‘Approaches for studying neuroimmune interactions in Alzheimer’s disease

Abstract

Peripheral immune cells play an important role in the pathology of Alzheimer’s disease (AD), impacting processes such as amyloid and tau protein aggregation, glial activation, neuronal integrity, and cognitive decline. Here, we examine cutting-edge strategies—encompassing animal and cellular models—used to investigate the roles of peripheral immune cells in AD. Approaches such as antibody-mediated depletion, genetic ablation, and bone marrow chimeras in mouse models have been instrumental in uncovering T, B, and innate immune cell disease-modifying functions. However, challenges such as specificity, off-target effects, and differences between human and mouse immune systems underscore the need for more human-relevant models. Emerging multicellular models replicating critical aspects of human brain tissue and neuroimmune interactions increasingly offer fresh insights into the role of immune cells in AD pathogenesis. Refining these methodologies can deepen our understanding of immune cell contributions to AD and support the development of novel immune-related therapeutic interventions.

Peripheral immune cells in AD

The evolving landscape of Alzheimer’s disease (AD) research increasingly highlights the intricate interplay between the brain and the immune system. Emerging evidence suggests that AD is not solely characterized by pathological protein aggregation—such as amyloid-β (Aβ) plaques and tau tangles—but also involves immune dysfunction, with peripheral immune cells playing pivotal roles in disease progression [1,2]. This shift in understanding is driving a new wave of research and therapeutic exploration in AD, expanding the focus beyond the central nervous system (CNS) to include contributions from the peripheral immune system.

Peripheral immune cells—including T and B cells from the adaptive immune system, as well as innate immune cells such as neutrophils, monocytes, NK cells, and others—are active participants in the AD process [3–10]. These cells can infiltrate the brain, adopt tissue-specific phenotypes, or influence brain pathology by releasing soluble effector molecules [11]. Elevated numbers of T cells have been observed in the cerebrospinal fluid (CSF), leptomeninges, and hippocampus of AD patients, as well as in mouse models of Aβ and tau pathology [12–18], contributing to neuroinflammatory responses and exacerbating neurodegeneration [1,2]. Immune depletion of T cell populations in tauopathy mouse models can significantly reduce brain atrophy, neuronal loss, and behavioral impairments, suggesting a direct pathogenic role for these cells and their potential as therapeutic targets [13]. Therapies that either inhibit harmful immune cell activities or enhance protective responses are under active investigation. For example, modulating immune cell responses or blocking specific cytokine pathways has shown promise in preclinical models [19,20].

Animal models (i.e., rodents) have been indispensable in elucidating the roles of peripheral immune cells in AD, utilizing approaches such as antibody-mediated depletion, genetic knockouts, and bone marrow (BM) chimeras. Studies using antibody-mediated depletion have demonstrated that targeting specific immune cell populations such as neutrophils, can reduce Aβ deposition and mitigate tau pathology, highlighting the disease-modifying potential of peripheral immune cells [6,7]. Genetic knockout models, such as those lacking Rag enzymes necessary for T and B cell development, have shown that the absence of specific adaptive immune cells can influence Aβ pathology and neuroinflammation [21,22]. Similarly, BM chimeras can help dissect the contributions of peripheral immune cells in modifying AD pathology by reconstituting AD mouse models with immune cells from genetically modified murine donors [23]. However, the limitations of animal models, particularly the differences between mouse and human immune systems, emphasize the need to establish or design more human-relevant models. This has spurred the development of innovative platforms such as brain organoids, three-dimensional (3D) cell culture systems, organ-on-chips, and other chimeric models, which are beginning to elucidate the molecular and cellular contributions of peripheral immune cells to AD neuropathology [14,24–29]. These emerging systems have the potential to bridge the gap between animal and human pathophysiology, offering deeper insights into the mechanisms underlying the disease.

This review examines the strategies and approaches used to study peripheral immune cells in AD, encompassing the successes and limitations of animal and human multicellular models. We discuss how these insights are guiding the development of immune-targeted therapies and consider future directions for harnessing these methodologies to better understand the roles of peripheral immune cells in AD. These advances have the potential to uncover new therapeutic strategies that leverage the immune system to combat this devastating disease.

Strategies for studying peripheral immune cells in AD mouse models

Emerging research underscores the pivotal role of peripheral immune cells in modulating the pathology and cognitive decline observed in AD mouse models [1,2]. Experimental depletion or disruption of these cells in amyloid and tau mouse models demonstrates profound impacts on pathological protein aggregation, glial activation, neuronal health, synaptic integrity, brain cytokine concentrations, as well as memory functions (Table 1). This body of evidence positions peripheral immune cells as key modulators of AD progression and putative therapeutic targets. Here, we discuss the approaches used to study the roles of peripheral immune cells in AD animal models (Figure 1), highlighting their strengths, limitations, and broader implications for AD.

Key table, Table 1.

Mouse studies investigating the role of peripheral immune cells in AD.

