁹. Although granulopoiesis relies on G-CSF both at baseline and under conditions of emergency, genetic deletion of this factor or its receptor only partially impairs granulopoiesis ⁴⁰ suggesting that the Granulopoietic compartment may follow more than one type of factor-driven pipeline. The possibility that different mechanisms of granulopoiesis exist, either through the differentiation of different progenitors and/or due to the presence of distinct factors in different medullary niches, remains unexplored. Interestingly, however, this notion is supported by the reported alterations of granulopoiesis in conditions of cancer or infections ^41–43, as discussed in more detail below.

The Mature compartment refers to the population of post-mitotic, morphologically and phenotypically mature neutrophils, including those with segmented or hypersegmented nuclei in mice and humans, and expressing the marker CD101 in mice or CD10 in humans ^28,31. The key distinction of the Mature compartment is its inability to proliferate, and lifetimes that range from hours (in peripheral tissues) to only a few days (in the bone marrow and under specific conditions associated with inflammation and cancer; reviewed in ²⁹). We propose that this is the defining feature of this compartment, as it implies that mature neutrophils must function in synchronized waves of effector cells, whose reprogramming potential is controlled by the extramedullary environment but cannot impact tissues long-term due to the short duration of each wave. Examples of neutrophils from the Mature compartment include those in the bloodstream of healthy individuals, or those that infiltrate an inflamed tissue and are rapidly cleared by phagocytes or undergo cell death within hours.

The partition of the collective into two compartments recognizes the biological and functional specialization of cells from each compartment, but also their interdependence. The Granulopoietic compartment is proliferative, organized in discrete anatomical sites, endowed with long lifespans (weeks to months), and is reprogrammed by the environment, allowing it to store immune memory. It is therefore multifunctional and long-acting. In contrast, neutrophils from the Mature compartment are non-proliferative and short-lived. Although these neutrophils may display context-dependent programs as they transmigrate and enter new tissues ^38,44–46, the extent to which they undergo subsequent reprogramming when moving to a new environment to shift between effector states remains unclear. The lack of experimental evidence for these transitions suggests that the program acquired by mature neutrophils in a tissue may be final. This is in contrast with other lineages in which these transitions have been reported ¹¹¹. These observations define the Mature compartment as an agile, yet short-lived unit optimized for rapid adaptation to individual environments.

It is possible that separate Granulopoietic compartments exist, each generating downstream clones of Mature neutrophils with distinct properties, potentially generating a collection of clone- and maturation-associated populations that contribute to the substantial transcriptional “noise” reported in circulating neutrophils ⁴⁷. In any instance, this partition of the collective highlights a key distinction between neutrophils and other immune lineages, such as lymphocytes or monocytes, in which the properties of the two compartments coexist within the same cell. As discussed in more detail below, this division has important implications in understanding the contribution of neutrophils to physiology, including inflammatory and malignant diseases in which the collective has memory of past insults, and in the design of therapies targeting either compartment.

Properties of the neutrophil compartments

Granulopoiesis is sensitive to systemic signals, a paradigm of which is G-CSF. Under normal conditions, multiple cell types produce tonic levels of G-CSF, and genetic deletion of the cytokine or its receptor causes neutropenia ^40,48. In contrast, increasing the levels of recombinant G-CSF induces the rapid expansion and mobilization of both mature and immature neutrophils into the circulation ⁴⁹. Physiological stress caused by certain infections, trauma, or chronic inflammation results in a sharp increase in G-CSF produced by endothelial cells ⁵⁰, leading to neutrophilia. Other factors, such as GM-CSF, SCF, TGF-β, or IL-6, can control, alone or in combination, the expansion of granulocytic precursors and the survival of mature cells ^51,52, but the regulation of these factors in the bone marrow or peripheral tissues remains less well defined. In addition to these, pleiotropic cytokines like IL-1α and β, IL-17 and IL-23 also regulate granulopoiesis, in part by regulating the expression of G-CSF ⁷. This regulatory axis is relevant because these cytokines are produced in response to stress across the organism ^7,15 and their producing cells therefore act as sensors that activate granulopoiesis locally or remotely ^53,54. In the context of murine mammary and lung cancers, IL-1β and soluble receptor for advanced glycation end products (sRAGE), respectively, relay signals for granulopoietic expansion and imprint an immunosuppressive and protumoral program ^55,56. In contrast, GM-CSF produced by murine glioblastoma cells favors the production of antigen-presenting neutrophils with anti-tumoral properties⁵⁷. Obesity also promotes alterations in myelopoiesis and neutrophil maturation via CXCL2, which in turn promotes tumor growth and metastasis ⁵⁸. In contrast, systemic IFNγ reprograms granulopoiesis to produce PDL1⁺ neutrophils with immunoregulatory functions ⁵⁹. While these regulatory axes have been mostly studied in murine systems, direct evidence for the systemic activation of granulopoiesis and epigenetic remodelling of the granulopoietic compartment has also been obtained from patients with all-cause sepsis and COVID-19 infection by examining the signature of hematopoietic precursors mobilized to the circulation ^43,60,61. Collectively, these and other lines of evidence suggest that the Granulopoietic compartment is uniquely sensitive to systemic perturbations through the production of factors that respond to organismal stress, and that this level of regulation controls the quantity and quality of neutrophil production on demand. Of particular interest is the demonstration that cancer, obesity, and other systemic perturbations can elicit long-term alterations and functional bias in neutrophils, at least in part by epigenetic reprogramming of myeloid progenitors ^62,63, raising the notion that the neutrophil collective can store inflammatory memory only when targeted at the level of the Granulopoietic compartment.

