Knowledge Gaps and Priorities for Future Research

Despite extensive experimental evidence linking NF-κB to AD, a central question remains unresolved: when, where, and in what molecular form does NF-κB shift from an adaptive response to a disease-promoting program? Protective effects are most consistently observed during transient neuronal stress responses, trophic signaling, c-Rel-dependent antioxidant programs, and selected astrocytic compensatory functions. By contrast, persistent IKKβ–RelA signaling is repeatedly associated with amyloidogenic processing, inflammatory secretion, tau propagation, blood–brain barrier dysfunction, and neuronal injury. It remains unclear whether these patterns represent a temporal transition during disease progression, distinct programs operating simultaneously in different cell types, or both.

Resolving this question will require longitudinal human evidence. Genetic findings such as SHARPIN-associated impaired inducible signaling suggest that abnormalities in NF-κB regulation can precede clinical disease, whereas postmortem studies predominantly reveal altered RelA/p65 activity after substantial pathology has developed. However, these approaches cannot determine whether impaired stress responsiveness in preclinical disease later evolves into persistent inflammatory activation. Longitudinal studies spanning preclinical AD, mild cognitive impairment (MCI), and dementia are therefore needed to relate NF-κB pathway state to amyloid, tau, neurodegeneration, and clinical progression within the same individuals.

Greater cellular and molecular resolution is equally important. Most human studies rely on bulk tissue, conventional immunohistochemistry, or generic markers such as p65 phosphorylation, nuclear translocation, IκB degradation, or κB-DNA binding. These measures cannot distinguish whether signaling originates from neurons, microglia, astrocytes, pericytes, endothelial cells, or infiltrating immune cells, nor can they resolve functionally distinct NF-κB complexes. This distinction is critical because c-Rel-dependent signaling can support neuronal survival, whereas persistent RelA-associated signaling is more often linked to amyloidogenic and inflammatory programs, and p50 can participate in either activating p50/RelA heterodimers or transcriptionally repressive p50/p50 complexes. Future studies should therefore combine spatial transcriptomics, single-nucleus multi-omics, phosphoproteomics, and cell-resolved protein analyses with measurements of dimer composition, promoter occupancy, cofactors, post-translational modifications, and downstream target genes.

The kinetics of signaling also remain poorly defined. Many protective experiments examine NF-κB activation over minutes or hours, whereas AD develops over decades. Acute preconditioning and trophic responses demonstrate that transient activation can be beneficial, but they provide little information about repeated or sustained signaling in the aging brain. Conversely, chronic genetic manipulations frequently reveal detrimental effects but cannot determine whether intermittent physiological NF-κB responses remain necessary. Stage-specific experiments that manipulate the same NF-κB program within the same cell type while independently varying signal intensity and duration are needed to determine whether adaptive signaling crosses a definable threshold into pathology or whether distinct molecular programs are engaged from the outset.

A related methodological issue is the frequent conflation of IKKβ activity with NF-κB-dependent transcription. IKKβ can regulate autophagy, proteostasis, receptor signaling, and cell death independently of canonical NF-κB transcription. This may explain why IKKβ activation can improve autophagic handling or reduce necroptosis in some acute models, whereas persistent neuronal or microglial IKKβ activity promotes amyloid production and tau propagation in chronic disease. Future experiments should combine IKKβ manipulation with dimer-specific blockade, κB-site mutation, target-gene perturbation, and downstream rescue before assigning an observed phenotype specifically to canonical NF-κB signaling.

Important uncertainties also remain within individual cell types. In microglia, uptake of Aβ or tau is often interpreted as successful clearance even though internalization, lysosomal degradation, intracellular retention, extracellular release, and subsequent seeding are biologically distinct processes. Future studies should follow pathological cargo through each of these steps while simultaneously measuring inflammation, neuronal integrity, and functional outcomes. In astrocytes, the key challenge is to distinguish compensatory responses from pathological secretory programs. Astrocytic NF-κB may support plaque-associated organization and selected homeostatic functions while also suppressing KLK7-dependent Aβ degradation or promoting COX−2/PGE2, complement C3, and HDAC7-dependent neurotoxicity. These processes should be measured together rather than inferred from plaque burden or astrogliosis alone.

