Integrated control of leukocyte compartments as a feature of adaptive physiology

Abstract

As a highly diverse and mobile organ, the immune system is uniquely equipped to participate in tissue responses in a tunable manner depending on the number, type and nature of cells deployed to the respective organ. Most acute organismal stressors that threaten host survival—predation, infection, poisoning, and others—induce pronounced redistribution of immune cells across tissue compartments. In many cases, the biological purpose of this trafficking is not entirely clear. This review discusses the current understanding of leukocyte compartmentalization under homeostatic and noxious conditions. We argue that leukocyte shuttling between compartments is a function of local tissue demands, which are linked to the organ’s specific contribution to the adaptive physiology in both the steady state and upon insults. We highlight the narrow set of known neuroendocrine signals that relay and organize this trafficking behavior and outline mechanisms through which functional diversification of leukocyte responses can be achieved. Lastly, we place these concepts into a broader context, highlight future avenues for scientific discoveries, and discuss the implications for clinical medicine in the era of targeted immunomodulation.

Introduction

Leukocyte trafficking from one site (compartment) to another, referred to as re-compartmentalization, is a hallmark of organismal responses to environmental insults¹. This response is best studied in the context of infection and injury, where leukocyte trafficking is required for eliminating invading pathogens and restoration of tissue homeostasis². However, re-compartmentalization of leukocytes is not specific to these challenges, but rather a stereotypical feature of both homeostasis and general stress physiology, suggesting an involvement of these cells in a much broader set of adaptive biological processes.

Growing evidence implicates key contributions of different leukocyte subsets to mammalian physiology beyond their canonical roles in immunity³. Indeed, at steady state in vertebrates and mammals, lymphocytes are even found in the central nervous system, typically considered an “immune privileged” site, where they impact cognition and mood and support physiological programs as complex as sleep^4–7. Other “immune privileged” organs such as reproductive tissues also harbor leukocyte populations, where their functions beyond surveillance are less well understood^8,9 Given their necessity in maintaining homeostasis, it is clear that trafficking of leukocytes between tissue compartments must be tightly controlled because translocation of cells from one site to another comes at the cost of functional impairment at the supplying tissue site. On the other hand, excessive influx of leukocytes, as is frequently seen in acute disease states, into a compartment can also be detrimental¹⁰, and persistent changes in leukocyte compartmentalization that endure long after an inciting insult has resolved is a hallmark of a wide range of chronic diseases^11,12. The overarching logic as to why certain immune cells are abundant in certain tissues, beyond pathogen control and tumor surveillance, remains largely unclear. And yet, interference with leukocyte compartmentalization has emerged as a key therapeutic concept of modern medicine¹³.

In this manuscript, we discuss leukocyte compartmentalization during homeostasis and in response to environmental insults from the perspective of physiology. We summarize mechanisms of leukocyte trafficking between tissue compartments and propose a coherent logic underlying these cell movements in which leukocytes contribute to the adaptive physiology of the respective tissue. We introduce the concept of supply and demand of leukocyte trafficking and explain why mismatches between these two would be maladaptive and cause disease. We then highlight the importance of neuroendocrine control of leukocyte trafficking and provide examples for dysregulation of the corresponding functional components as well as their impact on leukocyte compartments. Finally, we consider these ideas and concepts in the context of clinical medicine and close with an outlook on future avenues of experimental and translational inquiry.

Leukocyte re-compartmentalization is a general feature of adaptive physiology.

Acute infection is the best-studied trigger of leukocyte re-compartmentalization¹⁴. Upon detection of pathogens, a cascade of downstream events aiming to restore homeostasis is triggered. Tissue-resident leukocytes present at homeostasis are typically the first cells to sense invading microbes and their activation contributes to the clinical correlates of inflammation: rubor (redness), dolor (pain) and calor (warmth, swelling). These changes typically occur at the expense of homeostatic tissue functions (function laesa)^15–19. If the pathogen burden exceeds the capacity of resident cells to combat the challenge, a demand for additional cells arises. In such scenarios, leukocytes are recruited from organ reservoirs and infiltrate the site of injury, culminating in an expansion of the corresponding leukocyte tissue compartment.

The recruitment of leukocytes to infected tissues is required for pathogen clearance but also serves additional functions including resolution of tissue damage, irrespective of the cause of damage²⁰. Indeed, leukocytes sense and respond to cellular damage itself, which can be a sequalae of pathogen virulence, but may also arise from other types of insults including sterile injuries. Stereotypical signals of cellular damage and death, called damage associated molecular patterns or “DAMPs”, activate and recruit leukocytes to injured tissues, where they are required for restoration of tissue homeostasis^21,22. Consistent with this, localized damage across various organ systems induces an influx of myeloid cells to the site of injury as demonstrated in experimental models of stroke, myocardial infarction, malignancy, ischemia-reperfusion injury or chemical toxicity^23–27.

The most well-studied program modulating leukocyte compartments in response to infection or injury is known as “emergency myelopoiesis”. The key output of this cascade is the mobilization of myeloid cells (neutrophils, monocytes and immature precursors) from organ reservoirs, followed by their distribution to the site of the insult. In this case, the supplying reservoir is the bone marrow²⁸. Because the bone marrow compartment will consequently experience a depletion of mature myeloid cells and the demand for these cells does not immediately cease, hematopoiesis is activated and skewed towards myelopoiesis at the expense of lymphopoiesis, thereby ensuring sufficient supply^29,30. Across the trajectory of the inflammatory response and as a function of the type of infection or insult, the demands for leukocytes change such that the quality and quantity of cells required differs between the acute phase of pathogen clearance and tissue repair³¹. Thus, coordinated shuttling of leukocyte populations is necessary to restore homeostasis following infection and tissue injury.

Yet, leukocyte trafficking is not specific to infection or tissue injury, but instead observed across a range of environmental insults including predation, starvation, poisoning or noxious cold, all of which are combated by biological programs vastly different from infection physiology, but share re-compartmentalization of leukocytes as a common feature.

