Frank Paul Mora (@InTime_49) header
Frank Paul Mora (@InTime_49) profile picture

Retired Haskell programmer. Was pro drummer & still play hours/day. Science, Philosophy, Literature. DMs blocked unless shared interests.

New Mexico, USA
Joined December 2012
Photo by Frank Paul Mora (@InTime_49)Photo by Frank Paul Mora (@InTime_49)Photo by Frank Paul Mora (@InTime_49)Photo by Frank Paul Mora (@InTime_49)Photo by Frank Paul Mora (@InTime_49)Photo by Frank Paul Mora (@InTime_49)Photo by Frank Paul Mora (@InTime_49)Photo by Frank Paul Mora (@InTime_49)Photo by Frank Paul Mora (@InTime_49)Photo by Frank Paul Mora (@InTime_49)
Frank Paul Mora retweeted
Engineered Outer Membrane Vesicle Nanoengagers Potentiate NK Cell Immunotherapy through Immune Bridging and Mitochondrial Apoptosis Priming @MolTherapy sciencedirect.com/science/ar…
4
15
912
Frank Paul Mora retweeted
Transformer-based and biological processes may be understood as different instantiations of a single computational type phisci.info/jour/article/vie…
4
3
426
Frank Paul Mora retweeted
Most people on Twitter pretend they understand the Riemann Hypothesis. Here is an explanation probably simpler than you think: Start with prime numbers like 2, 3, 5, 7 and 11. There are 25 primes up to 100, and you can check that by hand. Now imagine counting them up to a trillion. Mathematicians can get a pretty good estimate using something called the logarithmic integral. It takes into account that primes get rarer as you count higher. Riemann wanted to know how far that estimate could be from the actual count. He found a way to study that gap using a formula called the zeta function. Before getting into that, try a much simpler rule. Take any number and subtract 3. Put in 5 and you get 2. Put in 3 and you get zero. That makes 3 a zero of that rule. A zero is simply an input that makes the answer zero. The zeta function can take inputs with two coordinates, like points on a map. Every time an input gives you zero, mark that input’s location with a dot. Set aside the easy zeros at minus two, minus four, minus six and so on. Riemann guessed that every remaining dot sits on the same vertical line, the one that crosses the horizontal axis at one-half. The dots can sit at different heights. They all have to be exactly as far to the right. Why does that matter for primes? The dots help explain why the actual prime count is sometimes above the estimate and sometimes below it. A dot farther to the right can make the estimate go further off as you count higher. If all the dots Riemann was talking about line up at one-half, there’s a tight limit on how fast that error can grow. Computers have checked trillions of zeros, and they’ve all lined up. A single zero off the line would prove Riemann wrong. A proof has to show why that can never happen, even for the zeros we’ll never get around to checking one by one. On November 20, Anthropic mathematician Levent Alpöge said he believes he found that proof with Claude’s help. He released the argument in a form computers can check step by step. Two independently developed programs accepted it, with further checks underway. That same day, Ravi Vakil, president of the American Mathematical Society, shared a separate 940-page draft by Alessandro Maria Masullo, who worked without AI. Masullo’s draft tackles the Generalized Riemann Hypothesis, a stronger statement that would settle the original too. Vakil later said readers had flagged problems that still need to be resolved.
2
2
5
403
Frank Paul Mora retweeted
Accumulating biological evidence showing that #LongCOVID involves deep, systemic shifts at the cellular level rather than just lingering mild symptoms When inflammation and cellular aging disrupt your bone marrow, the exhaustion isn't "in your head"—it's in your blood production
COVID-19 may disrupt the entire blood-production system—from stem cells to mature blood cells! ➡️ Beyond direct viral infection, spike protein and inflammatory signals may trigger bone-marrow dysfunction through inflammasome activation, pyroptosis, cellular senescence, epigenetic changes and altered marrow niches. 🔹 This may contribute to lymphopenia, thrombocytopenia, anemia and coagulopathy. 🔹 Some hematological abnormalities may persist long after viral clearance, potentially contributing to #LongCOVID. 1/
2
25
1
95
2,086
Frank Paul Mora retweeted
Some theories say a mind is conscious only if it not only represents the world but also "knows" about its own mental states. This paper asks what that means for AI. Real Consciousness or Real Illusion? A Higher-Order Approach to AI Consciousness philpapers.org/rec/BRORCO-5
8
14
54
2,721
Frank Paul Mora retweeted
A reminder that we vaccinate for ourselves yes - but also for others. A baby too young for the measles vaccine, a child going through chemo, a neighbor with a weakened immune system, someone’s aging parent. They may never know you but your choice can still help keep them safe.
