Macrophages on the margin: choroid plexus immune responses

Abstract

The choroid plexus (ChP), an epithelial bilayer tissue containing a network of mesenchymal, immune, and neuronal cells, forms the blood-cerebrospinal fluid (CSF) barrier. While best recognized for secreting CSF, the ChP is also a hotbed of immune cell activity and can provide circulating peripheral immune cells passage into the central nervous system (CNS). Here, we review recent studies on ChP immune cells, with a focus on the ontogeny, development, and behaviors of ChP macrophages, the principal resident immune cells of the ChP. We highlight the implications of immune cells on ChP barrier function, CSF cytokine and volume regulation, and their contribution to neurodevelopmental disorders, with possible age-specific features to be elucidated in the future.

The ChP mediates neuroimmune interactions

The choroid plexus (ChP, see Glossary) is located within each brain ventricle and forms the blood-cerebrospinal fluid (CSF)-barrier (BCSFB). The ChP is commonly recognized as a source and regulator of CSF [1–5]. In recent years, the ChP has also become increasingly appreciated as a key part of the central nervous system (CNS) neuroimmune network (Box 1, [6–9]). This review aims to summarize recent advances in our understanding of ChP immunity, with a focus on development. We refer the readers to several recent reviews on other aspects of ChP biology, including ChP regulation of the CSF and the molecular aspects of ChP barrier function [2,3,6,10–12].

Box 1. Neuroimmune compartments in the brain.

Brain resident macrophages comprise microglia in the brain parenchyma and border-associated macrophages (BAMs) within the brain barriers. BAMs have also been referred to as CNS-associated macrophages, or “CAMs” [24]; in this review we follow the “BAMs” terminology. Anatomically, BAMs are resident macrophage populations at the CNS borders, including meningeal macrophages at the brain surface, perivascular macrophages at the blood-brain barrier (BBB), and choroid plexus (ChP) macrophages at the blood-cerebrospinal fluid (CSF) barrier. During neuroinflammation, BAMs interact with T cells by presenting antigens to them at the brain barriers [20]. Notably, the CSF is also filled with a dynamic population of immune cells [17,18,110,111], which implies possible immune cell trafficking and the crosstalk of immune response among anatomically distinct barriers. Together, the immune cells located in all CNS compartments form a highly dynamic and complex network, maintaining the health and homeostasis of the CNS. The ability to access, target, and modulate the immune landscape at the barriers could provide additional therapeutic targets for disease interventions. A number of recent, excellent reviews have discussed the following topics in great detail: brain resident macrophages [21,24,25,112], meningeal immunity [23], immune activities at the brain barriers [7,113], the development and anatomy of the blood-CSF barrier and other brain barriers [3,6,10], and T cell immunity at the ChP during neuroinflammation [8,9]. Leveraging in vivo imaging, immune cell movement at the meninges has been characterized during neuroinflammation (e.g., T cell movement using a murine experimental autoimmune encephalomyelitis model [114]; macrophage movement using traumatic brain injury model [48]). However, current understanding of the similarities and differences among BAMs of different barriers is limited (Table 1). It is also unclear whether and how BAMs and other immune cells in different CNS compartments interact to maintain homeostasis. Lastly, the question of whether one barrier might out-perform the others, and if so, when and how, is a matter of ongoing discussion and debate [66,114].

Multiple studies have reported increased numbers of immune cells at the ChP and in the CSF following immune challenges to the CNS [13–18], fueling a growing interest in the role of the ChP in brain immunity. The ChP harbors a variety of immune cells including macrophages, basophils, mast cells, dendritic cells, monocytes, neutrophils, and lymphocytes [19]. In this review, we will focus on the most plentiful of these: macrophages, discussing their development and functions at different stages of life (Figure 1). Macrophages throughout the body are well-appreciated for their specialized roles in the detection, phagocytosis, and destruction of harmful entities. They can also present antigens to T cells, release cytokines that trigger signaling cascades to activate other immune cells, and play a role in tissue repair following injuries [13,20]. Resident macrophages at the other brain barriers (e.g., meninges and blood-brain barrier (BBB)) have been well studied and recently reviewed (Box 1, Table 1, [7,21–25]). However, macrophages at the ChP represent an immune population that has been historically underappreciated. Here, we review technical advances and new insights into ChP macrophage developmental origins, molecular and functional identities, and possible roles in health and disease.