It has long been observed that acute psychological stress provokes major shifts in leukocyte populations in both mice and humans^32–35. Various experimental models to induce stress in rodents exist, of which restraint – a model for predation - is the most commonly used³³. Mice exposed to restraint stress demonstrate rapid-onset neutrophilia with concomitant lymphopenia³⁶. Similar observations have been reported in other stress paradigms including repeated social defeat and predator odor^36,37. In contrast to the case of infection, where the biological purpose of neutrophil recruitment is not only intuitive, but also experimentally demonstrated³⁸, the reason for neutrophil mobilization upon predatory and psychological stress is not well understood. Conceptionally, “stress neutrophilia” could correspond to an anticipatory response, because pathogen invasion (via a wound) was frequently linked to predation throughout evolution. Preemptive neutrophil mobilization following threat perception would thus mitigate the risk of fatal infection. However, this idea is not supported by experimental studies because exposing mice to restraint stress impairs host defenses, increases susceptibility to infection and delays wound healing^36,39. Alternatively, neutrophils could contribute to, and may even be necessary for, stress physiology itself, but this hypothesis has yet to be rigorously tested.

Akin to psychological stress, fasting also elicits a stereotypical leukocyte response that is characterized by a contraction of the blood as well as most other tissue compartments, whereas cell numbers in the bone marrow remain unchanged or increase⁴⁰. This reduction in leukocyte numbers has been suggested to serve adaptive functions by conserving metabolic resources to enable survival of the host during starvation, while concurrently conferring susceptibility to infection⁴¹. On the other hand, recent work demonstrated a selective shuttling of T memory cells into the bone marrow during fasting, which supports their survival and provides enhanced protection against a subsequent infectious challenge⁴². The signal(s) governing these changes are incompletely understood but involve neuroendocrine commands as discussed below. It is also unclear to what extent leukocyte re-compartmentalization is a necessary aspect of fasting physiology itself.

Leukocyte compartments are also modulated in response to other insults including climatic challenges. Humans exposed to acute cold water-immersion exhibit increased neutrophils in the blood, whereas leukocyte re-compartmentalization to prolonged hypothermia phenotypically resembles fasting^43–46. In hibernating animals, lymphocytes are stored in secondary lymphoid organs during torpor and are rapidly mobilized upon arousal in a sphingosine-1-phosphate receptor-dependent fashion⁴⁷. Snake envenomation triggers neutrophilia and recruitment of neutrophils to the site of tissue injury⁴⁸. Similar observations have been made in the context of scorpion envenomation as well as poisoning with other substances such as acetaminophen or ethanol^26,49,50.

Thus, environmental stressors generally result in stereotypical patterns of leukocyte re-compartmentalization (Table 1). Whether or not this response is a prerequisite for induction of adaptive physiology to the respective threat, as it is for adaptation to injury, infection and potentially starvation, is currently unclear.

Table 1.

Schematic overview of leukocyte re-compartmentalization in response to environmental insults.

cell type	tissue compartment	insult				REFs
		systemic infection	fasting/starvation	acute stress	cold
Neutrophils	Blood	⇑	⇓	⇑	⇓	36,40,42,45,51–53
	BM	⇓ → ⇑	⇑	⇓	⇑⇔
	Lung	⇑	⇓	⇑	⇑
Monocytes	Blood	⇑	⇓	⇓	⇓
	BM	⇓ → ⇑	⇑	⇑	⇑⇔
	Lung	⇑	⇓	⇓	?
T cells	Blood	⇓	⇓	⇓	⇓
	BM	⇓	⇑	⇑	⇑⇔
	Lung	⇓	⇓	⇓	?
B cells	Blood	⇓	⇓	⇓	⇓
	BM	⇓	⇑	⇑	⇑⇔
	Lung	⇓	⇓	⇓	?

Blood, bone marrow (BM) and lung compartments are shown. Red corresponds to an increase in cell number, while blue reflects a decrease. The alternative red shading of the BM compartment in systemic infection denotes the transition from mobilization (decrease in cell number) to emergency myelopoiesis (increased proliferation of progenitors, yielding augmented myeloid cell numbers). Grey and white shading corresponds to undefined changes.

Leukocyte trafficking in homeostasis

Beyond the dramatic, host-wide remodeling of leukocyte compartments in response to severe perturbations, leukocytes also shuttle between tissues under steady state conditions. This behavior results in stereotypical, circadian oscillations in leukocyte numbers across organs⁵⁴.

During the active phase (dark cycle in nocturnal animals including rodents), leukocyte numbers in the blood are low whereas their abundance across tissues is high. Conversely, the behavioral rest phase (light cycle in rodents) is characterized by an expansion of the blood leukocyte compartment and a contraction of tissue compartments^55,56. In mice, blood leukocytes peak around Zeitgeber time 5 (referred to as ZT5; 5 hours after lights on) and exhibit a nadir at ZT13 (1 hour after lights off)^55,57. The molecular driver of this trafficking behavior is the circadian clock, which controls rhythmic gene expression of chemokines and adhesion molecules that guide leukocytes to their destinations⁵⁵. The so-called “master clock” resides in the suprachiasmatic nucleus of the hypothalamus, exhibits robust rhythmic gene expression, receives light sensation as a major signal input (Zeitgeber) and regulates whole-body physiology by coordinating circadian activity of endocrine and neuronal signals including the hypothalamus-pituitary-adrenal (HPA) axis^58–60. Peripheral clocks can partly function autonomously⁶¹. The trafficking of leukocytes is regulated by both cell autonomous and non-cell autonomous clock networks⁵⁵. Conversely, disruption of circadian rhythmicity leads to perturbed leukocyte compartmentalization, which confers susceptibility to disease⁶², suggesting that this circadian program is biologically meaningful for maintenance of homeostasis.