Mouse model	Age	Sex	Approach	Findings	Ref
3xTg-AD, 5xFAD	6 mo (3xTg AD) 4 mo (5xFAD)	♂ and ♀	Neutrophil depletion using anti-Ly6G (clone 1A8) or anti-Gr-1 (clone RB6–8C5) antibodies	Reduced Aβ and p-tau; rescued memory deficits	[5]
3xTg-AD	7–8 mo	♀	NK cell depletion using anti-NK 1.1 (clone PK136) antibody	Reduced inflammation; enhanced cognitive function	[6]
3xTg-AD, 5xFAD	60–70-wk	♀	B cell depletion using anti-CD20 (clone 5D2) and anti-B220 (clone TIB-146) antibodies	Reduced Aβ; increased numbers of TGFβ+ and IL10+ microglia	[32]
THY-Tau22	3–12 mo	♂ and ♀	T cell depletion using anti-CD3 (clone 145–2C11) antibody	Increased cognitive function; reduced proinflammatory gene expression in microglia; no changes in tau deposition	[2]
APOE4 tauopathy	6–9.5 mo	♂	T cell depletion using anti-CD4 (BP0003–1) and anti-CD8 (BP0061) antibodies	Reduced p-tau; enhanced memory performance	[33]
5xFAD	6 mo	♂	CD8+ T cell depletion using anti-CD8 (BP0061) antibody	Reduced myelin damage; enhanced memory and spatial learning	[14]
APP-PS1	12 mo	♂ and ♀	CD8+ T cell depletion using anti-CD8 antibody	No effect on Aβ burden or cognition	[34]
APP-PS1	6–12 mo	not specified	CD8+ T cell depletion using anti-CD8 antibody	Increased Aβ plaque size	[35]
Rag1 KO	3, 12, 18, 24 mo	not specified	Lack of functional lymphocytes (B and T cells)	Prevented the decline in white matter oligodendrocyte density	[21]
Rag2 KO X PSAPP	3–8 mo	♂ and ♀	Lack of functional lymphocytes (B and T cells)	Reduced Aβ pathology and myelin abnormalities	[23]
Rag1 KO X 5xFAD	4, 6, 10, 12 mo	not specified	Lack of functional lymphocytes (B and T cells)	Reduced Aβ burden	[14]
Rag2/II2rg KO X 5xFAD	6 mo	not specified	Lack of functional lymphocytes (B, T, and NK cells)	Increased Aβ	[22]
JHT mice X 3xTg-AD	50–60 wk	not specified	Lack of B cells	Reduced Aβ burden; improved cognitive performance	[32]
μMT mice X 5xFAD	3, 5, 8 mo	not specified	Lack of B cells	Aggravated AD-like pathology	[3]
β2m KO X 5xFAD	4 and 10 mo	♀	Lack of CD8 T cells	Increased Aβ burden	[38]
MHC II KO X 5xFAD	1,3, 6 mo	not specified	Lack of CD4 T cells	Worsened Aβ pathology	[39]
MR1 KO X 5xFAD	2, 4, 6, 8 mo	♂ and ♀	Lack of MAIT cells	Reduced Aβ burden	[9]
Wsh X APP-PS1	not specified	♂ and ♀	Lack of mast cells	Restored synaptophysin protein	[41]
Cpa3Cre/+ X 5xFAD	12–13 mo	♀	Lack of mast cells	Restored contextual fear; increased gene expression of disease-related microglia	[8]
APOE3NTKO and APOE4NTKO	12 mo	♀	Blockade of IL-17F pathways	Reduced neutrophil infiltration	[49]
Cxcr6 KO X 5xFAD	8–10 mo	♂ and ♀	Lack of CXCR6 chemokine receptor	Reduced CD8+ T cell infiltration; increased Aβ plaque and worsened cognitive function	[38]
JNPL3 tauopathy	10–12 mo	♀	PD-1 Blockade with anti-PD-1 antibody (clone RPM1–14)	Reduced Aβ plaques and improved cognitive impairment	[19]
Young and aged	Young: 2-3 mo Aged: 19-24 mo	♀	IL-33 or IL-5 administration	Reduced age-related cognitive decline	[7]
μMT X 5xFAD	4 mo	♂	IL-35 administration	Reduced Aβ burden and cognitive dysfunction	[3]
APP-PS1	not specified	♂ and ♀	Masitinib administration	Enhanced cognitive performance	[41]

KO: knockout; NTKO: neutrophil-specific deletion of APOE variants; APP-PS1: APP Swedish PS1 ΔE9; mo: months-old; wk: weeks-old; Ref: references

Figure 1. Strategies for studying peripheral immune cells in AD mouse models.

The schematics outline various experimental approaches used to study the role of peripheral immune cells in AD mouse models. These strategies encompass antibody-mediated depletion of immune cells, the use of genetically modified knockout and mutant mice, the creation of bone marrow chimeric mice, inhibition of peripheral immune cell infiltration, and modulation of peripheral immune cell effector functions, either through ablation or enhancement. Further descriptions of these studies can be found in Table 1. ↓, decrease; ↑, increase

Antibody-mediated immune cell depletion

One of the most direct approaches to dissecting the role of specific immune cells in AD is through antibody-mediated depletion. This strategy enables the targeted removal of particular cell types without the need for genetic manipulation, providing a flexible and clinically relevant approach. For example, the depletion of neutrophils using anti-Ly6G and anti-Gr-1 antibodies reduces amyloid and phospho-tau (p-tau) amounts and subsequently improves memory performance in fear conditioning and Y-maze tests in the 3xTg AD mouse model [6]. Similarly, the administration of anti-NK1.1 antibodies to deplete NK cells has been reported to enhance cognitive function in the water maze test and reduce the expression of inflammatory genes, as shown via RNA sequencing, in 3xTg AD mice [7]. Depleting B cells with anti-CD20 and anti-B220 antibodies not only reduces Aβ burden but also increases the presence of hippocampal microglia expressing immunoregulatory cytokines such as transforming growth factor beta (TGFβ) and interleukin 10 (IL10) in 5xFAD mice, suggesting that B cells may contribute to exacerbating AD pathology [30]. However, a separate study using B cell knockout mice reported the opposite findings (see section on ‘Genetically knockout and mutant mice‘).

The role of T cells in AD is particularly complex, as evidenced by mixed results across different studies. For example, in the THY-Tau22 model, treatment with anti-CD3 antibodies—targeting both CD4⁺ and CD8⁺ T cells—improved cognitive function in the Y-maze test, with reduced microglial proinflammatory gene expression via RT-PCR (Clec7a, Itgax, Ccl3), without altering tau deposition [3]. However, in an APOE4 tauopathy model, the co-injection of anti-CD4 and anti-CD8 antibodies significantly increased microglial Iba1⁺ and CD11c⁺ areas, reduced p-tau amounts, restored hippocampal and cortical brain volumes, and improved memory performance in Y-maze and fear conditioning tests [31]. However, contradictory results have been observed in CD8⁺ T cell ablation studies, with some showing no effect on Aβ burden or cognition [32] and others reporting modest increases in Aβ plaque sizes [33]. Two recent studies suggested that CD8⁺ T cells play a detrimental role in AD models. Specifically, one study reported that depleting CD8⁺ T cells resulted in less severe myelin damage and improvements in spatial learning and memory in the Barnes maze in 5xFAD mice relative to control 5xFAD mice bearing a physiological number of CD8⁺ T cells [15]. The other study showed that CD8⁺ T cell depletion reduced hippocampal Aβ plaques and microglia numbers and restored the spontaneous alternation rate in 5xFAD mice [34]. Although most current reports suggest a detrimental role for CD8⁺ T cells in AD, it is likely that CD8⁺ T cell subsets are heterogeneous, with some exhibiting protective effects.

