The choroid plexus (ChP) comprises sheets of secretory epithelial cells enclosing a vascularized stroma of mesenchymal and resident immune cells, all of which communicate extensively and conduct core ChP functions collaboratively (Figure 1). Although best known as the primary source of CSF and the principal blood–cerebrospinal fluid (CSF) barrier, the ChP’s immune environment and its role as a gateway for peripheral immune cell to enter the central nervous system (CNS) are also well recognized 1, 2, 3, 4. ChP inflammation and hypertrophy have been documented in an ever-growing list of diverse neurologic conditions, including multiple sclerosis (MS) [5], Alzheimer’s disease 6, 7•, amyotrophic lateral sclerosis [8], schizophrenia and psychosis [9], and COVID [10] (also reviewed in Refs. 11, 12). However, only in recent years has the ChP gained attention as an active neuroimmune modulator [13]. Given the critical role of the ChP to regulate the microenvironment of the brain, a better understanding of these changes is likely to inform disease etiologies as well as treatment and prevention strategies. In this review, we highlight key concepts and summarize current knowledge, focusing on recent insights gained through studies of human ChP samples and experimental animal models. We point out critical knowledge gaps and discuss future steps to overcome technical barriers to a fuller understanding of the interplay between ChP inflammation and brain pathophysiology.

Resident immune cells in the ChP are predominantly macrophages that tile the tissue in two key locations 4, 14. Stromal macrophages are so named according to their position in the space between epithelial cells and fenestrated endothelial cells, the stroma, whereas epiplexus (or Kolmer) cells are positioned on the apical, CSF-contacting side of the epithelial surface (Figure 1). The activation states and critical molecular signatures have been described for subsets of these cells 15, 16, and their core functions are beginning to be established. Like macrophages throughout the body, ChP macrophages are phagocytotic and play immunosurveillance roles. Epiplexus macrophages clean debris, including extravasated blood products, and are first responders to intraventricular hemorrhage. Stromal macrophages may additionally influence ChP barrier properties (see below) [17]. ChP macrophages also play key roles in defending against pathogen invasion because the ChP is a common site for various pathogens to invade the brain. One study of Trypanosoma brucei infection has highlighted the distinct role of brain resident and recruited macrophages in the response and restriction of parasitic infections [18]. The ChP is uniquely positioned to highlight these differences, as epiplexus cells, which are microglia like, expand during infection, whereas stromal macrophages are derived from recruited monocytes. Although all ChP macrophages upregulate inflammatory genes, the stromal cells have certain signatures (i.e. Nos2, Arg1, Cxcl14) indicating a difference in plasticity of recruited monocyte-derived macrophages. ChP macrophages persistently express inflammation-associated genes and in epiplexus cells, MHCII, long after infection resolution. In contrast, only a minority of microglia have increased MHCII and their inflammatory genes downregulate with the cells returning to a homeostatic state by 9 weeks, indicating a potential for increased trained immunity at the brain borders. Intriguingly, the importance of MHCIIhi macrophages for restricting pathogen brain invasion is highlighted by a study comparing neonatal and adult mice infected with lymphocytic choriomeningitis virus (LCMV). The study shows that neonatal border-associated macrophages (BAMs) in the dura, a primary site for viral invasion, have a limited number of MHCIIhi macrophages, enabling LCMV to infect the brain [19]. Understanding the role that MHCIIhi epiplexus cells have in preventing specific pathogens from infiltrating the brain is an important area for further investigation. Whether ChP immunosurveillance underlies differential vulnerability with other pathogens, in aging, or defends against chronic noninfectious conditions remains to be determined.

Immune cell infiltration into the CSF is a marker of neuroinflammation in various brain conditions, including infection and age-related neurodegeneration, and the ChP is one likely point of entry. Innate immune cells, mainly neutrophils and monocytes, accumulate at the ChP during brain inflammation of various causes (e.g. infection, brain injury, and autoimmunity). In Toxoplasma gondii infection [20], innate lymphoid cells (ILCs) and natural killer (NK) cells infiltrate the ChP early. While both cell types expand during acute infection, ILC1s are crucial for initiating interferon-gamma (IFN-γ) production, activating monocytes, and reducing the parasite burden — essential mechanisms for host survival. Additionally, NK cells isolated from postmortem ChP samples of patients with MS, an autoimmune condition, share high similarity in molecular signatures with those in the blood, consistent with the hypothesis that these cells infiltrate from blood into the ChP [21]. Recent imaging in an adult model of brain inflammation using lipopolysaccharides (LPS) showed for the first time that the neutrophils in the ChP enter the CSF, demonstrating that the ChP is indeed one gateway for peripheral immune cells to access the CNS [13].