To date, circadian leukocyte trafficking between compartments has mostly been studied in the context of diseases, where the resultant pathology is a function of the time of the day during which the insult occurs^55,63–66. Vice versa, the physiological purpose of oscillating leukocyte numbers across tissue compartments in homeostasis, much like with stress states that are not related to injury or infection, is not well defined. Given the growing body of evidence suggesting that tissue-resident leukocytes confer indispensable functions in the steady state⁶⁷, one might predict that trafficking leukocytes are likewise relevant to the biology of tissues. In fact, tissue-resident leukocyte populations participate in a range of housekeeping functions such as thermogenesis, metabolic control or nutrient absorption^3,68,69. While some of these are relayed via local mechanisms, others involve long-range communication via nerval pathways and soluble signals. For example, brain microglia orchestrate cephalic insulin release via interleukin 1 beta (IL1B)-dependent activation of efferent vagal fibers, while macrophage-derived, soluble IL1B induces postprandial insulin secretion from pancreatic beta cells^70,71. Homeostatic functions are not exclusively exerted by macrophages, but also by other leukocyte populations including mast cells, innate lymphocyte cells (ILCs), eosinophils, as well as conventional lymphocytes^72–77. A more detailed explanation of the contribution of these cells to homeostasis is beyond the scope of this manuscript and we refer the reader to excellent manuscripts^3,20,67 that have discussed the corresponding cell-type-specific biology in depth.

Akin to tissue-resident leukocyte programs, we argue, in the next section, that leukocyte trafficking follows a certain logic, which likely contributes to tissue and hence organismal homeostasis.

The logic of leukocyte re-compartmentalization

Pattens of leukocyte trafficking in homeostasis or upon non-infectious insults are complex and may appear unpredictable, but we suggest that they follow general principles. From a simplified perspective (reviewed in depth elsewhere¹), tissue leukocyte compartment sizes are a function of cell efflux, influx, proliferation, differentiation and half-life (Fig. 1). Efflux is typically mediated by downregulation of retention signals as exemplified by CXC motif chemokine ligand 12 (CXCL12)-signaling in bone marrow niches⁵⁷. Influx is primarily controlled at the level of adhesion molecules, which allow leukocytes to extravasate into tissue compartments as well as chemokine gradients and receptor expression^78,79. Proliferation and differentiation are regulated by cues within niches including paracrine factors and metabolites⁸⁰. Prolonging the half-life of cells usually requires their distribution to niches that can provide survival factors and suitable microenvironments.

Figure 1. Regulation of leukocyte compartment sizes.

Neuroendocrine signals and local cues control the principal mechanisms (labelled 1–5) by which the number of leukocytes within a given compartment can be adjusted. The sum of these changes determines the net size of the compartment.

It follows that whenever shuttling of leukocytes between compartments occurs, one site will experience a gain (influx), whereas the other site will show a loss in cell numbers (efflux), establishing a supplier and consumer relationship. This loss can be compensated by the three mechanisms described above or via reverse trafficking of leukocytes to their site of origin.

Due to the cost arising from the loss of homeostatic leukocyte functions in the supplying tissue, leukocytes- akin to oxygen, nutrients or metabolites- should only move from one site to another in response to a demand. We suggest that demands for leukocytes arise from an inability of the tissue to maintain its function in the face of a challenge. This challenge can be a noxious insult or occur as part of normal physiology.

Every tissue must handle local and systemic tasks, which we exemplify by describing these duties for four vital organs (summarized in table 2). The proposed distinction between local and systemic tasks is not absolute as many of these roles serve both the tissue as well as distant sites. The liver functions as a detoxifying (urea cycle, xenobiotics), excretory and degradative (bile, breakdown of hormones) as well as recycling organ (iron, heme), all of which are primarily local tasks. However, hepatocytes also release plasma proteins such as albumin, coagulation factors or lipoproteins, supply glucose to distant tissues, store triglycerides and glycogen and support platelet production in an endocrine manner^81,82. Likewise, the kidney filters large amounts of blood and reabsorbs many of the filtered solutes (e.g. glucose, electrolytes, peptides etc.), while excreting others along the tubular system ^83,84. Yet, the kidney is also the main source for erythropoietin (EPO) production in response to hypoxia, controls blood pressure through renin release, and activates Vitamin D to manipulate calcium and phosphate homeostasis^83,85. Similar principles apply to the heart, where cardiomyocytes deal with electromechanical coupling at the tissue level, but also regulate systemic plasma volume and blood pressure via ejection of blood into the arterial circulation and secretion of natriuretic peptides. The local function of the bone marrow is presumably ensuring the integrity of the hematopoietic niche, whereas the release of cells into the systemic circulation serves distant sites. Lymph nodes offer space for interaction between lymphocytes and antigen presenting cells, while also supplying leukocytes to other tissues.

Table 2.

Exemplary local and systemic tasks of different tissues.

tissue	local task	systemic task	alternative program of systemic task
liver	detoxification  • ammonia  • xenobiotics	synthesis  • plasma proteins (coagulation factors, albumin, lipoproteins etc.)	acute phase reaction
	degradation, recycling  • hormones  • heme/iron	supply  • gluconeogenesis	ketogenesis
	excretion/absorption  • bile	storage  • glycogen, triglycerides  • vitamin A, B12
kidney	filtration	blood pressure control  • renin release  • filtration, reabsorption	change in GFR quality or quantity
	solute reabsorption  • water, glucose, electrolyte, minerals	oxygen sensing  • EPO secretion	acidosis sensing  • HCO3-, NH4+ production
	solute excretion  • H+, organic acids, metabolites etc.	supply  • gluconeogenesis	excretion  • glucosuria
		calcium/phosphate homeostasis  • calcidiol activation	phosphate wasting
heart	electromechanical coupling	Plasma volume and pressure control  • stroke volume  • ANP/BNP release	inactive BNP/ANP release (“BNP paradox”)
bone marrow	maintenance of hematopoietic niche	cell supply	emergency myelopoiesis

Alternative programs of the systemic tasks are listed in the fourth column.