Despite the utility of antibody-mediated cell depletion, this method presents several technical challenges. The efficacy of cell depletion can vary depending on factors such as antibody dosage, treatment duration, administration routes, and the stage of disease at which depletion is initiated. Incomplete depletion may yield misleading results, and the binding of antibodies to cell surface markers could inadvertently activate target cells or induce an adaptive immune response [35,36]. Therefore, rigorous optimization and validation of treatment protocols are essential to ensure the reliability of experimental outcomes.

Genetic knockout and transgenic mutant mice

For more complete and sustained elimination of specific peripheral immune cell types, breeding genetically engineered knockout (KO) or transgenic mutant mice with AD models provides a powerful tool. Rag1^−/− or Rag2^−/− KO mice (which lack T and B cells due to the absence of Rag enzymes that are essential for their development), are commonly used to investigate the role of adaptive immune cells in AD [37]. In aged mice, Rag1 deficiency prevented the decline in brain white matter oligodendrocyte density observed in wild-type mice, suggesting a protective effect against age-related white matter loss by eliminating T and B cells [21]. Rag2 KO PSAPP mice exhibited reduced Aβ pathology as assessed by thioflavin-S and meso scale discovery (MSD) biochemistry assays of brain homogenates [38]. Additionally, electron microscopy analysis of Rag1 KO APP^NLGF mice demonstrated ameliorated myelin abnormalities relative to controls [39]. Together, these findings support the hypothesis that adaptive immune cells can contribute to AD pathology [15,23]. Of note, the combined KO of Rag2 and the common γ chain (Rag2^−/− Il2rγ^−/−), which also eliminates NK cells, results in an increased deposition of Aβ in the 5xFAD mouse model, suggesting that different immune cell types may have distinct and perhaps sometimes opposing roles during AD pathogenesis and/or progression [22].

The role of B cells in AD is still under debate. While antibody-mediated depletion of B cells suggests a detrimental role, genetic KO approaches have produced conflicting results. In the 3xTg and APP/PS1 models, B cell deficiency via immunoglobulin J_H locus deletion (J_HT-3xTg and J_HT-APP/PS1) reduced Aβ burden, which was assessed from histology and ELISA assays and improved cognitive performance in water maze tests [30]. However, in μMT 5xFAD mice, B cell deficiency led to increased Aβ burden (thioflavin-S and 6E10), glial activation (GFAP, Iba1, and CD68 immunofluorescence staining), and worsened cognition (novel object recognition test [NOR], Y maze, and Barnes maze tests) [4]. Relative to controls, a recent study showed via flow cytometry, reduced CD8⁺ T cell infiltration in the brains of μMT 5xFAD mice with fewer CD8⁺ T cell-producing proinflammatory cytokines such as TNF and IFNγ [34]. However, here, the authors did not report on Aβ pathology or neurodegeneration phenotypes.

T cells, which are known to accumulate in the brains of both AD patients and mouse models, continue to present a complex and unresolved puzzle regarding their contribution to pathology. Crossing 5xFAD mice with TCR alpha chain KO mice (TCRα^−/− 5xFAD; lacking CD4⁺ and CD8⁺ T cells), resulted in increased Aβ plaque number and burden compared to controls, suggesting that T cells might play a protective role in AD pathology [40]. Therefore, further studies using CD4 or CD8 KO AD models are needed to disentangle the distinct contributions of these T cell subsets. In one study, CD8⁺ T cell deficiency reduced the number of interferon (IFN)-responsive oligodendrocytes and microglia in 24-month-old mice compared to aged-matched controls, indicating that CD8⁺ T cells can act directly on these cells via IFNγ production in the aging brain [21]. To our knowledge, specific CD4 or CD8 KO models in AD have not yet been employed. Studies using 5xFAD mice that were deficient in major histocompatibility complex (MHC) molecules, and lacking CD4⁺ T cells, showed worsened Aβ pathology compared to 5xFAD mice with a normal number of mature CD4⁺ T cells, suggesting a protective role for CD4⁺ T cells against amyloidosis [41]. Similarly, β2-Microglobulin KO (B2m^−/−) mice, which lack MHC class I molecules and CD8⁺ T cells, showed increased Aβ accumulation by immunofluorescence and impaired memory performance in the Y-maze and NOR tests relative to 5xFAD controls, again suggesting that CD8⁺ T cells may have a protective function against AD [40]. However, β2m itself has been suggested to influence amyloid pathology independently of CD8⁺ T cells, because it can co-aggregate with Aβ in 5xFAD mouse brains and in vitro assays, potentially exacerbating disease progression [42].

Innate T cells, such as mucosal-associated invariant T (MAIT) cells, have also been implicated in AD pathogenesis. MAIT cells, which depend on the MHC class I-related molecule MR1 for their development, increased in number in 5xFAD mouse brain tissues relative to wild-type [10]. In MR1 KO 5xFAD mice (Mr1^−/−/5xFAD), which lack MAIT cells, reduced plaque burden in mice after 6 months of age was reported, pointing to a possible detrimental role for these cells in AD [10].

Mast cells, known for their proximity to the brain and rapid release of proinflammatory mediators, have also been associated with AD pathology. In APP mice with Kit mutation-mediated mast cell deficiency (APPswe/PSEN1dE9 Kit^W-sh), synaptic protein integrity was compromised (shown via immunochemistry), suggesting that mast cells might contribute to synaptic dysfunction in AD [43]. Additionally, in the specific mast cell-deficient “CreMaster” strain crossed with 5xFAD mice (Cpa3^Cre/+ 5xFAD), mast cell deficiency not only restored contextual fear conditioning but also enhanced the single-cell RNA-sequencing (scRNA-seq) gene profile associated with disease-associated microglia such as Apoe, B2m, Cd68, and Csf1 [9].