Despite the dominance of systemic regulation, local regulation of granulopoiesis can also occur. For example, the skull and vertebral marrows appear to produce neutrophils that are functionally distinct from those in the circulation or produced in other anatomical sites ⁶⁴. These granulopoietic sites supply mature neutrophils to specific areas of the central nervous system, both at steady state or during infection and acute ischemia ⁶⁵. Thus, undefined local factors associated with anatomically distinct bone marrows, spleen, or other specific hematopoietic sites, may influence the expansion, survival, and instruct distinct programs in the Granulopoietic compartment. These observations reveal possible sources of heterogeneity of the neutrophil collective and represent an emerging area of research with exciting implications.

In contrast to the predominant systemic regulation of the proliferative Granulopoietic compartment, mature neutrophils in the circulation or those that accumulate in healthy or injured tissues appear to be predominantly regulated by local signals. For example, direct contact with microbial products (PAMPs) or danger signals produced by injured tissues (DAMPs) elicits rapid programs in neutrophils present at those sites. This is illustrated by swarming-type behaviors in focal injuries of the skin or infected lymph nodes ¹¹, matrix-producing properties of neutrophils recruited to the naïve or injured skin elicited by TGF-β signaling ¹³, a distinct DcTrailR1⁺ neutrophil type endowed with angiogenic properties in pancreatic tumors ^38,66, or in the naïve lungs and intestine ⁴⁶. In tumors, such specialization can be driven by glycolytic reprogramming within hypoxic niches that instruct immunosuppressive and pro-angiogenic functions ^38,66, whereas lung neutrophils acquire similar angiogenic features in CXCL12-rich areas through CXCR4-dependent signaling ⁴⁶.

A fascinating example of local modulation of neutrophils is in the context of vascular inflammation. This is mediated not only by soluble factors but also by direct engagement of adhesion receptors present on the endothelial surface; interactions via the selectin receptor P-selectin glycoprotein 1 (PSGL-1) mediate their physical contact with the activated endothelium and initiate a signaling cascade required for firm adhesion and transmigration ⁶⁷. Similar signals via PSGL-1 are delivered by platelets that interact with neutrophils bound to the vascular wall ⁶⁸. Even mechanical signals delivered during transmigration across the vasculature can reprogram the neutrophil to express components of the NADPH oxidase needed to kill the microbes it may encounter in the tissue ⁶⁹. The variety of local cues that prime intravascular neutrophils is further illustrated by the diversity of morpho-kinetic responses that neutrophils display as they move on the inflamed vessel wall. These appear to be instructed by the differential expression or activation of receptors and signal transducers ⁷⁰, revealing that even in seemingly homogeneous areas, different cues or interactions with other cells can instruct a wide repertoire of responses in mature neutrophils.

Finally, we propose that a key property of the collective is the maturational “transit” of neutrophils through the two compartments in an irreversible (unidirectional) manner. This is supported by pseudotime and RNA-velocity analysis (which leverages the dynamics of mRNA splicing to infer the direction of transcriptional transitions in a population; ⁷¹) of single-cell transcriptomic datasets (¹⁸ and our unpublished observations). It is also supported by analyses of in vivo-tagged waves of blood neutrophils that undergo irreversible phenotypic transitions from a CD62L^HI CXCR4^LO to a CD62L^LO CXCR4^HI phenotype ⁷². Although formal demonstration that transitions along the collective are strictly unidirectional in all contexts is lacking, we propose that reversibility of the neutrophil state is hampered by the unique constraints of their chromatin ⁷³, short lifespan, and susceptibility to cell death (rather than transcriptional reprogramming) upon potent stimulation (Figure 1).

Emerging properties of the neutrophil collective

Advances in single-cell and multi-omics technologies have revealed an unprecedented molecular diversity within the neutrophil lineage. As transcriptomic and chromatin-level resolutions increased, new populations and states emerged, revealing the collective as a structured yet heterogeneous system. Below, we focus on three key features that emerge when considering the entire collective, rather than individual cells or pathophysiological scenarios.

