# Perivascular cells induce microglial phagocytic states and synaptic engulfment via SPP1 in mouse models of Alzheimer's disease.

- **DOI:** [10.1038/s41593-023-01257-z](https://doi.org/10.1038/s41593-023-01257-z)
- **Rating:** Strong (4.0 out of 5)
- **Verdict:** Identifies perivascular macrophage-derived SPP1 as an upstream driver of microglial synaptic engulfment in Alzheimer's mouse models, constrained by an untested translation gap to human therapy.
- **Framework version:** v5.0
- **Pipeline version:** v0.4
- **Evaluation date:** 2026-06-05
- **Evaluation page:** https://www.nabu.science/paper/10.1038%2Fs41593-023-01257-z

## Strengths

- **Orthogonal methods converge on one conclusion** STED super-resolution imaging, correlative light-electron microscopy, single-molecule FISH, single-cell RNA sequencing, genetic knockouts, and in vitro rescue all triangulate the same causal chain from perivascular SPP1 to microglial engulfment.
  - Dimension: methodological rigour
  - Location: Figs. 1-5 and Extended Data Figs. 1-6
- **Identifies a new cellular source of SPP1** The work localizes SPP1 predominantly to perivascular macrophages rather than microglia in the adult hippocampus, a reassignment supported by reporter mice, CLEM, and RiboTag data, reframing a known AD marker as an upstream functional signal.
  - Dimension: contribution
  - Location: Results (SPP1 is expressed by PVMs and fibroblasts); Fig. 2
- **Genetic necessity demonstrated with synapse rescue** Crossing Spp1 knockout to the AppNL-F model prevents microglial phagocytic marker upregulation and rescues synapse loss despite amyloid challenge, and extracellular SPP1 restores engulfment in knockout microglia, supporting an extrinsic signaling mechanism.
  - Dimension: contribution
  - Location: Fig. 4 and Extended Data Figs. 4-5

## Limitations

- **Small sample sizes in key experiments** Several pivotal readouts rest on low animal numbers, including n=2 for one Homer1 engulfment quantification and n=3 from a single independent experiment for the STED SPP1 measurement, which limits statistical confidence in individual findings.
  - Dimension: methodological rigour
  - Location: Fig. 1e,f; Fig. 3h; Fig. 5e
- **Computational network analysis is predictive only** The NicheNet ligand-target predictions are based on sparse sorted-cell numbers (around 200 PVMs and 350 PVFs per genotype) and Spp1 was not even detected in the sorted PVMs; only select hits are orthogonally validated, so the network claims carry substantial uncertainty.
  - Dimension: methodological rigour
  - Location: Fig. 5; Methods (scRNA-seq library preparation)
- **Translation to human disease untested** Human evidence is limited to descriptive postmortem perivascular SPP1 localization with no functional or pharmacological validation, and no therapeutic intervention targeting SPP1 is tested, leaving a substantial gap between mouse genetics and clinical utility.
  - Dimension: application pathway
  - Location: Fig. 1j, Fig. 2i; Discussion

## Reliability checks

Caveats (4 of 4 checks).

- **Internal coherence:** Noted. The paper is internally consistent across abstract, figures, and results, with multiple orthogonal methods converging on the same conclusion and the scRNA-seq vs smFISH/RiboTag Spp1 detection discrepancy explicitly explained as an isolation artifact. The one noted item is that Discussion passages generalize from mouse-model findings toward broader AD-relevant and disease-general crosstalk beyond the directly tested evidence.
- **Research conduct:** Noted. Ethics approvals are declared for animal work (UK Home Office/UCL, Jackson Laboratory IACUC, Weizmann Institute Animal Care Committee) and human postmortem tissue (NHS ethics committee, license 12,198; Queen Square Brain Bank with informed consent). These are declared-but-unverified at the editorial layer; no undisclosed COI was identified; data is shared only via linked source files with no named repository and no project-specific code, which is the noted item.
- **Reference integrity:** Clean. All 85 cited references resolved to real publications via PubMed, Crossref, or OpenAlex.
  - References not verified automatically are resolved manually within 24 hours.

## Post-publication record

- **Status:** Clean
- **Summary:** No retraction or expression of concern found in Crossref (includes Retraction Watch).
- **Correction published 17 Feb 2026**, listed in Crossref: https://doi.org/10.1038/s41593-025-02197-6
- **Last checked:** 2026-10-17