Figure 1. The diverse identities and functions of choroid plexus (ChP) macrophages across development.

Past studies and recent single-cell transcriptomics analyses revealed that the ChP harbors a variety of immune cells including macrophages, basophils, mast cells, dendritic cells, monocytes, neutrophils, and lymphocytes (top, center). For T cell activities at the ChP, we refer the readers to earlier reviews [7,9] Studies in mice showed that already during early brain development, ChP macrophages reside at two distinct anatomical positions—on the apical surface (epiplexus) and in the stromal space. In vivo live imaging of fluorescence-labeled macrophages (e.g., Cx₃cr₁^GFP/+; top, left and top, right) using two-photon microscopy paired with computational algorithms has revealed diverse behaviors of ChP macrophages [27,28]. During development (bottom, left), macrophages are round with short, stunted processes. They survey the surroundings and can enter the CSF via distal tips of ChP villi following inflammatory triggers. In mature ChP (bottom, right), macrophages have long, ramified processes. In addition to routine surveillance, mature macrophages can actively take up blood-borne substances including fluorescent dextrans. Stromal macrophages elongate in the periluminal region, as if coating the blood vessels, during LPS-induced systemic inflammation, while epiplexus macrophages immediately rush to laser-induced focal injury sites on the ChP. aCSF: artificial CSF; T1, T2 and T3: indicate different timeframes to demonstrate the processes are motile over time.

Table 1.

Features of macrophage at different brain barriers and in CSF

	ChP macrophages	Meningeal macrophages	Perivascular macrophages	CSF macrophages
Lineage	Yolk sac, fetal liver, bone marrow derived monocytes [31,32]	Yolk sac, fetal liver, skull bone marrow derived monocytes [31,32,116]	Yolk sac, fetal liver [32]	Unknown
Turnover	Partial turnover [31,32]	Long lived [32]		Unknown
Transcriptomic signature shift during development	Yes [19,41]	Unknown	Unknown	Unknown
Transcriptomic signature shift during aging and neuroinflammation	Yes [19,20,35]			Yes [18,111]
Homeostatic behavioral dynamics	Motile and mobile [27,28]	Motile [20,48]	Motile [32]	Unknown
Interactions between macrophages in different compartments	Unknown

The origin and molecular identities of ChP macrophages

Macrophages are myeloid phagocytic cells that reside in all tissues and organs, including the ChP. Determining the developmental trajectory of macrophages is challenging because of their propensity to move and adopt different morphological and transcriptomic features during development and aging, as well as in response to environmental cues. For example, bone-marrow derived monocytes can infiltrate tissues and differentiate into macrophages [26].

Both brain parenchyma and the ChP are seeded with macrophages originating from the yolk sac early in development (mouse embryonic day [E]10.5 in the brain parenchyma and E11.5 in the ChP), and they both establish an evenly tiled spacing pattern by mid-gestation [19,27–29]. ChP macrophages are found both in the stromal space between the epithelial and endothelial layers (stromal macrophages) and along the apical epithelial surface (epiplexus macrophages; Box 2). It is still unclear whether macrophages that seed brain parenchyma and the ChP originate from the same precursor cells [30,31]. Fate mapping studies revealed constant monocyte trafficking to the ChP throughout adult life to replenish resident macrophages [32]. The majority of ChP macrophages labeled by tamoxifen-induced fluorescence at E9 (in mice) are no longer detected by 6 weeks of age (P42), as shown by a decrease of labeling rate from ~50% to ~10%, suggesting a steady turnover of the resident macrophage population [32]. CCR2-expressing monocytes in the blood stream are not believed to contribute to the ChP macrophage population during embryonic development but are reported to routinely enter the ChP in adulthood, even in the absence of inflammatory stimuli [32–35]. In contrast, resident macrophages in other barrier tissues, such as perivascular macrophages and meningeal macrophages, have minimal turnover [32,36]. Recent work suggests that the development of epiplexus macrophages is Csf1r-dependent, similar to parenchymal microglia, whereas the development of stromal macrophages is only partially Csf1r-dependent [37]. These recent finding suggests more complex developmental regulation of ChP macrophages than previously appreciated, with location-specificity.