217
174
4
534
9,456
Frank Paul Mora retweeted
Cells photographed through a microscope by @EmmaKoory. #CellBiology
2
20
2
159
2,890
Frank Paul Mora retweeted
Macrophages shape healthy adipose expansion dlvr.​it/TWPLMZ
2
14
56
3,284
Frank Paul Mora retweeted
#Atherosclerosis #Lipoprotein(a) Expansion of a mesenchymalized (TGFβ-hi & matrix-depositing) Macrophage subpopulation🤩 (or a non-foamy, phagocytic smooth muscle cell state!?🤓) in Lp(a)-transgenic 🐭 atherosclerotic lesion A great piece from our nicest neighbor🥳 @JuliaStJohn253 @atvbahajournals 2026 ahajournals.org/doi/10.1161/…
1
8
1
22
917
Frank Paul Mora retweeted
Extracellular vesicle budding is a vital cellular process that allows cells to communicate with each other and the microenvironment. Release of vesicles to the extracellular space can occur through an internal pathway, whereby vesiclecontaining compartments fuse with the plasma membrane to release vesicles called exosomes, or by vesicle budding directly from the cell surface. The latter occurs even at the surface of a primary cilium, a tiny antenna-like organelle that projects from the surface of most vertebrate cells. Although it makes up less than a thousandth of the plasma membrane, the cilium is a major signaling hub. In a new Science study, researchers report that cilial budding does not rely on a complex machinery generally harnessed for extracellular vesicle formation but on the condensation of a single component of this machinery that drives the concentration of cargo and the release of vesicles from this structure. Learn more in this #SciencePerspective: scim.ag/4eGrrv7
6
27
2
152
19,686
Frank Paul Mora retweeted
Genomic surveillance reveals age-structured SARS-CoV-2 transmission across settings @CellCellPress cell.com/cell/fulltext/S0092… @GageKMoreno @broadinstitute @tbrockfi @lakrasil @PardisSabeti @bronwynmacinnis
5
4
667
Frank Paul Mora retweeted
CD19 CAR T cell therapy alters disease-associated immune signatures in multiple sclerosis @CellCellPress cell.com/cell/fulltext/S0092… 🇩🇪
7
16
1,226
Frank Paul Mora retweeted
Levent Alpöge claims to have solved the Riemann Hypothesis "It is natural to feel that this is an end ... but what I saw during the course of progress so amazed me that I believe it is one dawn across science of something far greater" alphaxiv.org/abs/2611.claime…
5
43
2
247
9,282
Frank Paul Mora retweeted
NLRP3 as a driver of post-traumatic microglial morphodynamics and support selective NLRP3 inhibition as a promising strategy to improve neurological outcomes after traumatic brain injury cell.com/cell-reports-medici… @CellRepMed @michael_heneka @CastroGomezSerg
7
30
1,770
Frank Paul Mora retweeted
✨New issue✨ Check out the latest in #Metabolism research from across all disciplines. Read about #ReproductiveHealth, #AdiposeTissue, #Aging, #ImmuneFunction, #Cancer and many more topics! 🧪👥🐭 nature.com/natmetab/volumes/…
42
1
153
7,517
Frank Paul Mora retweeted
Human iPS cell-derived thymus-seeding T cell progenitors support in vivo polyclonal naive-like T cell development @CellStemCell cell.com/cell-stem-cell/full… 🇯🇵
9
22
1,387
Frank Paul Mora retweeted
🧬🫀 What if a blood test could identify vascular aging before cardiovascular disease develops—and reveal proteins that actively regulate endothelial and pericyte senescence? A new study in npj Aging integrates plasma proteomics from 42,326 UK Biobank participants, machine learning, Mendelian randomization, cross-species analyses, and functional experiments to identify candidate regulators of vascular aging. The central discovery: 2,920 plasma proteins → 44 vascular aging-associated proteins → 11 genetically prioritized candidates → 9 expression-supported proteins → 5 functionally investigated regulators of vascular senescence Among the most compelling candidates: IGFBP2 and LAYN may protect against vascular cell senescence, whereas FAM20A, ERBB2, and SMAD5 exhibit context-dependent pro-senescence effects in experimental models. Importantly, the findings provide candidate biomarkers and mechanistic targets—not clinically validated anti-aging therapies. 🔬 1. A population-scale proteomic investigation of vascular aging Researchers analyzed data from the UK Biobank Pharma Proteomics Project. The discovery cohort included: • 42,326 participants • 2,920 plasma proteins • 18,807 individuals classified as having vascular aging • Median age: 58 years • Female participants: 53.7% Plasma proteins were quantified using the Olink proteomics platform. The investigators combined: • Proteome-wide association analysis • Age- and sex-stratified analyses • Machine learning • Longitudinal cardiovascular outcomes • Mendelian randomization • AlphaFold 3 structural modeling • Mouse aortic protein measurements • Human serum validation • Primate arterial single-cell RNA sequencing • Human vascular cell knockdown experiments This multilevel design allowed researchers to move beyond simple biomarker discovery toward functional prioritization. 