When dysfunctions of local tissue tasks arise, they are typically countered by local defense mechanisms as typified by the tubuloglomerular feedback mechanism in the kidney, fibroblast growth factor 15/19-signaling in the liver or paracrine epidermal growth factor (EGF) receptor activity regulation in the bone marrow^86–88. In contrast, challenges affecting the organ’s systemic tasks will require a systemic response. This creates what we refer to as the “local demand”, which results in leukocyte trafficking to the site of demand. When tissues produce a signal of local demand, the predicted outcome should be the maintenance of the systemic tasks. This demand may be part of normal physiology such that the tissue requires specific functions, information or other cargo transported by leukocytes to maintain its systemic task. This idea is supported by work demonstrating that time-of-the-day dependent trafficking of neutrophils into the liver supports metabolic homeostasis via controlling clock gene expression⁸⁹. Likewise, circadian neutrophil migration to the bone marrow controls the size of the hematopoietic niche and very late antigen 4 (VLA4)-dependent trafficking of T cells to CNS sites is required for memory consolidation and learning^7,77,90,91.

However, animals may also face insults that confer an immediate threat to survival including systemic infection, starvation, or predation. These insults require a unique adaptive physiology, which we refer to as “global demand”. This global demand is likely not just the sum of local demands of each tissue but rather a derivation of the latter. This derivation is likely generated by the CNS following integration of the sum of peripheral inputs (local demands), and requires a hierarchical decision which tissue demands should be preferentially met because organismal resources are finite. During homeostasis, the global demand likely corresponds to the maintenance of the general fitness of the animal, which likewise requires central integration of peripheral inputs to generate an instructive output.

In cases of immediate survival threats such as systemic infection, the systemic task of the tissue can be reprogrammed, enhanced or suppressed (Table 2). For example, nutrient scarcity triggers fasting metabolism in the liver characterized by ketogenesis, which does not occur under steady state conditions⁹². Upon infection, hepatocytes engage in the so-called acute phase reaction characterized by a strongly increased production of a set of proteins including those required for iron sequestration¹⁸. Glomerular filtration is reduced or even shut-down completely in sepsis and high-output heart failure, which bypasses filtration of large amounts of blood and solutes in the presence of hypotension^84,93,94. Changes in the glomerular filter components can alter the quality of the filtrate⁹⁵. Beyond infection or injury, how leukocytes contribute to this reprogramming of tissue function or global organismal physiology in general remains mostly unclear. Consistent with the stereotypical patterns of leukocyte re-compartmentalization, we predict that subsets of these cells have functions that are indispensable for adaptive stress physiology.

As previously discussed in the context of infection, a tissue’s demand (consumer) may change dynamically over the course of the insult. Additionally, the cells meeting the demand need to shuttle between compartments that are anatomically segregated. Therefore, a coordinated, adaptive reorganization of leukocyte compartments requires continuous integration of peripheral inputs from multiple sites that are relayed to at least two tissues (supplier vs. consumer).

Moreover, the distribution of leukocytes to the site of demand often requires additional, supportive changes such as adjustments of blood pressure, fluid dynamics or vascular permeability to even enable delivery of leukocytes to the affected tissue. This highlights the need for coordination of multiple organ systems in parallel to achieve a functional response. Further, any significant organ damage has downstream effects on other organ systems, which itself will create a demand for leukocytes. In support of this paradigm, tissues not directly damaged by an inciting insult (brain, heart and abdominal cavity) experience a striking reconfiguration of their myeloid cell compartment⁹⁶. Thus, a single organ insult can generate multiple consumer tissues that demand different cells from supplying organs including the bone marrow, which must be orchestrated in space and time. This coordination is achieved by integration of peripheral inputs in the central nervous system, followed by the release of neuroendocrine signals to instruct the periphery.

Central and distributed control of leukocyte compartmentalization

The logic for choosing neuroendocrine commands as key regulatory checkpoints of leukocyte compartments is intuitive because (1) these systems allow for manipulation of multiple tissues in parallel (suppliers and consumers), (2) neuronal pathways and endocrine signals are fast acting and can reach distant sites and (3) the distribution of these signals throughout the body enables concomitant engagement of functions that support the leukocyte response (e.g. fuel supply for energetically costly programs such as inflammation), and (4) they are directly controlled by the CNS, allowing for the generation of a centralized, coordinated response tailored to the demands of the periphery as well as the overall organismal state.

Neuroendocrine outputs affect both supplying and demanding sites and typically feed into local control mechanisms. This systems design is not limited to leukocyte shuttling, but rather a general feature of physiology, whereby integration may occur inside or outside the CNS (Fig. 2). Below, we discuss the four layers of regulation of leukocyte re-compartmentalization: central control of supply, central control of demand, local control of supply and local control of demand.

Figure 2. Dynamic monitoring of local demands allows for coordinated distribution of biological goods via instructive signals.

(A) Local demands of tissues (A and B) are communicated to integration sites (C,D) that function in a hierarchy (C>D). Following integration of local demands, and communication between the integration sites, a signal output is generated that is relayed to the periphery (A,B) and induces a movement of goods from a supplying (B) to a consuming site (A). The resultant local changes are again monitored by the integration sites, yielding a dynamic circuit design (B) Exemplary circuit in the context of tissue infection. Hypo denotes hypothalamus, GC/SNS denotes glucocorticoids and sympathetic nervous system outflow. The signals or pathways that are used for bottom-up communication between tissues and the CNS, beyond cytokines, are poorly defined (C) Glucose/insulin circuit in exercising muscle (D) Bile acid/cholecystokinin (CCK) circuit upon meal ingestion with integration at the peripheral level (liver).