Box 2. ChP macrophages and the mysteries of Kolmer cells.

The identities and functions of ChP macrophages are only beginning to be elucidated. Macrophages populate the ChP at two distinct anatomical positions—on the apical surface (epiplexus) where they are constantly exposed to the CSF, and in the stromal space, presumably in closer communication with mesenchymal cells and blood-borne factors. These cells exhibit location-specific behaviors (discussed in “ChP macrophages in motion”).

Epiplexus (apical) immune cells of the ChP have long intrigued researchers, for their unique positioning has the potential to facilitate their trafficking from the ChP throughout the CNS. According to most accounts, macrophages at the ChP-CSF interface were first described in 1921 by W. Kolmer, and they have been often referred to as “Kolmer cells,”although they are also commonly called “epiplexus cells” [6]. Early on, two types of Kolmer cells were proposed: Type 1 were fibrous with numerous processes radiating from the cell body, whereas Type 2 had fewer pseudopod-like processes. In this review, we refer to all apically positioned macrophages in the ChP as “epiplexus cells,” but whether these cells are strictly located at the ChP at all times remains to be determined. Cells with similar morphology to epiplexus cells are also found along the ventricle walls [115]. While their functions and identities remain largely elusive, the morphology and position of epiplexus cells leads to several key questions: Do epiplexus cells survey the entire ventricular system and transiently gather at the ChP? Do epiplexus cells travel into and out of the ChP stromal space? Are epiplexus and stromal cells two distinct populations that do not mix? Are Type 1 and Type 2 Kolmer cells distinct cell types, or rather, the same cell population reflecting differing activity/functional states? Macrophages are commonly detected in CSF samples [13,18,111]. Do Kolmer cells travel to different CNS locations via the CSF as well as the ventricle walls? Notably, a subset of embryonic epiplexus cells express Lyve1 [19], which has been proposed to be a meningeal macrophage marker [20]. This finding supports the diversity of epiplexus cells and raises the possibility that epiplexus cells may travel throughout the ventricular system under certain circumstances. Integrating transcriptomic and imaging approaches will help elucidate the identity and movement patterns of epiplexus (Kolmer) cells. Advanced live imaging approaches should provide additional information on the cellular behaviors and trafficking of BAMs more generally.

Studies of transcriptome signatures are beginning to elucidate the molecular and functional differences among macrophages from different CNS compartments. For example, adult border-associated macrophages (BAMs, [24]) including those in the ChP express CD206 (MRC1) while microglia lack CD206 [31,32]. ChP macrophages adopt transcriptomic signatures that distinguish them from microglia as early as E12.5 in mice [19,30–32]. Currently, a handful of genes have been proposed as ChP macrophage signature genes (Table 2). However, the temporal and spatial heterogeneity in the molecular signatures of macrophages from all parts of the CNS exceeds earlier expectations. As detailed below, many identified markers show differential expression over the course of development, and the tissue specificity of newly proposed markers remains uncertain.

Table 2.

Examples of currently identified transcriptional signatures of developing and mature ChP macrophages.

	Developing	Mature	References
Cx3Cr1	High	High	[19,24,27,28,32]
Ccr2	Yes*		[19,32,35]
Csf1r **	High		[19,21,37]
Iba1	High		[27,32]
Cd206 (Mrc1)	High	Low	[19,20,31]***
Lyve1	Low	None**	[19,20]
Pf4 **	High	Low	[19,117]
Slc40a1	High	Low**	[19]
Tgfbr1 **	Medium	High	[19]
Cd74	Low	High	[19,20]

^*Ccr2 is expressed by a small subset of macrophages.

