Endothelial identity under pressure: paper watch for July 6-13, 2026

Endothelial papers this week clustered around one problem: how an endothelial cell keeps, loses, or reassigns identity while it is being pushed by growth factors, flow, inflammation, disease genetics, or engineered tissue contexts. The clearest read is that several groups are moving past one-signal explanations. VEGF-family inputs still sit near the center, but the papers below add GPCR signaling, E2F cell-cycle control, BMP/Notch/ERK cross-talk, mechanical force, and multicellular organoid context to the story.

I would read three papers first if your time is tight: the Apelin lymphatic-sprouting preprint for developmental mechanism, the E2F1 pulmonary-hypertension preprint for pathological arterialization, and the vascularized brain-organoid preprint for methods.

The Apelin paper is the cleanest developmental signal in the batch. Vegfc/Vegfr3 remains the upstream lymphangiogenic driver, but the authors place Apelin/Aplnr downstream as a migration-competence module. In their zebrafish work, loss of Apelin signaling disrupted sprouting from the posterior cardinal vein and thoracic-duct formation, while the abstract states that lymphatic endothelial-cell specification itself was not impaired. 1

That distinction is useful. If a signal changes fate specification, it can rewrite the cell identity program. If it changes migration competence, it may be a more selective handle on where and when lymphatic vessels grow. The paper's proposed chain is Vegfc-driven ERK activation → Aplnr expression → Apelin-dependent lymphatic sprouting. The translational hook is obvious: GPCRs are druggable. The harder question is whether a developmental migration module can be tuned in disease without disturbing maintenance roles in mature lymphatics.

E2F1 pushes the same identity-remodeling question into disease. The PAH preprint integrates human PAH lung transcriptomics, endothelial-state scoring, mouse and rat PH models, single-cell RNA-seq, pseudotime, and cell-communication inference. The authors report that genetic E2f1 loss reduced right-ventricular systolic pressure, right-ventricular hypertrophy, vascular remodeling, and distal muscularization in an Egln1-driven PH model. They also report that pharmacological pan-E2F inhibition attenuated both Egln1-driven and monocrotaline-induced PH, including reversal of established MCT-PH. 2

The part to watch is not just proliferation. The claim is that E2F1 couples cell-cycle activation to arterial endothelial programming: CAP1-to-arterial trajectories, Notch-associated programs, ECM signaling, and VEGF communication move together. If that holds up, PAH endothelial remodeling looks less like generic overgrowth and more like a misassigned endothelial-state transition.

The HHT preprint is built around a convergence argument. The authors connect reduced BMP9/ALK1 signaling from ENG or ALK1 pathogenic variants, and overactivation through SMAD6 variants, to increased endothelial proliferation and high ERK MAPK activation in patient biopsies. In cell models, SMAD6 loss or SMAD1 knockdown reproduced parts of the phenotype, and the paper links BMP9/Notch control of ERK to PPP1R3C, a regulatory subunit of PP1 phosphatase. 4

The in vivo readout is the attention point: in an adult ALK1-loss mouse model, vascular failure and hemorrhages in lung and intestine were reduced by the MEK/ERK inhibitor selumetinib. 4 That does not make ERK inhibition a ready HHT therapy, but it gives the field a testable axis that cuts across multiple genetic entry points.

The PubMed-indexed SETD2 paper is more translational and narrower. Under diabetic stress, the authors report reduced SETD2/H3K36me3 in HUVECs, impaired endothelial proliferation, migration, and tube formation when SETD2 is inhibited, and partial rescue by exogenous ANGPT2 through AKT phosphorylation. In endothelial-specific SETD2 knockout diabetic mice, hindlimb ischemia recovery and capillary density were impaired. 6 I would treat this as a mechanism lead rather than a therapeutic lead until the vascular selectivity problem is clearer.

The vascularized-organoid preprint is a methods paper with developmental consequences. The authors use genetic code expansion technology, introduced through a PiggyBac system, to enable temporal control of ETV2 expression during endothelial differentiation inside hPSC-derived cerebral organoids. The resulting vascularized human cerebral organoids are reported to show BBB-like features, perfusable vascular networks after transplantation into immunodeficient mice, endothelial subclusters resembling fetal brain, and three microglia-like subtypes that participate in microglia-vascular interactions. 3

For vascular-development readers, the method is interesting because it tries to coordinate multiple missing compartments at once: endothelium, barrier-like features, immune-like microglia, and neurovascular signaling. The Zika-virus experiment is also a stress test of the model, since the paper reports neurovascular dysfunction and impaired microglia development after infection. 3

The complex-flow preprint is the other methods pick. The authors use four engineered well geometries on a rotational-flow endothelial culture platform and pair them with CFD maps of wall shear stress, pulsatility, and oscillatory shear. Endothelial alignment and functional assays then map back onto the predicted local hemodynamic environments. 5

That matters because a lot of EC mechanobiology still compresses flow into a single condition label. This platform gives labs a way to ask whether the same culture surface contains separable force niches, closer to what a branch, curve, or narrowing creates in vivo.

A PubMed-indexed UNC5B/Netrin review argues for vascular-bed context rather than a single pro- or anti-angiogenic label. It describes UNC5B as a regulator whose effects depend on ligand availability, co-receptor expression, and local signaling cues, with expression heterogeneity across placental, retinal, and CNS endothelial subtypes. 7

That is the right caution for this week's primary papers too. Apelin may look like a migration-competence module in lymphatic development; E2F1 may look like a pathological arterial-programming node in PAH; ERK may look like an HHT convergence point. Each claim still has to survive vascular-bed specificity, developmental stage, disease state, and model-system differences.

For next week, I would watch for two kinds of follow-up: single-cell or spatial evidence that separates endothelial identity from behavior, and perturbation experiments that ask whether these proposed nodes can be tuned without damaging baseline vascular maintenance.