Molecular heterogeneity

A poorly studied but salient aspect of neutrophils, revealed by multiple bulk and single-cell transcriptomic and epigenetic analyses, is the substantial degree of molecular heterogeneity across tissues and pathophysiological contexts. These analyses reveal that human circulating neutrophils display unusually broad transcriptional and DNA methylation variability, greater than in other immune lineages.⁷⁴. Intriguingly, this burst in transcriptional “noise” occurs during the transition from bone marrow to blood ⁴⁷, suggesting that it is not random deregulation but an actively regulated process. Mechanistically, it may relate to chromatin reorganization within the lobulated nucleus, where loss of the loop-extrusion factor NIPBL alters enhancer topology and inflammatory gene positioning ⁷³. This developmental coupling of lineage specification and transcriptional noise could favour population-level diversity that equips the collective with multiple programs to respond to a wide spectrum of environmental and microbial challenges. These observations, discussed elsewhere in more detail ⁴⁷, highlight the organization of the neutrophil lineage as a collective from which new functional properties and antimicrobial strategies emerge.

Dynamic anatomical distribution

A second feature of the collective is a broad and dynamic anatomical distribution. Unlike most immune lineages, mature neutrophils are widely dispersed across the organism even at steady state and redistribute rapidly during infection, injury, or natural circadian cycles. In contrast, the granulopoietic compartment remains largely confined to bone marrow and spleen at baseline. Traditionally, these distribution dynamics have been interpreted as evidence that mature neutrophils exist primarily to patrol and defend or repair injuries. We argue that this interpretation, rooted in the ancestral description of neutrophils as rapid effector cells, is inaccurate and only represents a fraction of the neutrophils’ evolutionary purpose. Supporting this contention, studies have shown that neutrophils patrol not only the circulation but also multiple organs during homeostasis, including sterile tissues and mucosal surfaces ^15,75. Neutrophils that infiltrate the microbiome-rich oral mucosa, for example, have been proposed to function as sentinels of exposed surfaces, even in the absence of infections, both in mice and humans. Disruption of their function or numbers at these sites can lead to severe diseases of mucosal tissues ⁷⁶. Complementing these findings, others have shown that the active uptake of neutrophils that patrol the healthy organism by tissue-resident phagocytes is necessary to rewire the normal inflammatory state of macrophages to one that preserves homeostasis ^7,77. One example of this paradigm is the local and remote control of hematopoietic progenitors in the bone marrow through signals produced during the homeostatic phagocytosis of neutrophils in the bone marrow and the large intestine^15,72. In other organs, such as the skin, recent studies have shown that even very low numbers of patrolling neutrophils are sufficient to remodel the extracellular matrix, enhance tissue stiffness, and build ad hoc defensive structures that prevent microbial invasion ¹³.

Plasticity and tissue-level specialization

A third emerging property of the neutrophil collective is plasticity and tissue-level specialization. These features are a corollary of the other two properties and result in a broad range of phenotypes and functions that have been discovered over the past decade (reviewed in ²⁸).

Early studies in the context of Staphylococcus aureus infection identified three phenotypically distinct types of neutrophils that produced different cytokines and were associated with differential resistance to MRSA infection ⁷⁸. Likewise, two functional and phenotypic states of neutrophils (N1 and N2) were reported in the context of murine lung cancer models based on their tumor-promoting and -suppressive activities driven by TGF-β; depletion of neutrophils in these models showed opposite effects depending on whether TGF-β signalling was inhibited or not ¹⁹.

Subsequent studies have extended this tailored adaptation of neutrophils to tissues by profiling phenotypes and transcriptional signatures associated to different organs ¹⁴. For example, lung neutrophils exhibited transcriptional profiles that predicted immune-suppressive and angiogenic properties, which were functionally validated and demonstrated to support the normal capillary density and recovery from radiation- or endotoxin-induced injury ^14,79,80. Notably, the signals that induced these properties appeared to combine structural and biochemical cues that are restricted to the lungs, including anatomical proximity to the lung vasculature, mechanical sensing caused by physical constraints inside lung capillaries, and high levels of prostaglandin E2 (PGE₂) ^14,66,79. Similar adaptations occur in the skin, where neutrophils contribute to extracellular matrix formation and barrier maintenance¹³, and in the spleen ⁸¹, where they support B-cell maturation.