^**Expression at different ages of these genes is not yet validated by conventional molecular or cell biological approaches (qPCR, immunoblotting and/or IHC).

^***Cd206 is enriched in all BAMs but not microglia. Other transcriptomic signatures of BAMs and microglia have been reviewed in detail [25,118].

Gene expression can be differentially regulated during development, which complicates the ability to distinguish BAMs from parenchymal microglia. For example, during embryonic to neonatal development, ChP macrophages and other BAMs share some transcriptomic signatures with parenchymal microglia, implying similarity in identity and function (e.g., Cd206 and Axl [19,38–40]). On the other hand, ChP macrophages themselves exhibit shifts in gene expression with maturation (e.g., Cd74 is enriched in adult and aged ChP macrophages, while Cd206 is downregulated in adult and aged ChP macrophages [19,38,41]). Therefore, dissecting the heterogeneity of BAM transcriptomes with enhanced temporal resolution is needed to advance our understanding of BAM dynamics across different developmental periods.

Even at a given developmental stage, discrepancies exist regarding the specificity of transcriptomic signatures of ChP macrophages vs. other cell types. Sall1 was proposed as a canonical parenchymal microglial marker [35,42,43]. However, data from another study [44] suggest that epiplexus cells (Box 2) also express Sall1. This study also suggests that these ChP epiplexus cells possess a microglial transcriptome signature and are not replenished by circulating monocytes [44], unlike their stromal counterparts [32]. Thus, determining whether Sall1 is a reliable epiplexus cell marker and whether ChP epiplexus cells share origins with microglia will require further characterization. Likewise, several markers previously suggested to be subdural BAM-specific have been identified in embryonic ChP macrophages (e.g., Lyve1, Ednrb, Colec12 [19,44]), suggesting either a shared origin or trafficking of these cells among different barrier structures.

ChP resident macrophages themselves possess additional transcriptomic signatures associated with distinct spatial niches beyond stromal vs. epiplexus location [19]. As an example, Slc40a1/Ferroportin, an iron exporter, is expressed by vessel-adjacent ChP macrophages, suggesting potential roles in iron metabolism [19]. Both bulk RNAseq and single cell RNAseq have revealed ventricle-related heterogeneity in other ChP cell types including epithelial and mesenchymal cells [1,19]. It remains to be determined if ChP immune populations possess ventricle-specific transcriptomic signatures, which would further diversify the ChP immune population and functional complexity. Altogether, it is tempting to speculate that the spatial positioning of ChP macrophages near the blood (stromal cells), the CSF (epiplexus cells), or in different ventricles endows the cells with particular roles. As an example, stromal ChP macrophages actively take up fluorescent dextrans from the blood, while a subpopulation of macrophages (epiplexus) traverse long distances across the ChP surface (discussed also below in “Macrophage surveillance activity during homeostasis,” [28]). Collectively, these observations represent the beginning of a journey to elucidate spatially specific macrophage behaviors.

Challenges in reconciling currently published findings into one simple model for ChP BAMs stem largely from technical limitations of transcriptomics approaches that vary among laboratories, including differences in sample age, sequencing depth of coverage, and controlling for cross-contamination. Epiplexus cells may be prone to detaching from the ChP tissue during dissection and manipulation, raising questions as to whether these cells have been successfully captured in all single cell transcriptomic analyses. Additionally, transthyretin (Ttr), a highly abundant marker gene of ChP epithelial cells, is sometimes detected in other cell types including immune cells [44]. This expression may be the result of the macrophages ingesting parts of epithelial cells, artifact from tissue dissociation, or uncorrected cytosolic contamination by the vast abundance of epithelial cells and Ttr transcripts that can be computationally accounted for [19]. CellBender represents one promising computational tool for removing ambient RNA contamination [19,45]. Additional sources of variability include the genetic background, sex, age of the animals used for each study, as well as ventricles being analyzed. Collectively, these caveats invite careful interpretation of existing data to improve ChP cellular profiling. Multiple approaches are necessary to validate findings, especially as development and transitions to disease pathology occur in a continuum.

