Escrever um comentário

The review is the result of a collaborative live review discussion held on October 28th, 2025, organized and hosted by the PhD school in Biomedical Sciences of the University of Padova as part of the Journal Club activity. The discussion was followed by additional asynchronous work by students and the JC committee, who jointly contributed to the composition and refinement of this report.

Summary

The manuscript entitled “Inflammatory reprogramming of human brain endothelial cells compromises blood-brain barrier integrity in Alzheimer’s disease” investigates how inflammation contributes to blood-brain barrier (BBB) dysfunction in Alzheimer’s disease (AD). By integrating post-mortem human single-nucleus transcriptomic analyses with induced pluripotent stem cell (iPSC)-derived BBB models and a perfusable BBB-Chip system, the authors identify an NF-κB-associated endothelial gene module (endoM2). The latter turns out to be elevated in AD and shows strong positive correlations with inflammatory pathways and endothelial-to-mesenchymal (EndMT) signatures. They show that inflammatory cytokine stimulation of iPSC-derived brain endothelial cells induces morphological and transcriptional changes consistent with EndMT, and compromises barrier integrity. These effects are mitigated by pharmacological inhibition of NF-κB using BAY11-7082. These findings link cerebrovascular inflammation to endothelial reprogramming and highlight NF-κB-driven endothelial plasticity as a potential therapeutic target to preserve BBB integrity in AD. The manuscript presents a well-conducted and novel investigation into how inflammatory processes contribute to BBB dysfunction in AD. The authors effectively integrate human post-mortem transcriptomic data with advanced in vitro models, providing valuable insights into the molecular mechanisms underlying cerebrovascular dysfunction in AD. The identification of the NF-κB-associated endothelial gene module (endoM2) and its association with cytoskeletal remodeling represents an important contribution to the field. These findings highlight a potentially actionable pathway for therapeutic strategies aimed at preserving BBB integrity in AD. Additionally, the pharmacological rescue by BAY11-7082 provides a potential intervention point and strengthens the translational relevance of the study. However, although the experimental approach is solid and the results are promising, the manuscript would benefit from clearer explanation of specific methodological details and a more thorough discussion of the limitations inherent to the in vitro models used. Addressing these points would strengthen the quality of the manuscript and enable readers to more accurately assess the applicability of the findings to broader aspects of AD pathophysiology.

List of major concerns and feedback

Below, we outline several major points that should be addressed by the authors:

1. APOE4 focus may limit generalizability. The manuscript relies heavily on APOE4 iPSC-derived endothelial cells. Yet the introduction provides only limited background on the role of APOE4 in cerebrovascular dysfunction and BBB impairment. Expanding this section by including current knowledge on APOE4-mediated effects would better contextualize the authors’ focus and more clearly frame the rationale underlying the study design. In addition, while APOE4 is a major AD risk factor, the study emphasizes this genotype without systematically addressing whether and to which extent these findings may apply to other familiar or sporadic AD cases.

2. Clarification of APOE4-specific mechanisms. The study shows that APOE4 iPSC-derived brain endothelial cells display heightened susceptibility to inflammatory cytokine stimulation and more pronounced activation of the endoM2 transcriptional program. However, the mechanistic basis of this APOE4-dependent vulnerability remains underexplored. Elaborating on how APOE4 influences endothelial signaling pathways that converge on NF-κB or EndMT-related processes would further strengthen the manuscript. For instance, previous works have linked APOE4 to lipid dysregulation, oxidative stress, and impaired receptor signaling (e.g., LRP1, TGF-β) (see references below), all of which could affect inflammatory responses. Additional discussion or supporting data clarifying whether these pathways contribute to the enhanced endothelial dysfunction in APOE4 backgrounds would improve mechanistic depth and contextual relevance of the study. To address this point, the authors could perform targeted assays such as: (1) NF-κB activation measurements (p65 nuclear translocation or reporter assays) in APOE3 vs APOE4 iBMECs under cytokine exposure; (2) ROS quantification and antioxidant rescue to test whether oxidative stress amplifies APOE4 vulnerability; and (3) modulation of lipid handling (cholesterol depletion or DGAT inhibition) to evaluate whether correcting lipid imbalance attenuates endoM2 activation.

1. Clarification on ROSMAP dataset stratification. The stratification of ROSMAP samples appears to rely primarily on statistical clustering of transcriptomic profiles rather than on neuropathological or clinical diagnoses of AD. This approach may introduce uncertainty regarding the assignment of individuals to “AD” versus “control” groups. The authors should clarify whether clinical or pathological diagnostic criteria were available and how these were integrated into the statistical stratification. Explicitly describing how diagnoses or group labels were defined would help readers to interpret the biological relevance of the identified endothelial modules.

2. Rationale for selecting prefrontal cortex samples. The authors base their single-nucleus RNA-seq analysis primarily on prefrontal cortex samples, despite observing correlations with vascular transcriptomic changes in other brain regions across independent datasets. The rationale for focusing on the prefrontal cortex should be clarified.