While gene knockout and mutant mouse models offer the advantage of completely eliminating specific peripheral immune cell types, they also come with inherent challenges and caveats. One key issue is the potential for unintended side effects. For example, knocking out CD4⁺ T cells also disrupts B cell antibody class switching, thus leading to broader immune dysregulation [44]. The specificity of these models can be problematic – e.g., mast cell-deficient mice created through Kit mutations also affect stem cells, complicating the interpretation of results. Moreover, the process of breeding genetically modified mice with AD models and waiting for pathology to develop is time-consuming and costly. Therefore, while these models are invaluable, it is important to carefully consider their limitations when designing and interpreting studies.

Generating bone marrow chimeric mice

In addition to breeding AD mouse models with immunodeficient strains, cell depletion can also be achieved by generating BM chimeric mice. This method involves reconstituting AD mice with BM cells harvested from immune cell-deficient donors following irradiation to eliminate the host’s hematopoietic cells. For example, reconstituting PSAPP mice with Rag2 KO (Rag2^−/−) BM resulted in a reduced Aβ load in the brains of recipient mice. Similar findings were observed when Rag2 KO was directly bred with PSAPP mice [23]. However, caution is needed when interpreting results from BM chimera experiments. The irradiation process used to eliminate hematopoietic cells can increase blood-brain barrier (BBB) permeability, potentially leading to an influx of peripheral immune cells and complicating the analysis of their effects on AD pathology [45]. Additionally, the antibiotics administered post-transplantation to prevent infections can disrupt the gut-brain axis [46,47], which can potentially further confound results [46,47]. To mitigate these challenges, alternative strategies for BM depletion, such as using anti-CD45 or anti-CD117 antibodies conjugated with the ribosome-inactivating protein saporin, have been used where no irradiation or antibiotics are required [48,49].

Inhibition of peripheral immune cell infiltration

Peripheral immune cells, although not native to brain tissue, infiltrate and accumulate in the brain in response to various environmental signals, contributing to AD pathology. Rather than depleting entire cell populations, targeting the infiltration of these cells into the CNS offers a strategic approach to studying their roles in AD mouse models. For example, targeting lymphocyte function-associated antigen 1 (LFA-1), a key integrin for leukocyte migration and adhesion (e.g., neutrophils, among others), was shown to reduce Aβ burden and improve contextual fear conditioning in 3xTg AD mice, replicating the effects of neutrophil depletion in mice [6]. Additionally, a recent study identified IL-17F as a promoter of neutrophil infiltration into the brain, with antibody-mediated blockade of this pathway effectively reducing neutrophil presence, restoring neurodegenerative microglia responses, and improving memory performance in the Y maze. These results have placed a focus on IL-17F as a putative therapeutic target meriting further investigation [20].

In the context of AD, CD8⁺ T cells in the brain express receptors such as CXCR6 and CXCR3, which respectively bind to CXCL16 and CXCL10, predominantly secreted by glial cells [14,40,50]. These interactions (CXCL16-CXCR6 and CXCL10-CXCR3) facilitate the entry of CD8⁺ T cells into the brain parenchyma in AD mouse models. In 5xFAD mice, the absence of CXCR6 significantly reduced CD8⁺ T cell infiltration into the brain yet resulted in increased Aβ plaque burden and exacerbated cognitive decline, as evidenced by immunofluorescence, and Y maze and NOR tests [40]. This indicated that CXCR6-expressing CD8⁺ T cells could have beneficial effects on AD pathology. This notion of CD8⁺ T cells playing a protective role aligns with observations in the same report of TCRα-deficient (TCRα^−/−) and β2m-deficient (B2m^−/−) 5xFAD mice. These findings underscore the complexity of T-cell involvement in AD and warrant further studies.

Given that the roles of lymphocytes in AD are still being unraveled, insights can be drawn from other neuroinflammatory conditions, such as multiple sclerosis (MS). Fingolimod (FTY720) is an FDA-approved drug that delays MS progression and has shown promising effects in the experimental autoimmune encephalomyelitis (EAE) mouse model of MS [51]. It downregulates S1P receptor 1 expression on lymphocytes (a molecule that regulates lymphocyte egress from lymphoid tissues), thereby reducing their infiltration into the CNS [52]. Considering emerging evidence of T and B cell infiltration in AD, investigating the effects of Fingolimod in AD pathologies may offer significant translational potential.

Since many cell migration and adhesion molecules, as well as chemokine-receptor interactions, are shared across multiple immune cell types, targeting these pathways may lead to off-target effects. To better understand the roles of specific signaling axes in AD pathology, it is important to adopt a multi-faceted approach. Additionally, it is crucial to better understand how each signaling pathway influences distinct immune cell populations and, therefore, their possible contributions to disease pathology.

Ablation or promotion of peripheral immune cell effector functions

Immune cells execute their effector functions through direct cell-cell interactions and the secretion of various signaling mediators. Instead of merely removing entire immune cell subsets or inhibiting their infiltration, modulating their effector functions might provide deeper insights into their precise roles in AD. Each immune cell type can exert both protective and harmful effects, making it necessary to carefully dissect their specific contributions to disease pathology.

Many peripheral immune cells exert their effects by orchestrating the release of cytokines. As mentioned, neutrophil-derived IL-17F has been implicated in impairing microglial responses to neurodegeneration [53]. Blocking IL-17F with specific antibodies restored disease-associated microglial phenotypes and improved cognitive performance in the Y maze in APP/PS1 mice, highlighting the detrimental role of this cytokine [53]. Conversely, administration of certain cytokines can have protective effects. IL-33, for example, activates type 2 innate lymphoid cells (ILC2), leading to the release of IL-5, which alleviates age-associated cognitive decline in object placement and water maze tests in mice [8]. Similarly, B cells in 5xFAD mice upregulate the protective cytokine IL-35. Administering IL-35 to μMT 5xFAD mice reduced Aβ burden, as evidenced from histology and Western blotting; it also mitigated cognitive dysfunction in Y and Barens maze texts while neutralizing IL-35 exacerbated AD pathology [4].

T cell-derived IFNγ has been proposed to play a dual role in AD. It may recruit myeloid cells to the brain, as seen in 5xFAD mice, in which neutralizing IFNγ diminished protective myeloid cell recruitment and Ccl2 mRNA expression mediated by PD-1 immune checkpoint blockade [19]. Yet, IFN-γ from CD8⁺ T cells (assessed via) scRNA-seq was identified as a factor that induced IFN-responsive microglia, leading to myelin and oligodendrocyte loss in aged mice, as shown via histology [21]. The exact effects of T cell-derived IFNγ in AD models remain uncertain, and the generation of CD4 or CD8 Cre-mediated IFNγ conditional KO mice in an AD background could provide crucial insights.