ChP macrophages in motion

Macrophage surveillance activity during homeostasis

Macrophages constantly survey their local environment via motile processes that can extend rapidly towards sites of tissue injury [46–49]. While meningeal macrophages and microglia have both been visualized in vivo a number of years ago [32,48], it is only recently that ChP macrophages have been visualized in their native environment (Figure 1) [28]. Two-photon microscopy paired with an imaging cannula and fluorescent labeling of macrophages (e.g., Cx₃cr₁^GFP/+) enabled imaging of lateral ventricle ChP deep inside the adult mouse brain. Following computational algorithms that correct and stabilize large-scale and non-rigid motion of the ChP, which floats in the ventricle’s CSF, individual macrophage movements can be extracted from volumetric imaging data at high resolution [28]. Using fluorescent dextran to label the vasculature, volumetric imaging also enables differentiation between stromal and epiplexus cells, providing a useful tool to interrogate the functional behaviors of these two subsets of ChP immune cells [28].

Mature ChP macrophages constantly explore their surroundings with their distal processes. Substantial differences in baseline movements distinguish stromal macrophages from epiplexus cells [28]. While stromal macrophages have relatively stationary cell bodies, their processes have high motility and continuously monitor their surroundings. Their processes show contact-mediated inhibition by retracting upon contact with neighboring macrophages, suggesting the ChP macrophages each survey their own territory. The processes of stromal ChP macrophages also frequently contact blood vessels within the ChP stromal space and take up foreign material from the circulation. These findings are consistent with the phagocytic signature of ChP macrophages, and their homeostatic surveillance is similar to other brain resident macrophages [40,41,47,49]. In contrast, epiplexus cells exhibit substantial cell body mobility, saltatory movements at times, and can travel at a speed of up to hundreds of micrometers per hour at the ChP surface [28]. Such mobility has not been observed in other BAMs or microglia. These observations raise questions regarding the identities, functions, and trafficking of spatially distinct macrophages across various brain compartments (Box 2; and section “ChP macrophage responses to environmental cues”).

Embryonic ChP macrophage have amoeboid morphologies with short, stunted processes, which contrasts with adult ChP macrophages that have ramified shapes with long thin processes [19,27,28,32]. Nonetheless, the cell processes are motile and cell mobility is evident in some cells [27]. Intriguingly, these embryonic ChP macrophages exhibit prolonged intercellular contact, suggesting that at E14.5, the contact-mediated inhibition seen in the adult ChP macrophages is not yet fully established. Additionally, the macrophages in embryonic ChP can be found near apoptotic cells [30], suggesting they participate in clearing cellular debris.

Postnatal ChP macrophages possess an antigen presenting signature (e.g., Cd74) and functionally interact with T cells during neuroinflammation [20,41,50]. Located right at the blood-CSF interface, the dynamic behaviors of ChP macrophages imply their active participation in ChP immunity and barrier function. Identifying the mechanisms that regulate the development and maturation of ChP macrophages will help elucidate their functional capabilities at different ages.

ChP macrophage responses to environmental cues

ChP macrophages respond to cytokines and chemokines, and possibly also to purines. For example, in response to elevated CSF-CCL2, a subset of embryonic ChP macrophages show higher motility and mobility. Considering the potential heterogeneity of the macrophage population already at early developmental stages (Box 2), these findings also raise the question of whether heterogeneity (e.g., epiplexus vs. stromal location) underlies the differential responses to CSF-CCL2. Histological analyses reveal that the stromal cells abandon their typical tiling pattern, and instead, cluster at the free margin of the lateral ventricle ChP and at anatomically distinct “hot spots” at the distal tips of 4^th ventricle ChP villi [27]. In a peripheral inflammation model using LPS, adult stromal macrophages flatten and extend along the periluminal region of vessels [28]. By contrast, epiplexus macrophages do not change their morphology in this peripheral LPS model [28]. These observations underscore how different stimuli trigger distinct actions in recruited immune cells (e.g., pro-inflammatory vs. anti-inflammatory [13,14,16]).