In tumors, additional regulators, such as the hypoxia-dependent transcriptional factor Bhlhe40, drive disease-specific transcriptional rewiring ⁸⁶. However, because the transition between the two compartments in the collective is so dynamic, cancer-derived signals may influence the entire collective, both priming progenitors and accelerating adaptation of mature cells to the tumor microenvironment. Findings from lung, pancreatic, and breast cancers support this view, suggesting that pathological cues can reprogram granulopoiesis itself to generate preconditioned neutrophil phenotypes.^38,55,56. Moving forward, it will be essential to define how neutrophils adapt to and support the normal function of other organs and disease contexts, and to investigate whether similar programs exist in human tissues.

A major outstanding gap in knowledge is the mechanistic basis for this remarkable plasticity. Solving this gap is challenging due to the reluctance of neutrophils to ex vivo culture and genetic manipulation. By analogy with macrophages ⁸², we have proposed that similar principles may apply to neutrophils, whereby lineage-determining transcription factors modify the chromatin during maturation to allow access to signal- and tissue-specific transcription factors (reviewed in ²⁸). To circumvent the challenge of working with primary cells, studies have used the HoxB8 immortalized murine progenitors in the context of inflammation to uncover transcription factors that are active during the differentiation of immature forms (such as Runx1 and Klf6), or exclusive of the mature compartment (such as RelB, IRF5, and JunB; ⁸³). More recently, human iPSC-derived neutrophil models have begun to provide a complementary platform to interrogate human-specific mechanisms of neutrophil differentiation and function under defined genetic and environmental conditions. These systems, while still limited in their maturation fidelity, represent a major step toward understanding neutrophil plasticity in humans ^25,84,85.

Revisiting neutrophils through the lens of the collective

Neutrophils have been traditionally studied under the prism of three fundamental traits: massive production in the bone marrow, prompt deployment during infections followed by trafficking to organs via the bloodstream, and short lifespans. These attributes, however, have often been interpreted through the lens of a homogeneous group of cells, yet advances in single-cell, imaging and computational technologies have made it clear that these features must be interpreted from the perspective of a heterogeneous and dynamic collective of cells. Below, we discuss these traditional features of neutrophils from the perspective of the collective, presenting a lineage that is structured by its developmental architecture, exposure to tissue-specific cues, and the constrain of a short lifespan.

A moldable production pipeline

Given the short lifetime of the Mature compartment, granulopoiesis provides the structural backbone of the neutrophil lineage. In both humans and mice, differentiation proceeds through sequential stages—from GMPs to pro-neutrophils, pre-neutrophils, and ultimately post-mitotic mature cells. This maturation pipeline is governed by a hierarchical transcriptional network: early multipotency is maintained by PU.1 and IRF8, which must be downregulated to enable granulocytic commitment. GFI1 enforces lineage restriction by repressing monocytic programs, while C/EBPα initiates neutrophil differentiation and arrests proliferation ^32,87,88.^. Terminal maturation is driven by C/EBPε, which activates granule gene expression and promotes functional specification ⁸⁹. Under stress, inflammatory cytokines such as IL-6 or G-CSF activate C/EBPβ- and STAT3-driven gene expression to accelerate neutrophil production ⁹⁰

A remarkable feature of this granulopoietic pipeline is that, despite its tight regulation, it is highly malleable. Systemic cues, including interferons, IL-1β, or tumor-derived factors, can reprogram granulopoiesis to produce neutrophils repurposed for specific functions. In such contexts, neutrophils appear to leave the marrow with a biased transcriptional program, reflecting upstream fate decisions rather than peripheral adaptations. Evidence for the relevance of these marrow-imprinted programs has been obtained in the context of cancer, in which neutrophils exhibit impaired production of reactive oxygen species (ROS) and NETs, enhanced suppressive and pro-metastatic potential, and even anti-tumoral phenotypes ^55,63,91. Further reflecting this developmental plasticity, recent studies have identified progenitor populations in tumor-bearing hosts that blur classical myeloid boundaries. For example, monocytic precursors differentiate into cells with neutrophil-like morphology and potent immunosuppressive capacity ^92,93. These cells retain transcriptional features of their monocytic-like origin, such as IRF8 dependence and M-CSF responsiveness, yet express late-stage granulocytic genes encoding for S100A8/9, Arg1, and Lox1. These findings suggest that granulopoiesis may involve fate-biased sublineages tuned by specific pathological signals. When, how, and at what level this divergence occurs is unclear; it may involve distinct progenitors and/or the exposure to different factors across the multiple granulopoietic sites reported to co-exist in the bone marrow ⁹⁴.