🧬 2. How was vascular aging defined? The investigators used an operational composite phenotype combining: Heart Vascular Age (HVA) ≥60 years and Estimated Pulse Wave Velocity (ePWV) ≥10 m/s HVA was calculated from the Framingham Risk Score, incorporating conventional cardiovascular risk factors. ePWV was estimated using age and blood pressure. Under these criteria: 44.4% of participants met the vascular aging definition. The prevalence differed by sex: • Men: 48.4% • Women: 41.0% It also differed substantially by age: • Age >60 years: 82.2% • Age ≤60 years: 19.0% Importantly, this definition represents an adverse vascular risk phenotype rather than a direct measurement of intrinsic biological vascular aging. ⚡ 3. The vascular aging phenotype corresponds to greater arterial stiffness To assess whether the operational definition captured vascular abnormalities, researchers examined arterial stiffness index (ASI). In a broader UK Biobank sample: Vascular aging group: ASI = 10.19 ± 4.74 Non-vascular aging group: ASI = 8.71 ± 3.28 Mean difference: 1.48 95% CI: 1.44–1.52 P < 0.001 These findings indicate that the operational vascular aging classification identifies individuals with greater arterial stiffness. However, ASI is not equivalent to reference-standard carotid–femoral pulse wave velocity. 🔋 4. Forty-four plasma proteins are associated with vascular aging Proteome-wide logistic regression identified: 44 proteins meeting Bonferroni-corrected significance. Of these: 12 proteins were associated with lower odds of vascular aging. Examples: IGFBP2 OR = 0.80 95% CI: 0.74–0.86 LAYN OR = 0.72 95% CI: 0.63–0.82 NTproBNP OR = 0.84 95% CI: 0.80–0.87 Conversely: 32 proteins were associated with higher odds of vascular aging. Examples: DCXR OR = 1.21 95% CI: 1.11–1.31 ERBB2 OR = 1.57 95% CI: 1.30–1.89 SMAD5 OR = 1.67 95% CI: 1.34–2.10 These findings suggest that vascular aging is associated with a broad molecular signature involving growth factors, extracellular matrix regulation, metabolic pathways, and immune signaling. However, associations between circulating proteins and vascular aging do not establish causal effects. 🧠 5. Most protein associations persist after adjusting for cardiovascular risk factors A critical question was whether the identified proteins merely reflected conventional cardiovascular risk. The investigators therefore performed sensitivity analyses using continuous ASI. Models adjusted for: • Age and sex • Blood pressure • BMI • Smoking • Alcohol intake • Diabetes • Hypertension • Lipid levels • Cardiovascular medication use • Socioeconomic deprivation • Additional sampling-related variables Among the 44 vascular aging-associated proteins: 38 remained significantly associated with ASI after FDR correction. 30 also passed Bonferroni correction. Furthermore: 39 of 44 proteins showed directionally consistent associations. These results suggest that many identified proteins capture information related to arterial stiffness beyond conventional cardiovascular risk factors. Nevertheless, residual confounding and limitations of the operational vascular aging phenotype remain important. 🧬 6. Vascular aging proteins form six functional networks Protein interaction and enrichment analyses identified six major biological modules. Module 1: VEGF signaling Associated with vascular growth and endothelial regulation. Module 2: Growth factor binding Associated with extracellular signaling and tissue remodeling. Module 3: Plasma lipoprotein particles Associated with lipid transport and metabolic homeostasis. Module 4: Humoral immune responses Associated with immune signaling and inflammation. Module 5: Endopeptidase activity Associated with proteolysis and extracellular protein turnover. Module 6: Ethanol oxidation Associated with metabolic enzyme activity. These findings suggest that vascular aging involves interconnected changes in: Vascular signaling + Immune regulation + Lipid metabolism + Extracellular matrix remodeling The functional networks were derived from enrichment and protein interaction analyses rather than direct pathway perturbation experiments. 