The role of mast cells, known for releasing a variety of pre-formed and de novo-synthesized proinflammatory mediators such as histamine and chymase upon activation, has also been investigated in AD. Although the specific mediators involved remain unidentified, drugs such as masitinib, imatinib, and cromolyn, which stabilize mast cells by targeting Kit signaling (masitinib and imatinib) or via unknown mechanisms (cromolyn), have shown efficacy in restoring synaptic integrity (histology) and improving cognitive performance in the water maze test in APPswe/PSEN1dE9 mice [9,43].

Another promising approach involves targeting glucose metabolism by blocking the prostaglandin E2 (PGE₂) receptor in peripheral myeloid cells. This strategy reduced plasma and hippocampal inflammatory cytokine expression in cytokine assays and restored cognitive function in object placement and the Barnes maze in aged mice [54]. Exploring the effects of PGE₂ receptor blockade in AD models might yield further therapeutic opportunities.

Of note, ablating or promoting the effector functions of peripheral immune cells poses significant challenges in achieving precise specificity. Immune cells often share common signaling pathways, surface markers, and cytokine profiles, making it difficult to selectively target one subset without impacting others. For example, IL-17F and IL-5 are produced by multiple cell types, including neutrophils, ILC2s, and CD4⁺ T cells. Similarly, masitinib, a tyrosine kinase inhibitor, affects both mast cells and microglia [55,56]. This overlap can complicate the interpretation of experimental results. Therefore, strategies aimed at modulating immune functions must carefully consider the potential for broad, non-specific effects, underscoring the need for precise targeting to avoid unintended consequences. Additionally, it is crucial to determine the optimal timing for targeting immune effector molecules, because their roles may differ across disease stages. Finally, when modulating immune functions, the ‘Goldilocks’ principle should be applied, because both underactivity or overactivity could result in detrimental effects.

Strategies for studying immune cells in human cellular models of AD

The differences between mice and humans highlight the limitations of animal models and the need for developing human-relevant platforms. This has led to the design of innovative systems such as brain organoids, 3D cell cultures, organ-on-chips, and chimeric models [1,18] (Figure 2). Although research directly examining the interplay between peripheral immune cells and brain cells in AD using human cellular models is still in its nascent stages [1], substantial insights can be drawn from other fields, including cancer, MS, and Parkinson’s disease (PD). Integrating these perspectives can accelerate the development of innovative cellular systems to enable a better understanding of neuroimmune interactions and the role of peripheral immune cells in AD. These cross-disciplinary approaches also hold promise for developing therapeutic strategies that target the peripheral immune system to mitigate AD.

Figure 2. Strategies for studying peripheral immune cells in human cellular systems.

Schematic overview of current in vitro models for investigating the interactions between peripheral immune cells and brain cells in AD or other neurodegenerative pathologies. These approaches (used in various biomedical fields) include, among others, monolayer culture systems [58], 3D cell cultures [74,76], organoids and assembloids [86], as well as organ-on-chip platforms [14,90].

Traditional in vitro co-culture systems, including monolayer cultures and transwell assays, have long been utilized to study cellular interactions by allowing direct or indirect contact between different cell types. These methods (which involve conditioned media) also enable the exploration of soluble factor-mediated communication. While useful, these approaches predominantly offer bulk-averaged data, which can obscure the cellular heterogeneity and dynamic nuances that constitute the full spectrum of immune-brain interactions.

Advancements in single-cell cell culture technologies have begun to address these limitations by facilitating high-resolution analysis of intercellular communication. For example, a microwell-based platform was developed that precisely co-cultures individual NK cells with K562 leukemia target cells [57]. This system allows real-time monitoring and quantitative assessments of single-cell interactions, including immunological synapse formation, target cell lysis, and cytokine secretion profiles [57]. Of note, the granular data have shown significant variability in individual NK cell cytotoxic responses and secretory behaviors, underscoring the functional heterogeneity within seemingly homogeneous cell populations. Similarly, a microdroplet-based co-culture system was designed to study the interactions between DCs and T cells at single-cell resolution [58]. This approach encapsulates individual DC-T cell pairs within microdroplets, permitting controlled and isolated examination of their interactions over time [58]. Similarly, a high degree of variability in T-cell activation and proliferation responses, driven by differences in antigen presentation and costimulatory signals provided by DCs, were reported. These findings highlight the importance of using single-cell analyses to capture the diverse and dynamic nature of immune cell interactions that are critical for effective immune responses. These platforms can further inform on interactions between immune cells and brain-resident cells, including neurons and glial cells, which could be relevant in AD pathology.

Emerging high-throughput sequencing technologies, including scRNA-seq and single-cell assay for transposase-accessible chromatin sequencing (scATAC-seq), are revolutionizing our ability to dissect the complex interplay between immune and brain cells at an unprecedented resolution [16,59]. Information on the molecular diversity and functional states of cells in various neuroinflammatory conditions, including AD, can be obtained [60,61]. For example, multiplex proteomics has identified novel CSF and plasma biomarkers associated with early AD, particularly proteins involved in innate and adaptive immunity, such as matrix metalloproteinases 9 (MMP-9) and MMP-10 in CSF, and elevated IL-2 concentrations in plasma from AD patients [62]. Additionally, scATAC-seq has uncovered a cis-regulatory element that is co-accessible with the CXCR3 gene promoter in CD8⁺ T cells from PBMCs of AD patients, implicating these genes in T cell-mediated neuroinflammation, and which may potentially exacerbate neuronal damage through inflammatory cascades [12]. Additionally, spatial transcriptomics is a powerful tool for mapping gene expression within the spatial context of brain tissue, allowing the visualization of cellular interactions and their localization within the brain’s anatomical architecture [63–66]. In 7-month-old 5xFAD mice, monocyte numbers were increased in the deep cortex and white matter, while plasmacytoid DCs were present in regions such as the thalamus, pyriform area, and striatum [67]. Moreover, combining these techniques can enable a more comprehensive view of how immune cells and brain-resident cells communicate and contribute to disease progression. Similarly, the CellChat platform [68] offers a computational framework for analyzing and visualizing cell-cell communication networks by integrating scRNA-seq data. It has uncovered dysregulated receptor-ligand interactions in various disease models [68]. In an animal study of traumatic spinal cord injury (SCI), CellChat analysis showed that microglia contributed to tissue homeostasis, indicating communication networks between microglia and astrocytes, and offering certain insights into the molecular mechanisms behind inflammation and tissue repair [69]. Such computational tools could similarly help detail cell-type-specific molecular interactions in AD.