A corollary of these findings is that, by imprinting lasting modifications in long-lived multilineage or myeloid progenitors, the granulopoietic pipeline should be able to produce a continuous supply of functionally biased neutrophils, thereby enabling memory of prior encounters at the level of the collective. Evidence for such long-term granulopoietic bias has been reported in many pathological contexts, including fungal and bacterial infections (through their derived PAMPs), during infections and in anemic patients, or in cancer, all of which elicit the production of endogenous alarmins, such as heme ^42,95–97, impacting the functionality of the produced neutrophils for many months after the initial exposure to a stimulus ^42,95. While the signaling pathways and affected progenitors are likely to vary between conditions, a common mechanism involves epigenetic marking of histones in the progenitors, leading to the production of neutrophils with distinct capacities to produce ROS, cytokines, and other effector functions that are generally protective against cancers or secondary infections, but can also exacerbate injury in the context of inflammatory diseases (reviewed in ⁹⁸). While this phenomenon may be similar to traditional immune training mechanisms reported for macrophages and other longer-lived cells ⁹⁹, for neutrophils it must necessarily target the Granulopoietic compartment.

We model this fundamental property of the neutrophil collective as a combination of distinct differentiation pathways, driven by transcriptional “branching” of the Granulopoietic compartment at multiple points under the control of by distinct environments and context-specific cues (Figure 2A–B). While at present we can only predict that the same branched granulopoietic structure might exist already in the steady state, it is possible that acute or chronic signals reinforce or suppress (i.e., reprogram) certain trajectories to favor the ad hoc production of preferred neutrophil types (Figure 2C). This model is compatible with the presence of dedicated progenitors that produce distinct neutrophil subsets committed to specific functions, but this possibility requires formal demonstration.

Figure 2. Plastic granulopoiesis and fate of the neutrophil collective.

(A) The neutrophil collective originates from a granulopoietic backbone proposed here to branch at multiple points, thereby enabling the production of a large repertoire of neutrophil types (N1-N8). (B) Neutrophils produced through the different granulopoietic trajectories (T) can be additionally modulated by signals present in different environments (E) to further increase the final diversity of mature neutrophils (N1-N8 neutrophils in E1-E8 environments). In the proposed model, the N1 state is specified by the E1 environment, N2 by E2, and so on. It is likely that upstream progenitors contribute to the production of distinct subsets of neutrophils through separate maturation pathways. (C) Acute or chronic stimuli (cues) associated with pathological contexts favour specific maturation pathways to favour the production of specific types of mature neutrophils.

Spatial and temporal dynamism

Neutrophil migration is regulated by developmental timing, local and systemic cues, and tissue context. Viewed from the perspective of the collective, their remarkable migratory properties serve as a deployment strategy to efficiently deliver mature neutrophils into tissues to sense perturbations, execute local functions, or deliver homeostatic signals. The first migratory step—mobilization from the bone marrow—is regulated by the antagonistic actions of CXCR4 and CXCR2. CXCR4 retains neutrophils via stroma-produced CXCL12, while CXCR2 promotes release in response to G-CSF and inflammatory signals, ensuring rapid delivery into circulation ^100,101. Intriguingly, the emigration of neutrophils from the bone marrow is intimately linked to the acquisition of functional competence, characterized by sharp transcriptional and metabolic changes immediately before leaving the bone marrow. This transition includes chromatin remodeling and regulation of genes controlling degranulation, oxidative burst, and apoptosis ^27,102. Simultaneously, mitochondrial respiration declines and cytoskeletal plasticity increases as neutrophils leave the bone marrow, priming them for vascular surveillance. This coordinated emigration and functional maturation may spare the marrow from the cytotoxic activity of neutrophils during systemic stress ¹⁰², and highlight the sharp transition and biological division of labor between the Granulopoietic and Mature compartments. Under conditions of stress, however, medullary neutrophils facilitate the repair of the medullary niche or activate the hematopoietic compartment, enabling emergency granulopoiesis in response to systemic insults ^103–105. These findings evidence the remarkable dynamism of neutrophils already in the bone marrow as active regulators of hematopoiesis and the hematopoietic niche.

A striking aspect of neutrophil dynamics is their alignment with the environment, as illustrated by their rhythmic release from the marrow and clearance from the blood in synchrony with the organism’s diurnal cycles. Intriguingly, while most immune lineages display antiphase oscillations between mice and humans, as expected given their respective nocturnal and diurnal activities, neutrophils show conserved patterns, peaking at daytime in both species ^106,107,109. This suggests that evolutionary drivers independent of organismal activity may have guided the diurnal dynamics of neutrophils. In mice, newly released neutrophils undergo temporally regulated changes in phenotype, morphology, and granule content that are controlled cell-intrinsically by a Bmal1-CXCR2-CXCR4 axis, dictating when and under what activation state they reach tissues, both under homeostatic and inflammatory conditions ^106,108. An intriguing finding from these studies is that homeostatic migration into tissues follows different mechanisms from the canonical extravasation cascade established in the context of inflammation ¹¹⁰, as it appears to be initiated by integrin activation rather than selectin-mediated interactions ¹⁰⁸. Overall, this temporal phenomenon suggests that synchronized waves of neutrophils are released into and cleared from the circulation daily, implying that functionally distinct populations of neutrophils populate the blood at different times of the day. It also raises the intriguing possibility that the neutrophil collective delivers circadian signals to the multiple tissues that they infiltrate, a possibility that has only been explored in the bone marrow and skin so far ^13,72 and may have important implications for understanding how neutrophils impact global organismal physiology.

