Introduction

Nuclear factor kappa B (NF-κB) is involved in several processes central to Alzheimer disease (AD), including neuronal stress responses, neuroinflammation, amyloid-β (Aβ) and tau pathology, synaptic dysfunction, and neurovascular injury (Kaltschmidt et al., 2024, Singh and Singh, 2020, Sun et al., 2022). Its role in AD, however, has been difficult to define because NF-κB signaling does not produce a single biological outcome. Under some conditions, it supports neuronal survival and adaptive stress responses, whereas under others it drives inflammatory and neurodegenerative processes. These apparently opposing effects are therefore better understood as context dependent rather than truly paradoxical; these findings suggest that the biological consequences of NF-κB depend on the signaling complex engaged, the responding cell, the initiating stimulus, and the duration and downstream consequences of activation.

This context dependence is a fundamental feature of the NF-κB signaling system. The NF-κB family comprises RelA/p65, RelB, c-Rel, p50/NFKB1, and p52/NFKB2, which can combine to form different homo- and heterodimers with distinct transcriptional properties. The activity of these dimers is further influenced by interacting cofactors, chromatin accessibility, post-translational modifications, and the timing and duration of signaling (Oeckinghaus and Ghosh, 2009). NF-κB activation occurs mainly through canonical and non-canonical pathways. Canonical signaling typically involves IκB kinase (IKK)-mediated release and activation of complexes containing RelA or c-Rel, whereas non-canonical signaling depends on NF-κB-inducing kinase (NIK), IKKα, processing of p100 to p52, and subsequent activation of RelB/p52 complexes (Sun, 2012). In AD, however, the non-canonical pathway remains relatively poorly characterized, as most mechanistic studies have focused on canonical NF-κB signaling (Kaltschmidt et al., 2024, Singh and Singh, 2020, Sun et al., 2022).

These distinctions are essential when interpreting experimental evidence, because commonly used readouts capture different stages of NF-κB signaling and should not be treated as equivalent. IκB degradation indicates release of NF-κB from cytoplasmic inhibition, whereas nuclear p65 accumulation reflects translocation, κB-DNA binding demonstrates interaction with regulatory DNA, and target-gene induction provides stronger evidence that transcriptional activity has actually occurred. Similar caution is required when interpreting IKKβ manipulation. Although IKKβ is a central component of canonical NF-κB signaling, it also phosphorylates substrates outside the NF-κB pathway and can independently influence autophagy, proteostasis, and cell-death processes (Antonia et al., 2021, Häcker and Karin, 2006). Therefore, findings based solely on IKKβ manipulation or nuclear localization of p65 should be interpreted cautiously, because neither observation by itself identifies the specific NF-κB transcriptional program responsible for the biological effect. IKKβ can regulate processes beyond canonical NF-κB transcription, and nuclear p65 does not reveal which NF-κB dimer is active or which target genes are functionally involved. Mechanistic evidence is therefore strongest when studies directly connect a defined NF-κB complex to a specific transcriptional target and biological outcome using complementary approaches such as gain- or loss-of-function experiments, promoter-binding or reporter assays, pathway inhibition, and rescue experiments.

Cellular context provides another important explanation for the apparently conflicting effects of NF-κB in AD. NF-κB signaling occurs in neurons, microglia, astrocytes, endothelial cells, and pericytes, but its biological consequences differ substantially across these cell types. In microglia, for example, NF-κB activation can promote the release and propagation of seeding-competent tau (Wang et al., 2022a). In astrocytes, its effects appear more context dependent: NF-κB may support selected adaptive responses once pathology is established, whereas persistent activation in a relatively healthy brain can disturb inflammatory and metabolic homeostasis (Jong Huat et al., 2024). NF-κB also contributes to neurovascular dysfunction. In APOE4 carriers, impaired suppression of the pericytic cyclophilin A (CypA)–NF-κB–MMP9 pathway can promote blood–brain barrier breakdown and cerebral hypoperfusion (Bell et al., 2012). These findings show why NF-κB activation cannot be classified as uniformly protective or pathogenic on the basis of a single molecular marker or outcome. Likewise, an improvement in one endpoint, such as reduced plaque burden or altered cytokine expression, does not necessarily indicate an overall beneficial effect. A meaningful interpretation therefore requires consideration of the molecular pathway involved, the responding cell type, the underlying pathological state, and the broader functional consequences.

Viewed within this framework, the dual effects of NF-κB in AD appear to be context dependent but not equally supported across biological settings. Protective effects are most consistently observed under relatively restricted conditions, particularly during transient neuronal stress responses, c-Rel-dependent survival signaling, and selected compensatory functions in glial cells. In contrast, sustained disease-associated IKK–RelA signaling has been linked across a wider range of experimental models to amyloidogenic processing, amplification of neuroinflammation, tau propagation, and neurovascular dysfunction.

Accordingly, this review evaluates primary studies in relation to their cellular and molecular context and the strength of the causal evidence they provide. Particular emphasis is placed on cell-specific gain- and loss-of-function studies and on experiments that use pathway inhibition, genetic manipulation, or rescue approaches to establish mechanistic links. The central question is therefore not whether NF-κB is inherently protective or pathogenic in AD, but rather which NF-κB signaling program is engaged, in which cell type, under what pathological conditions, and with what biological consequences.