While 2D (or organotypic) models have yielded important insights, they fall short of fully replicating the intricate 3D architecture and interactions of brain cells that occur naturally. To address this, studies have been bolstered by the development of 3D cellular models that better replicate the architecture of the brain [14,24,26–29,70–73]. One recent preliminary report (in preprint) of a 3D immuno-glial-neurovascular brain model showed the generation of patient-specific co-cultures of six major brain cell types. If validated, this brain model might provide unprecedented insights into the complex interplay between brain cells in diseased brain [74]. However, studying the interactions between peripheral immune cells and brain cells within 3D cell culture systems is still in its infancy, especially when compared to advancements in other fields. A recent study investigated the interactions between the immune system and tumor cells by co-culturing cancer spheroids with activated peripheral immune cells [75]. An algorithm was applied to quantitatively assess immune cell infiltration and spheroid growth, offering insights into immune evasion mechanisms in the context of breast tumors. The findings demonstrated that increased CD3⁺ T-cell infiltration correlated with reduced tumor spheroid volume, indicating effective immune-mediated tumor suppression [75]. Another study demonstrated that the integration of macrophages into breast cancer spheroids significantly enhanced tumor growth, as measured by light sheet microscopy [76]. The findings highlighted the ambivalent role of macrophages in promoting tumor growth while also modulating immune responses within the microenvironment [76]. The incorporation of peripheral immune cells into 3D cellular models is crucial for gaining a better understanding of neuroimmune interactions and the mechanisms by which peripheral immune cells, such as T cells and monocytes, infiltrate into the brain and contribute to AD pathogenesis. Early findings suggested that blocking the PD-1/PD-L1 immune checkpoint pathway increased immunomodulatory monocyte-derived macrophages within the brain and led to cognitive improvement in a tauopathy model, offering potential immunotherapy avenues for AD [77]. Therefore, developing new 3D human multicellular models that integrate peripheral immune cells with neural/glial cells can test interventions that modulate immune cell activation and ideally prevent neurodegenerative processes in AD or other related dementia.

In addition, the rapid advancement of organoid technology offers a powerful platform for modeling brain development and disease progression and for studying human-specific genetic variants across diverse disease states (reviewed in refs [78–81]). The next crucial step is integrating immune components into brain organoids to create a more comprehensive model of neuroimmune interactions. For example, the inclusion of CD4⁺ T cells in mouse and human brain organoids can exacerbate brain injury, again suggesting a detrimental role of infiltrating T cells [82]. Also, the knockdown of IFNγ released by CD4⁺ T cells in necrotizing enterocolitis (NEC) mice significantly alleviated neuroinflammation and improved neurological outcomes, indicating an important function for this cytokine in modulating brain damage [82]. However, co-culture models often rely on a single immune cell population, which limits their utility to fully capture the complexity of cell interactions within the CNS. Lessons from other organoid systems, such as intestinal models, should also be applied to brain organoids. For example, integrating epithelial organoids with autologous tissue-resident memory T (T_RM) cells showed dynamic heterogeneity among T_RM populations through scRNA-seq analysis [83]. This study identified activated CD8⁺ T cells and an increased population of T helper-1-like CD4⁺ T cells in epithelial organoids, underscoring the complexity and plasticity of immune responses within tissue-specific environments [83]. In addition to organoids, assembloids (an extension of organoid systems) merge multiple brain region-specific organoids to more accurately mimic the complex structures and functions of the human brain [78–81]. Examples include human 3D cortico-motor assembloids [84] and brain assembloids designed to interrogate human neural circuits [85]. Additionally, vascularized brain assembloids have been developed by fusing brain organoids with endothelial cell organoids [86]. However, the integration of peripheral immune cells into these assembloids remains underexplored.

Organ-on-chip technology marks a major leap in human pathophysiology modeling, enabling precise control of microenvironments and the study of complex cellular interactions [87–89]. A recent study used a microfluidic system to observe how glioblastoma-like tumors (GL261 and CT-2A) influenced immune-vascular interactions [90]. The tumors steered macrophages toward an ‘M2-like’ immunosuppressive phenotype, helping to create a proangiogenic (blood vessel-forming) environment. Research also showed that TGF-β1 and endothelial cell – macrophage interactions via surface integrin (αvβ3), were essential for inflammation-driven angiogenesis [91]. We previously developed a 3D human tri-culture system with neurons, astrocytes, and microglia that mimics AD features, including Aβ aggregation, p-tau formation, and neuroinflammation [27]. In this model, microglia were activated (shown via increased CD11b) and recruited by AD neurons and astrocytes. This activation contributed to neuronal damage through pathways involving IFNγ and Toll-like receptor 4 (TLR4) [27]. Expanding this, we recently developed a 3D human neuroimmune axis model, called the PiChip system, which incorporates neurons and astrocytes harboring familial AD mutations, induced pluripotent stem cell (iPSC)-derived microglia (iMGL), and peripheral immune cells such as CD8⁺ T cells and monocytes [14]. Our findings showed that increased CD8⁺ T-cell infiltration worsened AD pathology, and the astrocytic-derived CXCL10 chemokine (quantified via MSD and immunostaining) played a key role in recruiting CD8⁺ T cells into this AD cellular system. Blocking the CXCL10-CXCR3 signaling pathway with an anti-CXCR3-neutralizing antibody (MAB160) reduced CD8⁺ T-cell infiltration and prevented neuronal death, highlighting the potential of targeting this pathway for therapeutic development [14]. These studies underscore the versatility of organ-on-chip platforms to assess neuroimmune interactions.