Finally, these circadian cycles of tissue infiltration and elimination highlight a fundamentally different temporal scale between neutrophils and other immune cells. Tissue-resident macrophages, which can persist for weeks to years through local self-renewal, have enough time to integrate into their niches and acquire stable transcriptional and functional traits^111,112. In contrast, neutrophils that enter naïve tissues persist for only hours to days ⁴⁶, rely on continuous supply from the bone marrow, and are, for this reason, only transiently reprogrammed by local cues (Figure 3). This brief exposure to tissue-derived cues is sufficient, however, to induce transcriptional or metabolic imprinting ⁴⁶. Interestingly, in some instances, neutrophils can emigrate from tissues through “reverse” migration and relay inflammatory signals to other organs ^{44,45,113–115}, suggesting that they may function as messengers that deliver physiological or pathological signals between organs.

Figure 3. Temporal dynamics of the neutrophil collective in tissues.

Tissue-resident macrophages (top panels) originate during embryonic life, seed tissues, and persist for weeks to years through local self-renewal. Their longevity enables stable imprinting by local cues and the acquisition of specialized homeostatic functions. In contrast, neutrophils (bottom panels) are produced daily in the bone marrow and infiltrate tissues following brief but periodic circadian cycles. They undergo partial reprogramming before being cleared by efferocytosis or other mechanisms. The model proposes that the impact of neutrophils on tissues is also weaker (faint colors) and circadian (sinusoidal thin arrows), compared to the persistent and stronger signals (straight thick arrows) delivered by resident macrophages.

Effector activities in context

Mechanisms controlling neutrophil cytotoxicity

Mature neutrophils carry a powerful cytotoxic arsenal composed of granule-derived proteins—including antimicrobial peptides and proteolytic enzymes—together with the ability to generate ROS and release NETs. Such exuberant killing potential demands tight regulation in time and space, and therefore several mechanisms ensure that full activation occurs only when and where required. During granulopoiesis, the cytotoxic machinery is preassembled, but the delayed expression of active sensing receptors limits effector function within the bone marrow.¹⁰². Exposure to peripheral cues later enhances oxidative and cytotoxic readiness.¹¹⁶. Once in circulation, “proteome disarming” progressively attenuates the neutrophils’ cytotoxic potential through programmed release of granule content into the bloodstream, thereby reducing ROS production and NET formation¹¹⁷. Consequently, inflammatory injury is diminished when insults occur after the disarmed phase of the circadian cycle ¹⁰⁸. Complementing their protective disarming in blood, “aged” neutrophils can secrete extracellular vesicles that actively promote resolution by dampening complement activation, an effect observed both in mice and in human arthritis. ^{118 119}. Neutrophils also adjust their responses according to the physical characteristics of pathogens: small microbes are phagocytosed and degraded intracellularly, whereas large pathogens that cannot be engulfed (such as fungal hyphae or bacterial aggregates) trigger NETosis, extracellular ROS production, and IL-1β expression to recruit additional cells.^120,121. Together with the exquisite coordination of swarming neutrophils around focal wounds or infections within tissues ¹²², these examples reveal an immune collective that senses, calibrates, and resolves tissue stress through control of its effector output.

Context-specific effector programs

Comparable logic extends to non-infectious contexts in which neutrophils acquire angiogenic, immunosuppressive, antigen-presenting, or structural functions. For example, human and murine adenocarcinomas in the lung and pancreas, as well as various types of liver cancer, instruct pro-angiogenic and immunosuppressive effector functions on neutrophils ^{38,86,123–125}. These programs are driven by signaling pathways that appear to be context specific, such as hypoxia, ER stress and the transcription factor BHLHE40 in pancreatic cancers ⁸⁶, and STAT3 in head and neck cancer and melanoma ¹²⁶. In contrast, glioblastomas use interferons and GM-CSF to elicit antigen-presenting properties in neutrophils. This property counters tumor growth and has been proposed to be instructed in local medullary niches of the skull ¹²⁷.