Central control of supply refers to the coordinated movement of leukocytes from a site of supply (a reservoir) to the site of demand instructed by neuroendocrine commands. The bone marrow is the main reservoir (i.e. supplying site) for leukocytes and engages in the program of emergency myelopoiesis in response to a diverse set of insults^23,24,36,97, which is governed by neuroendocrine signals. The key peripheral effector molecule of this response is granulocyte colony stimulating factor (G-CSF), a cytokine that can be secreted by various cell types including bone marrow endothelial cells upon pathogen detection⁹⁸. Other soluble factors such as IL1B, tumor necrosis factor, interleukin 6 or interleukin 3 also stimulate emergency myelopoiesis through mechanisms that may be partly G-CSF-independent^29,99. Neuroendocrine signals including the HPA-axis and the sympathetic nervous system (SNS) are upstream controllers of these signals. HPA-axis activation culminates in glucocorticoid (GC) release. GC physiologically rise in response to infection and potently induce G-CSF^100,101. Moreover, GC are required for the effects of G-CSF on the bone marrow¹⁰². The same is true for the SNS¹⁰³. This implies a hierarchy in which neuroendocrine commands override peripheral effector molecules such that G-CSF cannot function without the “allowance” of upstream controllers. The continuous integration of peripheral inputs from multiple sites in the CNS thus ensures an optimized host response via modulation of peripheral effector signals that govern leukocyte shuttling. Although cross-regulation between the HPA-axis and the SNS exists, the signaling mechanisms of the two systems are distinct: the HPA-axis acts host-wide through endocrine transmitters, whereas the SNS can function both locally via adrenergic nerves (norepinephrine) as well as systemically through catecholamine release (mainly epinephrine) from the adrenal medulla^104–106. Direct innervation of tissues such as the bone marrow is conceivably the fastest route through which leukocyte compartments can be modulated. Beyond the HPA-axis and SNS, little is known about possible effects of other canonical endocrine axes on leukocyte trafficking. Neither “tropic” hormones (FSH, LH, TSH, GH), posterior pituitary-derived factors (AVP, oxytocin), nor gland-secreted mediators (e.g. T3/T4, sex hormones, IGF1 etc.) have known regulatory roles in leukocyte re-compartmentalization.

Once the insult has been resolved, the stimuli for emergency myelopoiesis including neuroendocrine signaling cease, which returns the size and composition of the leukocyte compartments to homeostatic set points. However, there may be cases where homeostasis cannot be fully restored and the insult persists (e.g. tumor or chronic infection). In such scenarios, peripheral “islets of anarchy” may form, which are unresponsive to top-down commands and function autonomously (discussed below).

Neuroendocrine commands also affect the site of demand. GC regulate the expression of integrins and other adhesion molecules, chemokines as well as their receptors, which collectively shape leukocyte extravasation into tissues^107–109. Likewise, the SNS controls rhythmic recruitment of leukocytes to tissue compartments via adrenergic nerve-dependent regulation of chemokines and adhesion molecules such as ICAM1, VCAM1, E-selectin, CXCL12 and others^110,111. These mechanisms are strongly augmented in response to various types of stress, allowing for the stereotypical leukocyte re-compartmentalization discussed previously^32,33,36,111.

Although most mechanisms controlling the supply of leukocytes from organ reservoirs are subject to top-down control, these factors are also regulated in a tissue-intrinsic fashion as exemplified by tonic CXCL12 or G-CSF signaling in the bone marrow^112,113. Other signals that function in a paracrine fashion such as Dickkopf1 (DKK1) or Wnt ligands exist^86,114. Moreover, bone marrow adipocytes negatively regulate hematopoiesis via tissue autonomous mechanisms demonstrating that local crosstalk between different cell types can restrict supply¹¹⁵. Autonomous local control that is no longer amenable to central commands is usually a feature of disease as typified by extramedullary hematopoiesis accompanying bone marrow failure¹¹⁶ or local formation of tertiary lymphoid structures in chronic inflammatory diseases¹¹⁷. In the latter case, leukocytes are no longer recruited from distant sites (i.e. the bone marrow), but rather supplied locally by the newly formed structures. This strategy has two obvious advantages: first; cells are continuously and immediately available and second; the functional consequences for the formerly supplying tissue (the bone marrow) are mitigated because the demand is met by the local provider. On the other hand, the autonomy prevents coordinated control and higher integration, which can be costly and lead to chronic inflammatory states.

In contrast to supply, which is primarily coordinated centrally, demands arise locally in the periphery and are thus also controlled at the tissue level. These demands are dynamic during homeostasis and across the trajectory of an insult and specific for a certain number and quality of cells (further discussed below)¹¹⁸. In the case of infection, sensing of PAMPs and/or DAMPs through conserved sensors culminates in the induction of cytokines, chemokines and adhesion molecules that allow extravasation of leukocytes to the site of the insult¹¹⁹. The same effector hardware is used to recruit cells upon other types of injuries. During homeostasis, the oscillatory expression pattern of integrins and chemokines is established by the circadian clock, which enables time-of-the-day-dependent recruitment of leukocytes to tissue compartments⁵⁵. The circadian clock functions cell autonomously within leukocytes and endothelial cells⁵⁵, but also at the organismal levels, where the same molecular machinery synchronizes biological activities across tissues¹²⁰. Cytokines, chemokines and local cell-cell-interactions can rewire the function of morphologically identical cells in a way that allows for an optimal response to the demand of the tissue microenvironment^121,122. Local production of survival signals such as M-CSF to prolong the half-life of cells is another strategy to meet leukocyte demands¹²³.

According to these observations it becomes clear that supply and demand are physiologically matched to enable functional leukocyte trafficking between tissue compartments, which relies on local and central control mechanisms. Vice versa, a mismatch between supply and demand or global and local demand will be maladaptive. There are multiple conceivable mechanisms how such a mismatch may occur including loss of function of peripheral sensors, failure of transmission of the demand to the integration site, defective central integration, loss of signal output, impaired sensing at the supplying site, loss of function of supplier, loss of function of supplied goods or impaired trafficking of goods to the site of demand. These will ultimately prevent induction of the required physiological program. One can envision that this may be problematic in the case of local demand and most certainly fatal if the global demand cannot be met. Failure to meet global demands such as “starvation mode”, fighting invading pathogens or maintenance of thermoregulation all cause death. Although the roles of leukocytes are only well-defined for infection, all of the aforementioned programs are characterized by leukocyte re-compartmentalization and thus, the demanding sites should experience dysfunction when the leukocyte shuttling does not occur. Studying the contributions of leukocytes to tissue physiology in homeostasis and various stress states will thus be a field of interest for future studies.