🤖 7. Machine learning improves classification of the operational vascular aging phenotype Researchers developed LightGBM models using nested cross-validation. Three approaches were compared. Protein-only model AUC = 0.778 95% CI: 0.774–0.783 Framingham Risk Score alone AUC = 0.847 95% CI: 0.844–0.851 FRS + protein panel AUC = 0.911 95% CI: 0.909–0.914 Adding proteins improved classification performance substantially. The five highest-ranked protein features were: • PLAT • NTproBNP • DCN • GIP • REN The integrated model also reclassified: 6,260 participants Net reclassification improvement: 14.79% 95% CI: 14.0–15.6% However, an important methodological limitation must be emphasized. FRS contributed directly to the definition of vascular aging. Therefore, the strong AUC and reclassification improvement partly reflect overlap between the predictor and the outcome definition. These findings should not be interpreted as proof of clinical utility. 📊 8. Predicting arterial stiffness independently of the FRS-based phenotype To address this issue, researchers evaluated continuous ASI in 14,831 participants. Model performance: FRS alone R² = 0.063 Protein panel alone R² = 0.065 FRS + proteins R² = 0.083 Adding proteins increased R² by: 0.0207 95% CI: 0.0149–0.0269 RMSE decreased by: 0.0349 m/s These results demonstrate that the protein panel provides additional information for predicting arterial stiffness. However, the improvement was modest. The findings support further biomarker development rather than immediate clinical implementation. 🫀 9. The proteomic vascular aging score predicts future cardiovascular disease Researchers evaluated whether the protein-derived vascular aging score was associated with subsequent disease. Participants were divided into: • Low-risk tertile • Intermediate-risk tertile • High-risk tertile Over approximately 15 years, higher scores were associated with greater cumulative incidence of six vascular outcomes: Coronary artery disease Atrial fibrillation Ischemic stroke Peripheral artery disease Heart failure Atherosclerosis All six Kaplan–Meier comparisons showed: P < 0.001 The survival curves in Figure 3 demonstrated progressively greater disease incidence across increasing score tertiles. These results support the prospective relevance of the proteomic signature. However, the study did not establish that the score improves clinical decision-making beyond established risk prediction tools. 🧪 10. Mendelian randomization prioritizes 11 candidate proteins To identify proteins with genetic evidence relevant to arterial stiffness, researchers used: MR-SPI — Mendelian Randomization with Selection and Permutation Inference The analysis incorporated protein quantitative trait loci (pQTLs) and genetic associations with ASI. Eleven proteins were prioritized. Genetically predicted higher protein levels associated with lower ASI: • IGFBP2 • LAYN • NTproBNP • NPPB Genetically predicted higher protein levels associated with higher ASI: • ADH1B • ERBB2 • DCXR • FAM20A • SMAD5 • ADH4 • CCL26 The genetic instruments had F-statistics exceeding 10. These results provide genetic evidence supporting the relevance of selected proteins to arterial stiffness. However, Mendelian randomization depends on instrumental-variable assumptions, and the results do not definitively establish causal effects on vascular aging. 🧬 11. Candidate genes are enriched in vascular smooth muscle and calcification pathways Tissue enrichment analysis showed that the 11 prioritized genes were most strongly enriched in: Vascular smooth muscle tissue Phenotype enrichment highlighted: Vascular calcification This is biologically relevant because vascular aging is associated with: • Increased arterial stiffness • Extracellular matrix remodeling • Vascular smooth muscle phenotypic changes • Calcification • Endothelial dysfunction The findings suggest that the prioritized proteins may participate in vascular remodeling processes. However, enrichment analyses identify statistical associations with known biological annotations rather than directly demonstrating mechanisms. 🧠 12. AlphaFold 3 provides exploratory structural context Researchers used AlphaFold 3 to compare predicted protein structures associated with selected missense pQTLs. One example involved: ADH1B rs2066702 The variant was associated with an amino acid substitution: R370S Arginine → Serine Predicted structures showed local conformational differences near the substituted residue. A second example involved: DCXR rs61746217 The analysis similarly compared predicted structures corresponding to different alleles. These findings provide structural hypotheses for future mechanistic studies. However, the investigators explicitly cautioned that static AlphaFold models do not establish altered protein activity, stability, dynamics, or physiological function. 