3. Discussion of endoM6 and endoM7 endothelial modules. While the manuscript focuses mainly on endoM2 as the inflammation-associated endothelial module, endoM6 and endoM7 modules are also mentioned but not discussed in detail. The analysis would be strengthened by briefly describing the biological signatures or pathways represented in these modules, especially if they reflect endothelial states (e.g., angiogenic, metabolic, or stress-related) that complement or contrast with the inflammatory endoM2 profile.

4. LPS treatment effects and transcriptomic-functional discrepancy. The authors report that LPS exposure upregulates inflammatory and EndMT-like transcriptomic profiles in iPSC-derived endothelial cells, yet no corresponding morphological or barrier functional changes are observed. This apparent dissociation between gene expression and cellular phenotype warrants clarification. The authors should discuss possible explanations, including whether the timing or dosage of LPS treatment may have contributed to this discrepancy.

5. Limited mechanistic specificity of NF-κB signaling. The study identifies NF-κB inhibition as protective against cytokine-induced endothelial dysfunction; however, the downstream effectors responsible for mediating this rescue remain undefined. Given that NF-κB interacts extensively with other signaling axes such as STAT, TGF-β, and oxidative stress pathways (see references below), all of which are relevant to EndMT and AD pathophysiology, further exploration of these interactions would enhance mechanistic clarity of the study. For example, transcriptomic or phosphoproteomic analyses following NF-κB inhibition could help determine whether key inflammatory and mesenchymal regulators are directly modulated. Without such analyses, the protective effect of BAY11-7082 remains relatively broad and could reflect general anti-inflammatory or stress-suppressive actions rather than a specific NF-κB-dependent mechanism.

6. Pharmacological specificity and translational implications of BAY11-7082. The identification of BAY11-7082 as a protective NF-κB inhibitor is intriguing; however, this compound is known to have off-target effects on other pathways, including inflammasome activation and ubiquitination cascades (see references below). The authors should acknowledge these limitations and discuss whether more selective NF-κB inhibitors (e.g., IKKβ-specific inhibitors such as TPCA-1 or MLN120B), or genetic perturbations (e.g., siRNA or CRISPR-mediated knockdown of NFKB1), could provide complementary evidence for the specificity of the observed rescue effects. Additionally, discussing the translational feasibility and potential neurovascular safety of BAY11-7082 in vivo would further enhance the clinical relevance of the study.

7. Barrier dysfunction claims without direct permeability measurements. In both the 2D iBMEC monolayers and the 3D BBB-Chip, barrier dysfunction is checked from morphological remodeling (nuclear elongation, vascular retraction), lipid accumulation and junctional changes (ZO-1 redistribution), but direct functional readouts of barrier integrity (e.g. TEER or tracer permeability) are not reported. Given that junctional disorganization does not always translate into increased permeability, the absence of TEER or lumen-to-matrix leak measurements represents an important limitation.

List of minor concerns and feedback

In addition to the major points outlined above, we have few minor comments:

1. Experimental controls. The manuscript would benefit from clearer specification of the experimental controls used in each assay. In particular, indicating which control conditions (e.g., untreated, vehicle-treated, or genotype-specific) were applied in every figure and analysis would improve transparency and reproducibility.

2. Figure prioritization. Several figures include extensive data panels, which can make the main findings less immediately apparent (e.g., Figures 1, 3 and 5). Consider prioritizing or consolidating figures to highlight the most critical results and relocating supporting data to supplementary materials where appropriate.

3. Endothelial module nomenclature. The naming and labeling of endothelial gene modules (e.g., endoM2, endoM4, turquoise module, endoBBB-M1) could be simplified and standardized throughout the text and figures to reduce confusion and improve interpretability across datasets.

Concluding remarks

Overall, this manuscript provides a compelling and methodologically robust investigation into inflammatory endothelial reprogramming and BBB dysfunction in AD. Addressing the points outlined above, particularly those concerning APOE4-specific mechanisms, would further strengthen the translational impact of the study.

Competing interests

The authors declare that they have no competing interests.

References

Previous work has implicated APOE4 in lipid dysregulation, oxidative stress, and impaired receptor signaling

Lipid dysregulation: https://pubmed.ncbi.nlm.nih.gov/38480892/

Oxidative stress: https://pubmed.ncbi.nlm.nih.gov/32036033/

Impaired receptor signal: https://pubmed.ncbi.nlm.nih.gov/28434655/

NF-κB interacts extensively with other signaling axes such as STAT, TGF-β, and oxidative stress pathways

Oxidative stress: https://pubmed.ncbi.nlm.nih.gov/37761028/

STAT and TGF-B: https://www.nature.com/articles/s41392-024-01757-9

BAY11-7082 has known off-target effects on other pathways, including inflammasome and ubiquitination cascades

Ubiquitination: https://pubmed.ncbi.nlm.nih.gov/40765826/

Inflammasome: https://pubmed.ncbi.nlm.nih.gov/20093358/

Competing interests

The authors declare that they have no competing interests.

Use of Artificial Intelligence (AI)

The authors declare that they did not use generative AI to come up with new ideas for their review.