Although multiple other transcriptional states have been identified associated with tumors, a large study that profiled neutrophils across 17 different types of human cancer revealed a dominance of VEGFA+ angiogenic and HLA-DR+ CD74+ antigen-presenting phenotypes and further identified distinct metabolic dependence and epigenetic rewiring of these antigen-presenting neutrophils ¹²⁵. It is likely that this remarkable variety of transcriptional and functional states reflects adaptations that also exist during development and normal tissue physiology. Indeed, angiogenic and immunosuppressive neutrophils have been reported in the context of tissue ischemia and in the naïve lungs ^46,128–130, in the mouse embryo ¹³¹, and in mouse and human neonates ¹³². Further studies in breast cancer suggest that mutations associated with malignant or premalignant states, including clonal hematopoiesis, can further instruct the distinct expansion and reprogramming of the Granulopoietic compartment ^133,134.

Structural and reparative functions

Beyond immunomodulation, neutrophils can contribute directly to tissue architecture. Matrix-producing and remodeling neutrophils have been described in barrier organs, including the skin, lung, and intestine ¹³, and during renal fibrosis ¹³⁵. In internal wounds, neutrophils are capable of transporting preformed matrix ¹³⁶. In the naïve skin, this structural program is driven locally by TGF-β signaling and restricted to the subepidermal region, where diurnal modulation of matrix stiffness occurs. Following injury, neutrophils accumulate around lesions to build a fibrotic structure that protects against microbial invasion¹³. The emerging notion that neutrophils are endowed with “structural” roles in anatomically and temporally defined contexts ¹³⁷ highlights the principles of dynamism and functional adaptability as hallmarks of the neutrophil collective and underlines the potential impact of these programs in multiple pathophysiology-relevant contexts, including cancer, fibrosis, and wound repair.

These studies highlight the ability of neutrophils to mimic specialized cell types, acting like dendritic cells for antigen presentation or fibroblasts for matrix production, and raise intriguing evolutionary questions. Their abundance, mobility, and capacity for rapid transcriptional reprogramming may provide an evolutionary advantage by enabling a fast, localized deployment of essential biological functions across the body. In this view, neutrophils serve as an agile, multipurpose population of cells that can cooperate in non-redundant ways with specialized cells.

Death is not the end

Perhaps the most defining process of neutrophils is their early death. For the collective, death is not simply the terminal fate of the individual cells but a lineage-level checkpoint that governs the rhythm of renewal, calibrates the inflammatory tone, and relays signals throughout the organism ^7,15,72,138. With billions of neutrophils cleared daily in the human body, the impact of their elimination on the organism can be massive, as illustrated by the relevance of their uptake by resident phagocytes to inhibit their default inflammatory bias ⁷. Therefore, neutrophil death is an integral component of the collective’s function, as it delivers signals (in the form of their own corpses) to the tissues. Ironically, while pathways leading to the death of neutrophils have been thoroughly studied in vitro (reviewed in ¹³⁹), the mechanisms of homeostatic neutrophil removal from tissues remain poorly understood. Intriguingly, however, these mechanisms are extremely effective as apoptotic neutrophils are virtually undetected in living tissues ¹⁴⁰.

Apoptosis, followed by efferocytosis, is widely regarded as the principal route of neutrophil clearance under both homeostatic and resolving conditions. In vivo studies with labeled neutrophils revealed routine removal in the liver, spleen, and bone marrow ^72,141. In parallel, barrier tissues such as the gut, skin, and airway mucosa appear to support alternative, non-phagocytic disposal routes, where neutrophils are expelled and eliminated via stool, mucus, or sloughed epithelia ^75,142. The immunological consequences of efferocytosis are well documented. In the gut, uptake of apoptotic neutrophils suppresses IL-23 production, reducing downstream IL-17 and G-CSF levels and thereby curbing granulopoiesis ^7,15. In the bone marrow, the uptake of aged neutrophils by tissue-resident macrophages modulates niche activity and the egress of hematopoietic progenitors by inhibiting the production of CXCL12 ⁷². Inflammation further elevates the impact of timely apoptosis to facilitate resolution, as indeed delayed neutrophil death or enforced survival promotes chronic inflammation ^143–145, whereas apoptotic neutrophils harboring pathogens enhance antigen presentation and prime T cell responses ¹³⁸, or can release epidermal growth factor to elicit the differentiation of monocytes into antigen-presenting cells during viral infections ¹⁴⁶, linking neutrophil death with adaptive immunity.

Neutrophils have been reported to undergo other forms of death, including NETosis, pyroptosis, necroptosis, and ferroptosis. In some contexts, these forms of cell death elicit the release of active signals in certain tissues, such as IL1β in the bone marrow to stimulate myelopoiesis ¹¹³ or NET-associated proteases that “awaken” dormant cancer cells and induce metastasis in the lungs ^147,148, However, these lytic modes of cell death are typically inflammatory and can promote vascular injury, thrombosis, and autoimmunity ^149–152. In cancer, ferroptotic neutrophils release oxidized lipids that suppress T cell immunity and facilitate metastasis ¹⁵³, whereas neutrophils that resist ferroptosis may persist in tumors and blunt responses to immune checkpoint therapy ¹⁵⁴.