Adaptive immune cells also use the ChP as a portal into the brain. T cells have long been implicated in MS and experimental neuropsychiatric lupus as drivers of brain lesions 22, 23. Postmortem ChP samples from patients with MS show increased numbers of T cells (CD3+) and B cells (CD20+) [24]. The extent of infiltration has been linked to the severity of periventricular lesions and inflammatory signals in the CSF, including presence of fibrinogen, a key driver of inflammatory neurodegeneration. Although ChP inflammation does not correlate with clinical severity in progressive MS, the ChP may nonetheless provide a significant immune infiltration path at early disease stages. Mouse models of autoimmunity may allow a better understanding of the disease at various stages. For example, in an experimental autoimmune encephalomyelitis (EAE) model of MS, T cells accumulated in the ChP prior to infiltrating the brain and spinal cord [25]. These observations support the hypothesis that the ChP acts as a reservoir and entry point for leukocytes into the CNS. Similarly, B cells from patients with MS migrate across monolayer human ChP epithelial culture [26], further implicating the ChP as an entry point of B cells as well. However, in the mouse EAE model, B cells were reported to accumulate only in the spinal cord but not observed in the ChP [25]. Future analyses using patient specimens obtained shortly after disease onset will help resolving discrepancies between humans and animal models.

Blood–CSF barrier permeability is likely to be regulated by tight junctions between ChP epithelial cells. But surprisingly, and despite consistent reports of inflammation-related ChP barrier disruption, roles of nonimmune cells in coordinating immune responses have been historically overlooked. Our recent work characterizing gene expression patterns in ChP epithelial cells following LPS administration has generated new interest in this topic. Termed inf-Epi for their responses to inflammation, this specialized population of ChP epithelial cells secrete chemokines to recruit leukocytes and loosen the epithelial barrier via extracellular matrix metalloproteases [13]. Later, inf-Epi cells express colony-stimulating factor 1 and adhesion molecules that support macrophage survival and adhesion. These findings show that interaction between epithelial cells and immune cells at the ChP is required to orchestrate the breaking down and restoration of the ChP barrier in coordination with leukocyte recruitment. They also show that there is no clear division between CSF secretory, barrier, and immune modulatory ChP functions but rather ChP cells collectively synergize to respond to exogenous challenges.

Other corroborating evidence for active roles of ChP epithelial cells in immune cell recruitment includes the finding that inactivation of adenosine A2A receptors in the ChP prevents T infiltration and ameliorates EAE, likely through inhibiting the upregulation of CCL20 in ChP epithelial cells [27]. ChP-specific IFN-γ receptor (IFN-gR1) knockout has similar protective effects [28]. In a mouse model of Alzheimer’s disease, increased expression of type-I IFN-stimulated genes in the ChP epithelial cells was associated with clonal expansion of T cells [29]. Intriguingly, recent work suggests that CCR6-CCL20 axis may promote accumulation of Th17 cells to the basolateral side of the ChP, rather than escorting them across the epithelial layer [25]. Further studies are needed to dissect the mechanisms underlying leukocyte recruitment versus migration to the ChP and the contributions of epithelial and other ChP cells to these processes in neuroinflammatory conditions. Other cell types in the ChP stroma, including fibroblasts and endothelial cells, also participate in signaling with immune cells under both homeostatic and inflammatory conditions (Figure 1, also 13•, 30). This inchoate area of research provides ample opportunities for future investigations to elucidate the functional relationship between ChP inflammation and brain barrier regulation.

Interest in the roles of all brain barriers including the ChP during normal aging and age-associated neurodegeneration such as Alzheimer’s disease has increased tremendously over the past few years (e.g. 31, 32, 33). Although barrier alterations are clearly necessary to allow immune cell entry into the brain when needed, increased ChP barrier permeability commonly associated with inflammation may also permit entry of harmful bloodborne molecules, making it important to understand the relevant cellular mechanisms. Most studies to date focus on modulation of epithelial cell tight junctions as the key mechanism by which the ChP adjusts its barrier permeability, but other ChP cell types likely contribute. For example, in a mouse model of inflammatory bowel disease, peripheral inflammation was suggested to modulate ChP barrier permeability by modulating fenestrae of ChP capillaries [34]. Another example comes from an LPS model of peripheral inflammation, processes of stromal macrophages elongate along ChP capillary walls, appearing to physically block access of blood contents to the stroma [17]. It is tempting to speculate that these responses to peripheral inflammation are protective. However, the molecular mechanisms and physiological effects are still poorly understood.