From the earliest stages of development, ChP macrophages express an array of purinergic receptors [19,51,52], which are known to mediate inflammatory responses in other immune cells. For example, signaling through ionotropic (P2X4 and P2X7), metabotropic (P2Y1, P2Y2 and P2Y12), and purinergic receptors regulates cell membrane ruffling, motility, mobility, and migration of microglia. T cells and neutrophils also respond to changes in levels of extracellular ATP and its metabolites following tissue injury and inflammation [47–49,53,54]. ChP macrophages, especially epiplexus cells, are first responders that rapidly home to sites of focal injury, as demonstrated in response to acute heat [28]. Purine-dependent mechanisms may contribute to the regulation of this ChP macrophage response.

ChP as a coordinator of CNS inflammatory responses

Inflammation is an important defensive response to infection or injury, but excessive or prolonged inflammation can be detrimental and is correlated with the vulnerability to a number of CNS disorders throughout life [55,56]. The ChP has been shown to be an entry site for pathogens, a checkpoint for peripheral immune cells into the CNS, and a regulator of cytokines and other signaling molecules in the CSF [8,57,58]. In light of this evidence, it is becoming increasingly clear that the ChP may play critical roles in optimizing the amplitude and duration of brain inflammation. Below we review ChP’s roles in viral entry and immune cell recruitment, with an emphasis on ChP involvement in neurodevelopmental disorders.

ChP as gateway for pathogens into the brain

The ChP is suggested to serve as an entry site for pathogens, such as Coxsackievirus B3 (CVB3) and Haemophilus influenza type b (Hib), which cause viral and bacterial meningitis, respectively, and many others [57]. In Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), the novel coronavirus responsible for the recent COVID-19 pandemic and reported to cause neurological symptoms, has been detected in patient CSF [59]. Recent studies using human brain organoids suggest that SARS-CoV-2 viral particles can enter CSF space through the ChP [60,61]. Postmortem ChP from COVID patients and mouse ChP mesenchymal cells, and epithelial cells to a lesser extent, express Ace2, the gene that encodes the receptor for SARS-CoV-2 cell entry, as well as accessory proteins (e.g., Ctsbl, Ctsn and Nrp1 [19,62,63]). Collectively, these studies suggest that the ChP along with the nasal mucosa may contributes to neurological sequelae associated with SARS-CoV-2 infection [64], further highlighting the ChP as a pathogen entry point into the CNS.

ChP shepherds immune cells into the brain: evidence from neuroinflammation models

Following CNS injury or infection, a surge of CSF cytokines propagates an inflammatory response [13,14,27,65]. Cytokines and chemokines in the brain and CSF mediate the local accumulation of ChP immune cells as well as the recruitment of immune cells into the CNS through the ChP (e.g., CCR2 signaling for monocyte infiltration in stroke and maternal inflammation, CCR6 signaling for Th17 recruitment in EAE; [16,27,66]). Leukocyte trafficking across the ChP is accompanied by altered expression of cell adhesion molecules and disruption of barrier junctions [13,27,67–69]. During inflammation, the ChP also releases cytokines into the CSF, including but not limited to CCL2, CCL5, IL6, and TNFα [13,14,27,65], providing a chemoattractant driving force for immune cells to enter the CNS. Accordingly, neuroinflammation leads to increased immune cells in the CSF [17,18,70].