Overall, organ-on-chip models offer a unique platform for integrating various cell types and creating controlled microenvironments, making them useful tools for studying complex neuroimmune interactions. While animal models may still be necessary for validating the clinical relevance of these findings, the precision and versatility of organ-on-chip technology can significantly advance our understanding of molecular mechanisms underlying neuroimmune interactions and AD progression. Additionally, these human cellular platforms hold great potential for improving the screening of immunotherapies aimed at mitigating neurodegeneration in AD.

Concluding remarks

Peripheral immune cells play a crucial role in the pathogenesis of AD. The interplay between immune cells and brain resident cells is complex and can have both protective and harmful effects on neurons, depending on the stage of the disease and the immune cell type. Understanding these dynamics is pivotal for developing targeted therapies that can modulate immune responses and slow or halt AD progression (see Outstanding Questions).

Outstanding questions.

Conflicting findings regarding the roles of B and T cells in AD mouse models raise a key question: Do these peripheral lymphocytes exacerbate or mitigate AD pathology, and how might this vary with age or across different animal models?

Given the variability in models, disease stages, and ages used across different studies, how can the findings from AD mouse studies be standardized?

What mechanisms govern the migration and infiltration of peripheral immune cells into the AD brain, and how do these processes influence disease progression?

How do specific peripheral immune cell subtypes, such as T cell populations, contribute to neurodegeneration, and what are their roles during different stages of AD development?

Once peripheral immune cells infiltrate the brain, what are the molecular mechanisms by which they interact with brain-resident cells, such as microglia or astrocytes?

What antigen-specific receptors are expressed by infiltrating T cells in AD, and what are the antigens that are potentially presented by microglia during neurodegeneration?

Could targeting the interactions between peripheral immune cells and glial cells offer a putative therapeutic approach to mitigating AD-related neurodegeneration?

How do distinct T cell subpopulations, including effector and regulatory CD4⁺ T cells, cytotoxic and exhausted CD8⁺ T cells, as well as immune checkpoint regulators expressed by these cells influence AD pathology? Could immunomodulation therapies targeting these be effective in altering disease outcomes?

Can peripheral immune cells influence the brain indirectly through the release of soluble factors? If so, what are such mediators and how do they contribute to AD pathogenesis?

How can co-culture systems involving human peripheral immune cells and brain models be optimized to better replicate the human AD environment?

To enhance the relevance and output of AD models, what strategies might be employed to improve in vitro assays, including the addition of vascularized structures?

How can humanized systems and chimeric models be incorporated to explore immune cell function in AD, and what insights on the roles of peripheral immune cells in disease progression might these systems offer?

What lessons might be drawn from immunomodulating therapies used in other diseases? Could these approaches be adapted for AD to enhance therapeutic success or overcome limitations?

We suggest a two-step approach for studying the role of peripheral immune cells in AD. First, we propose a dissection of the disease-modifying roles of each immune cell type, ensuring consistency and reproducibility across findings. Although various adaptive and innate immune cells have been investigated in animal and cellular models of AD, results have been inconsistent or even contradictory. For example, while one study reported that depleting CD8⁺ T cells had no impact on AD pathology in animal models [32], other studies, including our own, have demonstrated the detrimental effects of CD8⁺ T cells in different models, such as the APOE4 tau mouse, 5xFAD amyloidosis mouse, and the human 3D multicellular model of AD containing neurons, glial cells, and peripheral immune cells [13,14,34]. Conversely, another group found CD8⁺ T cells to be protective in AD using various genetically knockout mouse strains crossed with an amyloidosis model [40]. These studies have consistently shown increased numbers of CD8⁺ T cells in AD mouse brains and cellular systems, reflecting similar increases in human patients presenting with mild cognitive impairment (MCI), AD, or tauopathies [13,18,92,93]. It is plausible that different CD8⁺ T cell subtypes present in the brain may complicate distinguishing between protective or harmful roles, making it challenging to fully understand their impact on disease. Additionally, these discrepancies might also arise from variations in AD models, strains, or species differences (e.g., humans vs mice), age and sex factors, as well as the diverse methodologies used to study CD8⁺ T cells (or other immune cell populations). Mouse model findings may not always translate well to human AD. For example, many studies rely on models of either amyloidosis or tauopathy, but not both simultaneously, which are key hallmarks of human AD. While each approach offers unique insights, there are clear limitations. We advocate for using a combination of models and carefully designed controls to gain a comprehensive understanding of how peripheral immune cells influence AD, ideally enabling translatable insights. Second, we propose a targeted exploration into the roles and mechanisms underlying each peripheral immune cell type in AD, focusing on their sites of action and effector functions. While these cells may be protective in the periphery, they can cause harm upon entering the brain by interacting with neural/glial cells and releasing cytokines that may worsen neurodegeneration [13,14,94]. By understanding their migration and specific functions, we can potentially mitigate their detrimental effects on the brain while preserving their protective roles in AD.

As the field of neuroimmunology advances, a significant challenge lies in developing models that accurately capture the role of peripheral immunity in AD as well as the complexity of the disease. Beyond AD-like animal models, emerging systems such as human iPSC (hiPSC)-derived organ-on-chip systems and chimeric models are essential. These innovative approaches allow for rigorous testing of how peripheral immune cells either protect neurons or contribute to their damage. Additionally, they can also help identify key stages, e.g., early phases of the disease, where the immune system influences disease progression and helps predict outcomes. There is optimism that the emerging technologies described here can more accurately replicate the human brain, and facilitate AD studies as well as target discovery. Emerging chimeric models, in which human cellular models are integrated into the mouse brain [25,95,96], may offer another innovative approach to creating a brain environment of complex cellular interactions that mimics human physiology and disease.

From another angle, the use of hiPSCs holds great promise because the reprogramming process erases somatic epigenetic signatures, including those linked to aging and disease. Indeed cerebral organoids created from 4-month-old hiPSCs resemble a human embryonic brain at 13 weeks post-conception [97]. Alternatively, directly converting fibroblasts into neurons (iNs) preserves age-related features and produces neurons with more mature properties than iPSC-derived cells [98]. However, each model has its limitations – hiPSCs can harbor DNA damage or oncogenic mutations [99], while direct cellular conversion faces challenges such as obtaining sufficient biopsy material and controlling genetic variability. It is also unclear if the epigenetic profiles of fibroblasts used to generate iNs accurately reflect those of aged neurons and other brain cells. Despite these challenges, hiPSC-derived neurons from both sporadic and familial AD reliably capture disease-relevant markers [100,101].