Functional diversification of leukocyte responses

The system design principles outlined in the previous sections reveal several interesting features. First, the periphery must communicate a wide range of environmental challenges requiring distinct counterregulatory responses to central integration sites. Second, despite this heterogeneity, the output relies, as far as we know today, on a pretty narrow set of instructive signals and effector cells, although recent studies challenge this paradigm¹²⁴. Third, the functional outcomes of the induced responses are yet distinct (e.g. HPA-axis activation in the context of fasting or infection). Accordingly, several predictions can be made. The signals released by the site of demand likely have some specificity to the insult encountered and are a function of the sensors that monitor the corresponding variable. In other words, what is sensed determines the communication of what is needed. For example, upon fasting, leptin levels drop (presumably secondary to lipolysis¹²⁵), while infection triggers cytokine release following pathogen recognition receptor ligation¹²⁶. Both signals activate the HPA-axis, but trigger opposing patterns of leukocyte redistribution, demonstrating distinct functional outcomes^40,127–129. The amplitude, kinetics and spatial distribution of the induced signals can be tuned to change the type of information transmitted to the integration site. While the case for infection or fasting is clear, the “bottom-up” communication between tissues and the CNS is loosely defined, especially under homeostatic settings. Neuronal pathways, endocrine signals and cell trafficking are probable mechanisms involved in this communication.

Although the response generated to meet a demand in the periphery may be tailored to the respective challenge, this is not necessarily mirrored by the morphology of leukocytes recruited such that neutrophilia can be a feature of both psychological stress and infection, raising the question of how neutrophilia differs functionally in the two settings.

Two features of a response can be tuned: the quantity and the quality. Because the quantity (amplitude) by which leukocyte compartments are modulated is comparable across different conditions and this variable poorly correlates with biological outcomes^130,131, it is unlikely the main operating mechanism. On the other hand, adjusting the quality of a response can be achieved by several means and offers more room for fine tuning. We propose 5 categories that determine the quality of leukocyte responses (summarized in Fig. 3). First, the cell composition, which can differ between distinct insults and may involve appearance of cell types in the tissue compartment that are not present under homeostatic conditions. Second, the “state” of the cell that is recruited to the site of demand. We refer to “state” as a functional category of a cell at a given time, that is independent of its gross morphology. This state might be the result of inherent differences between cells within the same population (e.g. influenced by characteristics of the supplying organ reservoir) or alternatively, may arise secondarily within a given population due to variable microenvironments that these cells encounter^132,133. From the latter perspective, the neuroendocrine signal would send functionally similar cells to the site of demand, which then shapes these cells through local cues in a way that allows for optimal resolution of the insult. In both cases, the resulting phenotypic differences can be summarized as “functional heterogeneity”. Third, the kinetics of the response can differ. The speed by which cells reach their target, the duration of their supply, as well as the time required to return the donor and recipient compartment to homeostatic set points may all contribute to functional outcomes. The secretory patterns of endocrine signals determine their biological effects (pulsatile vs. continuous)¹³⁴ and we speculate that the same is true for leukocyte responses. Fourth, the systems that are engaged in parallel to the leukocyte response are critical (“coupling” or contextualization) for determining the type of overall response. For example, mobile cells are typically distributed via the blood stream and are thus dependent on fluid dynamics. Co-activation of neurohumoral pathways that promote vasoconstriction will result in different biological outcomes when compared to a vasodilative and vascular leakiness program, even if categories 1–3 are matched between responses. Finally, the microenvironment to which the cells home (site of demand) can create unique niches for the recruited cells that will shape their effector properties. Increasing evidence suggests that this mechanism is particularly important, at least for functional neutrophil diversification¹²¹.

Figure 3. Strategies for functional diversification of leukocyte responses.

General principles how the quality of a response can be adjusted are shown and labelled (1–5).

As all of the described components interact with each other, it is unlikely that the biological response can be predicted from the individual properties of the components, corresponding to the phenomenon of emergence in complex systems^135,136.

Immunodeficiency diseases and leukocyte tissue functions

If leukocyte trafficking critically contributes to physiology, the prediction follows that genetic diseases resulting in qualitative or quantitative defects of this process should impair homeostatic functions. Unsurprisingly, in most cases in humans, the phenotype resulting from such defects is primarily characterized by recurrent infections, which is also the main concern in the clinics. In contrast, more subtle defects in homeostatic functions may be clouded or simply be overlooked due to the severity of the symptoms caused by impaired host defenses.

Children suffering from severe congenital neutropenia (typically caused by loss of function mutations of ELANE) experience life-threatening infections that result in early-onset mortality without appropriate therapeutic interventions (G-CSF treatment)¹³⁷. Affected individuals are also susceptible to developing osteoporosis¹³⁸, which is difficult to explain with our current knowledge on neutrophils. Whether this complication is indeed a result of neutropenia or rather an epiphenomenon (inflammation secondary to recurrent infections) remains unknown. In analogy, the most prominent phenotype of patients suffering from severe combined immunodeficiencies (SCID), who are devoid of functional T cells, is an impairment of host defenses and resulting infections¹³⁹. Yet, experimental studies suggest that T cell defects entail a much broader deficit in physiology (see previous sections).