🩸 13. Mouse aortic tissue and human serum support selected candidates The investigators evaluated the prioritized proteins using two experimental systems. Mouse aortic tissue Young mice: 2 months Aged mice: 24 months n = 3 per group Western blotting demonstrated: Increased with aging: • LAYN • IGFBP2 Decreased with aging: • DCXR • FAM20A • ERBB2 • SMAD5 Human serum The validation cohort included: • Vascular aging: n = 20 • Non-vascular aging: n = 20 ELISA identified significant differences in several proteins. Across the two experimental approaches: Nine of the eleven genetically prioritized proteins showed differential expression in at least one validation system. LAYN, IGFBP2, and ERBB2 were significantly different in both mouse aortic tissue and human serum. However, the human groups differed in cardiovascular risk burden, particularly hypertension. Therefore, serum differences cannot be attributed exclusively to intrinsic vascular aging. 🌍 14. Primate single-cell RNA sequencing reveals cell-specific expression patterns The investigators reanalyzed arterial single-cell RNA-seq data from: 16 cynomolgus monkeys including: 8 aged animals The dataset identified six major vascular cell populations. Nine of the eleven candidate genes showed cell-type-associated expression patterns. Several changes were concentrated in: Endothelial cells and Pericytes In aged primate pericytes: ERBB2 ↓ SMAD5 ↓ DCXR ↓ In aged endothelial cells: LAYN ↑ FAM20A ↓ IGFBP2 ↓ Interestingly, IGFBP2 increased in aged pericytes while decreasing in endothelial cells. This demonstrates that vascular aging-associated protein regulation may differ substantially across cell types. 🛡️ 15. IGFBP2 protects against senescence-associated phenotypes in two vascular cell types IGFBP2 emerged as one of the most compelling candidates. Researchers used siRNA to reduce IGFBP2 expression in: • Human umbilical vein endothelial cells (HUVECs) • Human cerebral pericytes Cells were subsequently exposed to hydrogen peroxide to induce senescence-associated stress. IGFBP2 knockdown increased: CDKN1A/p21 ↑ CDKN2A/p16 ↑ Senescence-associated β-galactosidase activity ↑ These effects were observed in both endothelial cells and pericytes. The results support a protective role for IGFBP2 in the experimental regulation of vascular cell senescence. IGFBP2 is also relevant to metabolic disease because previous studies have associated it with insulin sensitivity and arterial stiffness. However, the present study did not directly test insulin signaling or establish that IGFBP2 protects vascular cells through improved metabolic regulation. 🧬 16. LAYN may preserve endothelial homeostasis LAYN encodes layilin, a hyaluronic acid-binding receptor associated with cell–matrix interactions. Researchers knocked down LAYN in HUVECs before hydrogen peroxide exposure. The intervention increased: CDKN1A/p21 CDKN2A/p16 SA-β-gal-positive cells These findings suggest that LAYN may restrain endothelial senescence-associated responses. Potential biological connections include: • Cell–matrix adhesion • Endothelial barrier regulation • Cytoskeletal organization • Mechanotransduction • Responses to extracellular matrix stress However, the study did not directly demonstrate which of these mechanisms mediates the observed effects. 🔥 17. FAM20A, ERBB2, and SMAD5 show opposite functional effects Three additional candidates exhibited potentially pro-senescence effects in the tested cell models. FAM20A In endothelial cells: FAM20A knockdown reduced: • CDKN1A • CDKN2A • SA-β-gal activity ERBB2 In pericytes: ERBB2 knockdown reduced senescence-associated markers. SMAD5 In pericytes: SMAD5 knockdown similarly attenuated senescence-associated phenotypes. These results suggest that the proteins may participate in pathways promoting stress-associated cellular senescence. Potential biological contexts include: FAM20A: Extracellular protein regulation and matrix homeostasis. ERBB2: Growth factor receptor signaling. SMAD5: BMP–SMAD-associated transcriptional signaling. However, the experiments used acute siRNA knockdown followed by hydrogen peroxide exposure. They do not establish that inhibiting these proteins would prevent vascular aging in vivo. ⚠️ 18. ERBB2 reveals why biomarker direction and cellular function cannot be equated ERBB2 exhibited an especially complex pattern. In the UK Biobank analysis: Higher circulating ERBB2 → Greater odds of vascular aging In Mendelian randomization: Higher genetically predicted ERBB2 → Greater ASI However: ERBB2 protein decreased in aged mouse aortas. Circulating ERBB2 decreased in the human serum validation cohort. ERBB2 expression decreased in aged primate pericytes. Furthermore: ERBB2 knockdown reduced senescence-associated phenotypes in pericytes. These findings may appear contradictory. However, circulating protein abundance and membrane receptor signaling are not necessarily equivalent. Potential explanations include: • Cell-type-specific expression • Receptor shedding • Ligand-dependent signaling • Compensatory regulation • Differences between circulating and tissue-localized protein pools • Disease-stage-dependent effects The study did not resolve these possibilities experimentally. The important lesson is that a circulating biomarker cannot automatically be interpreted as a direct therapeutic target. 