The neurovascular compartment requires similar integration. Existing studies implicate APOE4-dependent CypA–NF-κB–MMP9 signaling, soluble Aβ-induced endothelial activation, and RAGE–NF-κB-dependent loss of P-glycoprotein in blood–brain barrier disruption, hypoperfusion, leukocyte trafficking, and impaired Aβ efflux. Yet these mechanisms have largely been investigated separately. More physiologically complete human neurovascular models incorporating endothelial cells, pericytes, astrocytic endfeet, basement membrane, physiological shear stress, APOE genotype, and aging-related states are needed to determine how these pathways interact. In parallel, longitudinal measures of BBB integrity and cerebral perfusion should be integrated with amyloid, tau, and inflammatory biomarkers to establish whether neurovascular NF-κB dysfunction precedes or accelerates neuronal degeneration.

Translation is further limited by the heavy reliance on familial AD models, acute Aβ exposure, and relatively young experimental systems. APP/PS1, 5xFAD, APP23, TgCRND8, PS19, and related models have been invaluable for establishing causality but reproduce only selected components of predominantly sporadic late-onset AD. Likewise, high-dose or intracerebroventricular Aβ exposure does not reproduce decades of progressive amyloid accumulation, vascular aging, metabolic dysfunction, and immune adaptation. Human iPSC models reduce species differences but also lose important aging-associated features during reprogramming. Greater use of aged humanized models, directly converted cells that retain aging signatures, multicellular organoid or assembloid systems, and sporadic-risk backgrounds such as APOE4 will therefore be important. Sex should also be incorporated explicitly, because aging, mitochondrial function, immune responsiveness, lipid metabolism, and vascular biology can differ substantially between males and females.

Another major gap is the limited understanding of intercellular NF-κB communication. AD develops within an interconnected neurovascular system rather than within isolated cell populations. Neuronal proteotoxic stress can activate glial NF-κB; astrocytes can modify microglial responses; microglia can promote neuronal injury or tau propagation; astrocytic APOE can regulate pericytic signaling; and endothelial dysfunction can increase exposure of the brain to circulating inflammatory factors. It remains unclear which cellular compartment initiates these interactions, whether the dominant source of pathological NF-κB changes with disease stage, and whether suppressing signaling in one cell type alters pathology in another. Multicellular models allowing independently inducible NF-κB manipulation in neurons, astrocytes, microglia, pericytes, and endothelial cells will be needed to establish causal direction across these networks.

Human genetics may help define such cell-specific NF-κB states. Rare SHARPIN variants suggest that impaired inducible pathway competence can increase LOAD susceptibility, whereas APOE4 promotes exaggerated inflammatory responses in astrocytes and removes restraint from pericytic CypA–NF-κB signaling. These findings imply that genetic risk factors may perturb different components of NF-κB regulation rather than uniformly increasing or decreasing the pathway. Genotype-stratified cellular models and larger genetic studies linked to cell-specific functional assays could determine whether distinct NF-κB endophenotypes identify biologically meaningful subgroups of AD.

A major barrier to addressing these questions in patients is the absence of validated biomarkers of NF-κB state in the living brain. Current human evidence depends largely on genetics, peripheral immune assays, indirect inflammatory markers, or postmortem tissue. None can reliably distinguish preserved inducible signaling from chronic RelA-biased activation or identify the responsible cell type. Development of cerebrospinal-fluid, extracellular-vesicle, molecular-imaging, phosphoproteomic, or transcriptional signatures of specific NF-κB programs would permit longitudinal testing of experimental mechanisms and provide essential pharmacodynamic markers for future intervention studies.

These gaps converge on the principal therapeutic challenge: pathological NF-κB signaling must be suppressed without eliminating physiological functions required for neuronal stress resistance, immune defence, tissue repair, proteostasis, glial support, and vascular homeostasis. The available evidence therefore argues against treating NF-κB as a single drug target. More promising strategies would selectively target disease-conditioned upstream receptors, cell-specific IKK–RelA complexes, pathological cofactors or post-translational modifications, or downstream effectors such as BACE1, C3/C3aR, MMP9, NLRP3, and tau-release pathways. The key objective for future research is therefore not simply to determine whether NF-κB is protective or detrimental, but to identify which NF-κB program is active, in which cell, at which disease stage, and with what biological consequence. Achieving this level of resolution will be essential for moving from global NF-κB modulation toward precision targeting of pathological signaling while preserving its physiological and compensatory functions.