The ChP coordinates immune cell recruitment into the CNS in various pathological conditions including tumor metastasis [71], infectious diseases [57,72], autoimmune disorders [66,73], neurodegenerative disorders and aging [50,74–78], and other forms of CNS injury [13,14,70,79,80]. For additional discussions on these topics, we refer the reader to several thorough reviews on the topic of neuroinflammation and pathophysiology of disease [6–9]. From a translational perspective, it has been suggested that intervening with ChP-mediated immune infiltration could provide an avenue for treating neuroinflammatory conditions. Notably, in mouse models of Alzheimer’s disease, either promoting monocyte infiltration across the ChP or blocking ChP IFNγ signaling can improve cognitive function [74,76,81]. With current limitations in the knowledge of ChP immune function, it is still early to make a definitive argument on whether the ChP immune response can be considered a “friend” or “foe” in different disease settings. Improved understanding of ChP immune dynamics will provide clearer mechanistic insights for developing potential therapeutic strategies that harness ChP immunity to provide optimal protection for the CNS in various diseases.

ChP immune activation in neurodevelopmental psychiatric disorders

Extensive epidemiological evidence supports the model that exposure to inflammatory illnesses during development, possibly together with underlying genetic susceptibilities, can contribute to enduring neurodevelopmental conditions including psychiatric illness [56,82,83]. ChP and/or CSF inflammatory abnormalities are reported in many of these disorders. For example, a study of 4–10 year old children with autism spectrum disorder (ASD) found increased levels of inflammatory cytokines in their CSF [84]. In another ASD cohort (5–68 years old), postmortem histology shows T cell infiltration at the brain-CSF interface [85]. The ChP from patients with schizophrenia exhibits an inflammatory transcriptomic signature [86]. This observation is consistent with the broad inflammatory features associated with schizophrenia [87], including excessive expression of complement component 4 (C4), a key molecule in innate immunity and synaptic pruning. BCSFB permeability to serum proteins is also increased in schizophrenia [88].

To seek further mechanistic understanding of how ChP inflammation may participate in these conditions, the ChP has been investigated in several rodent models of neurodevelopmental disease. Maternal immune activation (MIA) can be modeled in rodents by mid-gestational exposure to a synthetic viral genome mimetic, polyinosine-polycytidylic acid (polyI:C), which leads to the dysregulation of fetal brain development and alterations of core behavioral phenotypes that share certain features with neurodevelopmental disorders including ASD [89–92]. In the MIA model, the embryonic ChP can propagate peripheral inflammatory signals into the developing brain by releasing cytokines such as CCL2 and recruiting immune cells to enter the CSF at the tips of ChP villi [27]. Some macrophages at the ChP are proliferative, demonstrating that the ChP can also be a site for immune cell proliferation during inflammation [27,69]. Compromised ChP barrier integrity caused by ChP-associated inflammation contributes to an altered neuroimmune landscape in the developing brain. The fates and final destinations of immune cells recruited to enter the CNS from the ChP remain to be elucidated. However, growing evidence supports a model that once in the brain, these cells derail normal development, such as by dysregulating cortical progenitor cell proliferation [27,91–94]. The modulation of immune cell CNS entry by the ChP is also reported in neonatal brain injury models [16,95].

ChP may also contribute to pathological hallmarks of neuropsychiatric disorders by altering CSF dynamics. In patients with psychosis, cognitive deficits are associated with increased ChP volume and enlarged ventricles [96]. Patients with schizophrenia can have enlarged brain ventricles as early as their first episode [97,98]. Similarly, some individuals with ASDs (0–32 years old) also have enlarged ChP and ventricles [99]. Excessive and persisting extra-axial CSF in young children is associated with the later diagnosis of ASD and has been argued to be predictive of the severity of outcome [100–103], but such associations have not been reported in schizophrenia [104]. Fluid secretion can be dysregulated by inflammation in other secretory epithelia, such as the intestine, by interfering with ion transporter activities [105–107]. The regulation of ion transporter activities is also cardinal to developmental CSF regulation by the ChP, raising the hypothesis that CSF dysregulation could be related to ChP inflammation [5]. This hypothesis was also brought to light in the studies of acquired hydrocephalus, a condition of excess CSF accumulation in the brain that can be caused by infection or brain hemorrhage [79,108], but more investigation is required to establish the connection. We refer the reader to several recent reviews on the proposed mechanisms participating in CSF-water secretion, movement, and removal [2,4,10,109]. Collectively, this growing body of research implicates the ChP and CSF as important immune mediators in neurodevelopmental disorders that involve neuroinflammation. The extent to which the ChP directly participates in neuroimmune crosstalk in the clinical context requires further investigation.