DiGeorge Syndrome, caused by a microdeletion on the long arm of chromosome 22, is another form of SCID associated impaired host defenses. This deletion involves 30 genes or more, yielding a distinct clinical phenotype, characterized by distinct facial features, congenital heart disease, intellectual disability, endocrine dysfunction and schizophrenia, among others. T cell deficiencies are a hallmark of DiGeorge syndrome, which results from disrupted thymic development¹⁴⁰. Whether T cell dysfunctions account for some of the clinical phenotypes remains to be shown.

The gut harbors large numbers of leukocytes including both cells of the innate and adaptive immune system¹⁴¹. In common variable immunodeficiency (CVID), the latter populations are perturbed, clinically manifesting as low immunoglobulin levels, which requires substitution¹⁴². While infectious complications are also the major concern in these individuals, a significant proportion of affected patients (approximately 10%) develop non-infectious enteropathy, which may resemble celiac disease and can transition into a chronic condition that is difficult to tackle therapeutically¹⁴³. While the underlying pathogenesis may be partly explained by dysbiosis and/or small intestinal bacterial overgrowth due to IgA deficiency, the mechanistic underpinnings of inflammatory villous atrophy with resultant malabsorption are mostly unclear¹⁴⁴. Likewise, the spectrum of genetic defects culminating in defective oxidative burst of neutrophil granulocytes (collectively referred to as chronic granulomatosis disease) are frequently characterized by non-infectious gastrointestinal pathology, that has an inflammatory component¹⁴³. Because inflammation is a conserved response to numerous insults including loss of tissue architecture and function¹¹⁸, the biological trigger of the aforementioned clinical correlate is unclear. These case studies collectively suggest that dysfunctions of both the innate and adaptive immune system are sufficient to perturb gut physiology.

Of note, none of these diseases allow for the distinction between defects resulting from loss of tissue-resident vs. trafficking functions and their low prevalence hinders large scale studies. Mice, on the other hand, can be housed in defined environments with low pathogen exposure, which can be leveraged to reveal leukocyte functions beyond host defense such as those being perturbed in the context of inborn errors of immunity.

Neuroendocrine diseases highlight the importance of central control of leukocyte compartmentalization

Another intuitive conclusion that arises from our discussions is that diseases and drugs that interfere with neuroendocrine signals provoke changes in leukocyte compartmentalization. Because the SNS and HPA axis are the main neuroendocrine effectors controlling leukocyte trafficking, we will now discuss some clinical conditions affecting these two systems.

In patients suffering from pheochromocytoma, autonomous growth of chromaffin cells of the adrenal medulla results in recurrent episodes of systemic catecholamine excess. Clinically, pheochromocytoma is characterized by symptoms of SNS hyperactivity such as tachycardia (heart racing), sweating, tremor (shaky hands) and hypertension (high blood pressure)¹⁴⁵. Consistent with the established function of the SNS in central control of supply and demand of leukocytes, affected individuals exhibit higher leukocyte and neutrophil counts than controls¹⁴⁶. Marked neutrophilia may even precede the development of overt symptoms of catecholamine overproduction in some of these patients¹⁴⁷. Because preclinical data suggests that the SNS modulates leukocyte compartments via adrenergic nerves, rather than systemic catecholamine release^32,57,110, the large amounts of catecholamines secreted by phaeochromocytomas are either sufficient to override this local regulation or activate an indirect pathway (e.g. the HPA axis). Data on leukocyte counts in patients with paraganglioma, a tumor of the SNS outside the adrenal medulla, are scarce. Supply of exogenous catecholamines is likewise associated with neutrophilia, a phenomenon that was first described over a century ago¹⁴⁸. On the other hand, inhibition of beta-adrenergic receptors lowers neutrophil counts in patients with heart disease¹⁴⁹.

Stereotypic changes in leukocyte compartmentalization are also observed in patients affected by conditions resulting in hyperactivity of the HPA axis. The common theme here is that blood neutrophil counts are directly correlated to the degree of hypercortisolism, while the inverse is true for lymphocyte numbers¹⁵⁰. As such, individuals with non-pituitary tumors producing excessive amounts of ACTH (referred to as ectopic Cushing’s syndrome, ECS) have severe hypercortisolism and neutrophilia, while lymphocyte counts are low¹⁵⁰. Markedly elevated ACTH levels in patients with ECS may further contribute to neutrophilia independently of GC. This idea is supported by human studies in which injection of ACTH induced neutrophilia in the absence of a functional GC response, whereas lymphocyte counts did not change¹⁵¹. Similar results were recently obtained in rodents, revealing that leukocyte subsets are controlled by divergent effector mechanisms of the HPA axis¹⁵². High ACTH levels may also be linked to neutrophilia in patients with subclinical Cushing’s syndrome, in whom GC levels are only minimally elevated¹⁵³. In contrast, GC excess is generally less pronounced in the context of adrenal tumors including malignant adrenal carcinomas. Consistently, changes in leukocyte compartmentalization are also milder¹⁵⁰. Upon therapeutic reversal of HPA axis pathology, leukocyte numbers return to homeostatic set points^150,154. As expected, blood leukocyte patterns in presence of loss of function of the adrenal gland due to autoimmunity (also known as Addison’s disease) mirror those described above and may be paralleled by neutropenia and lymphocytosis. Glucocorticoid replacement reverses these changes¹⁵⁵.