A more conventional understanding of ChP inflammation relates to rates of CSF secretion. Hydrocephalus, a disease of excessive and dysregulated CSF production, is associated with immune activation at the ChP [35], possibly because Toll-like receptor 4 activation causes both macrophage cytokine production and increased activities of epithelial ion transporters involved in CSF secretion [35,36]. Moreover, increased CSF accumulation and ventriculomegaly occurred in a rat posthemorrhagic hydrocephalus model when the chemokine CCL2 was overexpressed in the ChP epithelial cells [37].

Such modulation of CSF contents in response to inflammation, particularly during the early stages of brain development in a model of maternal immune activation, can have harmful, lifelong consequences on the brain [38]. Evidence that factors of ChP origin also penetrate the adult brain parenchyma and influence cellular activities is sparse but compelling. For example, Klotho, a longevity protein with CNS expression almost exclusively by ChP epithelial cells in the CNS, reduces microglial inflammation in the hippocampus [39]. ChP-derived extracellular vesicles containing various cargo may reach the hippocampus and other parts of the brain parenchyma 40, 41, 42, and ChP-secreted antioxidants can protect the hippocampus from learning and memory deficits [43]. Emerging evidence suggests that the tight correlation between ChP-CSF inflammation, brain aging, and dementia may be linked to CSF proteins and other factors 44, 45, 46. Proteomics data show an association of altered levels of ChP-derived proteins (e.g. Transthyretin and extracellular matrix proteins) with a subcategory of Alzheimer’s disease [7]. Transcriptomic analyses of tissue samples from patients with COVID-19 revealed upregulation of immune signaling in the ChP that are strongly correlated with receptor upregulation in the cerebral cortices, suggesting that the ChP may have long-range immune modulating functions throughout the brain [47].

Altogether, these tantalizing data warrant further study of the interplay between different ChP systems to understand the overall impact of ChP immune activities on the brain. It is likely that as in other parts of the body, ChP inflammation may be a double-edged sword, both adaptive and potentially harmful, depending upon timing and magnitude. Additional studies are needed to elucidate which aspects of ChP changes contribute to pathophysiology or represent compensatory responses or both.

Recent advances have positioned the ChP as a key neuroimmune organ in brain disorders. With the rapid increase in the knowledge of ChP inflammation in various types of brain conditions and the development of new imaging, genetic, and bioinformatic tools, the field is poised for significant breakthroughs.

Looking forward, the critical next steps are to connect molecular and cellular changes in the inflamed ChP to its altered functions and further discern their relationship to brain pathology as causal or compensatory. New research tools will enable the comprehensive characterization of ChP inflammation necessary to understand its role in disease onset and progression. Molecular labeling strategies are needed to track and quantify movements of specific molecules across the ChP. Improved cell manipulation approaches are needed to build causal relationships by conducting gain- and loss-of-functions studies. For example, recent genetics models have enabled selective elimination of ChP epithelial cells 48, 49. Advances in live imaging combining optogenetics and chemogenetics methods should provide new ways to examine dynamic ChP activities with high spatiotemporal precision [50]. Advancements in noninvasive physiological monitoring techniques for CSF dynamics and tissue inflammation will help unravel the sequence of events linking ChP inflammation to brain dysfunction. Newly reported in a porcine model of traumatic brain injury, intravenously administered macrophage-adhering gadolinium-loaded anisotropic micropatches allowed identification of infiltrated immune cells at the ChP by magnetic resonance imaging in live animals [51]. This new technology may be adaptable to rodent models and human patients to aid in the investigation of ChP inflammation and immune infiltration of the brain. Finally, disease models that more faithfully mimic human diseases are needed. Human pathology samples mainly capture later stages of diseases, making it challenging to discern causality. Widely used rodent models have their own limitations in replicating human disease and determining influences of aging (e.g. 5xFAD Alzheimer’s disease model [52]). Humanized models such as iPSC and organoids [53] and/or chimeric mice with transplanted human cells may surmount some of these barriers. Cross-disciplinary collaborations will be essential. Ultimately, these advances will elucidate the ChP–brain connection in inflammation and place the ChP within the broader context of neuroinflammation and disease for mechanistic investigations and development of novel therapies.