Concluding remarks

Recent technical advances in imaging and transcriptomics have begun to reveal the functional roles of ChP macrophages. Although currently there are more questions than answers (see Outstanding Questions), the emerging model highlights the ChP as coordinating communication between the nervous and immune systems. In this capacity the ChP is likely to be a key player in the etiology of neurodevelopmental and other brain disorders with inflammatory features such as ASD and schizophrenia. We expect further improvements in single-cell level -omics, in vivo imaging, and computational techniques will continue to expand molecular, spatial, and temporal resolution limits allowing interrogation of the heterogeneity and functional behaviors of immune cells, other ChP cell types, as well as cell-cell interactions in their native context. Then, tissue-specific genetic manipulation approaches could be leveraged to address questions regarding immune cell trafficking across the ChP, and into other CNS compartments (such as the CSF and brain parenchyma). Collectively, we envision future research will provide more insights on the role of ChP as a CNS immune barrier and consequently its potential as a therapeutic target for disrupted brain development, aging, and neuroimmune related diseases.

Outstanding questions:

• How does gene expression track with choroid plexus (ChP) immune cell identity and functional state?

• Do transient immune cell populations inhabit the brain parenchyma or CNS barriers during development?

• Are there age-specific roles for ChP immune cells?

• How do immune cells influence ChP epithelial, endothelial, and mesenchymal cell interactions?

• What are the ultimate fates of immune cells recruited to the ChP during neuroinflammation?

• Do different CNS compartments cooperate to keep CNS immunity in check?

HIGHLIGHTS.

The choroid plexus (ChP) provides a critical brain barrier that is also home to diverse and dynamic immune cell populations throughout life, as revealed by recent transcriptomic studies.

ChP macrophages represent the largest class of ChP immune cells and show heterogeneity.

Advances in in vivo imaging provide unobstructed views of the ChP immune cells in their physiological setting in the brain’s ventricles.

Environmental cues and tightly knit receptor-ligand networks at the ChP govern immune cell recruitment and infiltration from the periphery into the central nervous system (CNS).

The relevance of immune cell trafficking and dynamics at the ChP is increasingly appreciated in the context of neurological conditions involving neuroinflammation.

GLOSSARY

Blood-brain barrier (BBB)

a barrier formed by endothelial cells, astrocytic endfeet, and pericytes that restricts the passage of blood-borne factors into the brain.

Blood-cerebrospinal fluid barrier (BCSFB)

a barrier formed primarily by ChP epithelial cells that restricts the passage of blood-borne factors into the CSF.

Border-associated macrophages (BAMs)

Macrophages that reside in the CNS barrier tissues, including the meninges, the ChP, and the perivascular space. They are also referred to as “CNS-associated macrophages, or CAMs.

Central nervous system (CNS)

a principal part of the nervous system that consists of the brain and spinal cord.

Cerebrospinal fluid (CSF)

a body fluid that fills the brain ventricles and surrounds the brain and spinal cord. It contains ions and biomolecules including growth factors and neurotransmitters.

Choroid plexus (ChP)

a vascularized epithelial tissue located in each brain ventricle. It is a key source of CSF and forms the BCSFB.

Chemokines

a family of cytokines that induces chemotaxis in responsive cells.

Cytokines

secreted signaling proteins that regulate the functions of cells and/or interactions between cells.

Epiplexus macrophages (Kolmer cells)

cells that are located on the apical, CSF-contacting surface of the ChP.

Macrophages

mononuclear phagocytes found in all tissues and organs. They are scavengers of cellular debris and harmful substances.

Microglia

resident macrophages in the brain parenchyma with key roles in brain development and function.

Mobility

the total trajectory length that a cell body migrates.

Motility

the rate at which a cell moves its processes.

Stromal macrophages

macrophages located in the stromal space between the ChP epithelial and endothelial cell layers.