Additionally, integrating single-cell culture platforms with high-throughput sequencing and proteomic analyses can further enhance the resolution of neuroimmune studies. Such combinatorial approaches enable the comprehensive profiling of transcriptional and translational changes during immune-brain cell interactions, providing a holistic view of cellular crosstalk and its impact on neuronal function and survival. As these technologies advance, they will become more important in personalized medicine for neurodegenerative diseases such as AD. They may help develop targeted therapies that address the complex interactions between the CNS and the immune system.

Evidently, while no single model can fully replicate the complexity of human brain biology, the strategic combination of human cellular models and AD-like mouse models, supported by rigorous validation studies, offers a robust approach for advancing AD research (as well as other neurodegenerative conditions). Thoughtful integration of various models that are tailored to the specific neuroimmune questions will likely yield the most accurate and translatable data. This strategy may bridge the gap between experimental findings and clinical applications, potentially paving the way for more effective therapeutic interventions.

Highlights.

Emerging studies demonstrate the crucial role of peripheral immune cells in shaping AD pathology.

Targeting peripheral immune cells presents promising therapeutic potential for AD treatment.

In vivo and in vitro models are being used to study the specific role of peripheral immune cells in AD.

Although research on neuroimmune interactions is in its early stages, we require a comprehensive understanding of how these interactions (involving peripheral immune cells) contribute to AD pathology. This can be achieved by integrating various cutting-edge approaches, including newer (e.g. humanized) animal models, human organoids, organ-on-chips, single-cell analyses, high-resolution imaging, and computational biology.

Significance.

Despite significant progress in understanding the roles of microglia (brain’s resident immune cells) in Alzheimer’s disease (AD), much less is known about the contribution of peripheral immune cells to influencing AD pathology. Recent studies using rodent models and innovative human-based cellular systems are beginning to shed light on how peripheral immune cells infiltrate the brain and modulate disease progression. These promising cutting-edge approaches can expand our knowledge of the disease-modifying roles of these immune cells, paving the way for developing novel putative AD therapies.

Glossary

Amyloid-β (Aβ) plaques

Protein fragments formed by amyloid-β peptides in the brains of AD patients

Tau tangles

twisted protein fibers inside neurons linked to Alzheimer’s disease

Organoids

Self-organizing 3D cellular cultures derived from stem cells used to model organ function or disease

Organ-on-chip

multicellular microfluidic system that simulates the microenvironments of tissues or organs, facilitating the study of biological processes in a controlled setting

Chimeric models

Experimental models containing cells or tissues from genetically distinct sources, used to study human disease mechanisms in a living organism while preserving the physiological context of a mouse

Y maze

three-arm maze used to assess rodents’ spatial memory and exploratory behavior by evaluating the ability to remember and avoid previously visited arms

3xTg AD

AD mouse model characterized by both amyloid and tau pathology

5xFAD

widely used AD mouse model carrying human APP and PS1 mutations

THY-Tau22

tau mouse model expressing the MAPT transgene to study tau aggregation and its effects in AD and other tauopathies

APOE4 tauopathy

genetically modified mouse expressing human tau protein and the human APOE4 gene variant; used to study the interaction between APOE4 function and tau pathology in various neurodegenerative diseases

Barnes maze

behavioral test assessing spatial learning and memory in rodents using an elevated circular platform with escape holes and visual cues

Spontaneous alternation rate

cognitive test measuring working memory by evaluating the natural tendency of rodents to alternate between different arms during maze navigation

APP^NLGF

APP knock-in AD mouse model

μMT 5xFAD

5xFAD mice lacking B cells

Novel object recognition test

behavioral test evaluating memory in rodents by measuring their preference for exploring new versus familiar objects

Mucosal-associated invariant T (MAIT) cells

subset of innate T cells recognizing microbially-derived metabolites presented by the MR1 molecule; contribute to immune responses at mucosal surfaces

Antibody class switching

biological process where B cells change the antibody isotype they produce from one class (isotype) to another, contributing to diversity in immune responses

PSAPP mice

transgenic AD mouse model expressing mutations in APP and presenilin

Saporin

ribosome-inactivating protein that induces cell death upon internalization

Multiple sclerosis

chronic disabling disease of the CNS with autoimmune etiology

Experimental autoimmune encephalomyelitis (EAE)

animal model for MS that leads to immune cell infiltration, demyelination, and paralysis

Disease-associated microglia

subset of microglial cells that are linked to neurodegenerative diseases; characterized by a unique gene expression profile, with roles in the diseased brain

Type 2 innate lymphoid cells (ILC2)

innate immune cells that promote type 2 immune responses; involved in allergy, asthma, and tissue repair

PD-1 immune checkpoint

inhibitory receptor that when engaged (i.e. PD-L1) inhibits T cell activation, regulates tolerance, and prevents excessive immune activation that might lead to autoimmune disorders. Considered an important signaling checkpoint in T cells

Immunological synapse

Specialized junction that forms between an immune cell (e.g. T cell) and an antigen-presenting cell (APC) or target cell, facilitating communication and triggering immune activation

scATAC-Seq

technique to assess chromatin accessibility at the single-cell level, enabling the study of cell-specific regulatory elements and cellular heterogeneity within tissues

Cellchat platform

designed to analyze and interpret cell-cell communication by integrating single-cell data, such as scRNA-seq or scATAC-seq

Cancer spheroids

3D cell culture models derived from clusters of cancer cells

Assembloids

3D cellular systems formed by integrating multiple organoids

Tissue-resident memory T cells

subset of memory T cells that permanently reside within peripheral tissues, including but not limited to the brain, providing rapid, localized, immune protection by responding to invading pathogens or other threats

Toll-like receptor

class of pattern recognition receptors that play a key role in the innate immune system by recognizing pathogen-associated molecular patterns (PAMPs) and triggering immune pathways

Sporadic and familial AD

Sporadic or late-onset AD, occurs later in life with no clear genetic cause; familial AD (FAD) is a rare, early-onset form caused by inherited mutations in genes such as APP, PSEN1, or PSEN2. Both share common pathological features but differ in genetic origins and onset age