The most frequently encountered entity clinically among this spectrum is leukocyte re-compartmentalization in response to therapeutic doses of exogenous GC, which is characterized by rapid-onset neutrophilia, lymphopenia and eosinopenia¹⁵⁶. The implications of these changes for physiology and disease have not been thoroughly defined. There may also be cases where cells and tissues no longer respond to GC, a state referred to as glucocorticoid resistance (GCR), which most notably occurs in the context of inflammatory diseases including sepsis. This resistance is typically defined as an inability of GC to modulate transcriptional programs including those involving inflammatory pathways¹⁵⁷. However, GCR is also considered to occur in non-inflammatory diseases such as depression¹⁵⁸. It is not surprising that leukocyte trafficking is also affected by this state. In humans with GCR, the stereotypical associations between GC, neutrophils and lymphocytes are lost such that increased levels of GC are no longer linked to higher neutrophil and lower lymphocyte counts, respectively¹⁵⁹. It is currently unclear why a state of GCR develops and whether this is an adaptive feature or a bug of the system. Experimental data and correlative clinical evidence suggest that disrupted GC signaling is generally associated with unfavorable outcomes and deterioration of health^160,161. For example, GCR in murine sepsis entails an impairment of hepatic gluconeogenesis with resultant lactate build up, which drives mortality¹⁶². Alternatively, one could hypothesize that tissue-specific induction of GCR is adaptive by temporarily allowing for an augmented immune response, which comes at the risk of developing immunopathology¹⁶³. Another advantage of GCR could be uncoupling of downstream effector (GC) from its upstream controller (ACTH). This would, for example, enable selective relay of physiological changes via ACTH, while avoiding GC effects. These may include ACTH-dependent neutrophil mobilization, a process, that we are only beginning to understand¹²⁴.

Implications for clinical medicine

The framework discussed herein has implications for patient care. Pharmacological agents directly or indirectly interfering with leukocyte compartmentalization are now increasingly being used in the clinics. Such drugs include antibodies directed against VLA-4 (Natalizumab, Vedolizumab), inhibitors of sphingosine-1-phosphate receptor (S1PR, e.g. fingolimod), antagonists of cytokines or cytokine signaling targeting the IL1B (e.g. Canakinumab), IL6 (e.g. tocilizumab), TNF (infliximab, adalimumab, golimumab, etanercept etc.), IL4/13 (e.g. dupilumab) or IL12/23 pathway (e.g. ustekinumab), just to name a few^164,165. Moreover, small molecules simultaneously interfering with multiple of these signaling pathways are also available and GC exerting a wide range of immunomodulatory effects are still the treatment of choice for many inflammatory conditions^166,167. Furthermore, some of these drugs have shown promising results for the treatment of non-traditional inflammatory diseases that affect a large share of the population as exemplified by the CANTOS trial, where canakinumab ameliorated atherosclerosis and its sequalae¹⁶⁸. We thus predict that the clinical use of these drugs in unexpected settings will likely further increase. This naturally emphasizes the need to develop a deeper understanding how the immune system contributes to general tissue and organismal physiology as these functions will experience perturbations in response to such therapies, resulting from a mismatch between supply and demand.

For example, the plethora of side effects that come with prolonged, high-dose GC therapy are well-known¹⁶⁹, but whether and how leukocytes contribute to these symptoms is poorly understood. Impaired lymphocyte trafficking to the CNS compartment secondary to downregulation of vascular adhesion factors, leukocyte integrins as well as lymphocyte sequestration in the bone marrow, all of which are consequences of GC signaling (see previous sections) may negatively impact mood and cognition. Likewise, metabolic derangements could arise from dysregulated leukocyte compartmentalization in the liver because local demands are not being met⁸⁹. Along the same lines, physical weakness in patients treated with S1PR antagonists might be a result of unmet local demands of muscle tissue because the requirement of regulatory T lymphocyte homing to this compartment to maintain its functions has been elegantly demonstrated in mice^170–172. Tackling the biological basis of the symptom complex of fatigue, a common drug side effect and symptom of disease, from leukocyte-centric perspective could prove valuable. Finally, whether and how dysregulated leukocyte trafficking contributes to abnormal liver tests and hepatotoxicity, a shared side-effect of most pharmaceuticals, could be tested experimentally.

Concluding Thoughts and Future Directions

The concepts of global and local demand, central integration, balanced decision making and interorgan communication outlined in this article are widely applicable to a range of disciplines and go beyond leukocyte compartmentalization. The supplier-consumer relationship is obviously not restricted to cells moving from one site to another but rather applies to any “goods” including nutrients, metabolites, or information (e.g. encoded as electrical or chemical signals)¹⁷³. Studying the “bottom up” communication between peripheral tissues and the CNS is of interest from both a conceptual as well as therapeutic point-of-view. The prediction is that classes of signals that are induced in response to certain categories of challenges, akin to inflammatory cytokines upon infection, are centrally integrated and translated into a neuroendocrine output, which shapes global physiology. Experimental evidence for this paradigm is starting to emerge^174,175. The signals communicating demand as a function of normal physiology are particularly understudied and worth exploring as are the contributions of leukocytes to dealing with the corresponding challenges.

It will further be interesting to better define at which point central integration sites decide to sacrifice global physiology in favor of local demands. For example, sleep is essential to the maintenance of homeostasis but several inflammatory conditions are linked to perturbed sleep, despite sleep’s established functions in promoting recovery^4,176–179. Beyond the apparent confounding factors, this may suggest that sleep needs to be actively suppressed to meet the demands of inflammation if these are exceptionally high. From an extreme point of view, this concept suggests that not only physiological functions, but also human behavior is the collective result of tissue states in the periphery, which may also be a helpful way to think about host-microbiota-interactions. Additionally, tissue biology can be strongly affected by the CNS, even if peripheral inputs no longer communicate demand¹⁸⁰, providing avenues to explore how the brain fuels, maintains or incites (e.g. autoimmunity or psychosomatic disorders) disease states. The endocrine relay of CNS decisions also requires further investigations. Many endocrine mediators abundantly present in the circulation have narrowly defined actions such that “-tropic” hormones including ACTH or TSH are considered to exclusively control the production of downstream effectors. We have recently provided evidence that this likely not true for ACTH¹²⁴ and others have reported similar observations for FSH^181–183.

In summary, leukocyte re-compartmentalization is a feature of homeostasis and a hallmark of adaptive stress physiology. This re-compartmentalization is a function of local and/or global demand and requires dynamically coordinated leukocyte shuttling in space and time via neuroendocrine commands to avoid maladaptive mismatches between supply and demand. Accordingly, we predict that the next decade of scientific discoveries will reveal many contributions of the immune system to general and adaptive physiology.