Thus, rather than a passive phenomenon, neutrophil death is a programmable, context-responsive switch. It defines the output of the Mature compartment while shaping signals that feed back to the Granulopoietic compartment. Apoptosis resets the system and promotes resolution, whereas lytic deaths escalate defense but also promote pathogenic inflammation. More importantly, programmed cell death defines the dynamics of the neutrophil collective by enforcing continued production, large numbers, and transient reprogramming and functional states that can be reset at every circadian cycle or in response to systemic perturbations.

Beyond this classical view of terminal death in a tissue, there is now evidence that neutrophils can, in some instances, return to the circulation through reverse migration. First described in zebrafish and later reported in mice ^44,115, these findings suggested a cellular feedback loop whereby peripheral education can influence granulopoiesis and inflammation. Indeed, reverse migration can limit local neutrophil accumulation and accelerate inflammation resolution^115,155, but can also relay inflammatory signals that promote injury in distant organs ^113,114.

Open questions and a look to the future

We have focused this review on key principles of the neutrophil lineage, intentionally leaving out many important aspects of their biology to highlight the numerous unknowns that still populate the field. We start by asking what a neutrophil is and provide only a partial answer, recognizing the diversity of the immune lineage and the lack of a reference state. We conclude by proposing that neutrophil death is a fundamental component of the cell’s physiology and highlight the conspicuous absence of apoptotic neutrophils in living tissues, challenging existing models and inviting reconsideration of the importance of their active removal through mucosal surfaces. Both examples highlight dynamism as a critical feature of the neutrophil lineage, which we model as a collective of cells in which each element is rescindable but, when combined, acquires unexpected new features. For example, the presence of antagonistic populations of neutrophils in tumors, the tight circadian regulation despite lifetimes of less than one day, the programmed recruitment of functionally distinct waves of neutrophils in wounds, or the long-term memory of neutrophil responses, which greatly surpasses the lifespan of the individual cells.

This review does not specifically discuss disease, yet neutrophils show perhaps the strongest link to disease outcome among all immune lineages ^156–158. This is because we find more value in understanding how neutrophils preserve tissue fitness, not only by removing the occasional microbe, but also by orchestrating a multitude of tissue-specific processes, including stem cell trafficking, reinforcement of tissue mechanics, and stimulation of myelopoiesis in the marrow during stress. In contrast, the cytotoxic, angiogenic, or immunosuppressive properties that contribute to inflammatory injury or cancer, for example, appear to be manifestations of similar protective activities during homeostasis that we do not fully understand. The lack of successful translation of neutrophil research into the clinic may reflect this biased approach to studying their biology.

A significant area of contention in the field is the issue of nomenclature. Inspired by the well-delineated structure of the T cell lineage, their genetically tractable subsets, and their success in the clinic, the field has pushed for a similar partitioning of neutrophils into subsets that can be easily tagged. We reason, however, that the neutrophil lineage cannot use the same approach because its biological design is entirely different from that of lymphocytes and other lineages. The use of the term “collective” emphasizes the interconnection between all elements but also the lack of genetically separable subsets that can be isolated or manipulated (and if they exist, they are yet to be identified). Instead, appreciating the dynamism, plasticity, and long life of the neutrophil precursors may spark interest in targeting the Granulocytic compartment to favour desired outputs -for example, pro-angiogenic neutrophils in patients with diabetes, or those with antigen-presenting capacity in cancer patients.

Beyond these conceptual challenges, systemic variables such as organismal age and biological sex remain understudied in neutrophil biology. During aging, hematopoietic progenitors acquire myeloid bias and a basal inflammatory state that favors the production of neutrophils in mice and humans. Paradoxically, while neutrophils from aged individuals are functionally impaired, they are associated with organismal inflammation, likely due to their blunted migration and reduced control of macrophage homeostasis (reviewed in ¹⁵⁹). These studies, however, have largely focused on classical features of neutrophils, making it important to define how organismal aging influences other fundamental properties associated with angiogenesis, matrix remodelling, or antigen presentation. Sex also introduces a functional bias; neutrophils from males tend to be less mature and are functionally less active, whereas those from females have enhanced type I interferon signatures and NET formation^160,161. Understanding how these systemic axes intersect with the broader collective will be important for precise targeting of neutrophils across the lifespan and sexes.

Finally, we propose that defining the overall architecture of the neutrophil collective by combining multiple physiological and pathological conditions (for example, sex and age, health, inflammation, and cancer) and readouts (transcripts, chromatin or proteins) will enable a better understanding of its biological principles, delineate the extent of its plasticity and functional diversity, and identify actionable targets to tame this remarkable army of cells.