🫀 19. NTproBNP also exhibits context-dependent associations NTproBNP showed another important example of biological complexity. In the overall UK Biobank analysis: Higher NTproBNP was associated with lower odds of the operational vascular aging phenotype. However, in participants aged 60 years or younger, the association was positive. The human serum validation cohort also exhibited higher NTproBNP in the vascular aging group. Meanwhile, Mendelian randomization associated genetically predicted higher NTproBNP and NPPB with lower ASI. These results highlight the distinction between: Circulating NTproBNP as a marker of cardiac wall stress and Natriuretic peptide signaling as a potentially protective cardiovascular pathway. The findings should not be interpreted as evidence that NTproBNP is uniformly protective or harmful. Its association with vascular aging may depend on age, hemodynamic burden, cardiovascular risk, and biological context. 🎯 20. Integrated model of vascular aging The study supports a multilevel framework. Population-level vascular aging Age + Blood pressure + Cardiovascular risk factors ↓ Adverse vascular phenotype ↓ Changes in circulating plasma proteins Proteomic signature 44 associated proteins ↓ VEGF signaling + Growth factor regulation + Immune–metabolic pathways + Extracellular matrix remodeling Genetic prioritization 11 proteins associated with arterial stiffness through MR-SPI ↓ Candidate biological regulators Vascular cell-specific mechanisms Endothelial cells + Pericytes ↓ IGFBP2 / LAYN-associated protective effects versus FAM20A / ERBB2 / SMAD5-associated pro-senescence effects under tested conditions Potential consequences Endothelial dysfunction + Pericyte dysregulation + Altered matrix homeostasis ↓ Vascular remodeling and arterial stiffness This is an integrated mechanistic hypothesis supported by multiple evidence layers. It is not a fully validated causal pathway. 💊 21. Implications for vascular aging and cardiovascular therapeutics The findings suggest several future directions. Proteomic vascular age assessment Plasma proteins may complement conventional cardiovascular risk factors in characterizing vascular health. IGFBP2-based research Investigating whether IGFBP2 signaling preserves endothelial and pericyte function could reveal new protective mechanisms. LAYN-associated mechanotransduction Understanding layilin-dependent cell–matrix signaling may provide insight into vascular resilience. SMAD5 and vascular calcification Cell-specific BMP–SMAD signaling may represent a candidate pathway linking vascular remodeling and senescence. ERBB2 signaling The divergent circulating and cellular findings highlight the need for tissue-specific mechanistic studies. Precision vascular aging Combining proteomics, genetics, and cell-type-specific functional assays may help distinguish biomarkers from potential disease mediators. However, no candidate protein was validated as a safe or effective therapeutic target in humans. 🚨 22. Critical limitations • The primary vascular aging phenotype was constructed from HVA and ePWV rather than directly measured biological vascular age. • HVA incorporates Framingham cardiovascular risk factors. • ePWV is derived from age and blood pressure. • Consequently, the primary classification model contains definitional overlap. • The integrated model's high AUC does not establish independent clinical diagnostic utility. • Continuous ASI prediction improved only modestly after adding proteins. • All machine-learning models underwent internal validation within UK Biobank; no independent external cohort was used. • The cohort predominantly represented individuals of European ancestry. • Plasma protein levels do not necessarily reflect vascular tissue expression or protein activity. • Mendelian randomization was performed using ASI rather than the primary vascular aging phenotype. • Horizontal pleiotropy and other instrumental-variable violations cannot be fully excluded. • AlphaFold 3 structural comparisons were exploratory. • Mouse aortic validation included only three animals per age group. • Human serum validation involved only 20 participants per group. • The human validation groups differed in cardiovascular risk burden. • Primate single-cell analyses were based on previously published data. • Functional experiments used hydrogen peroxide-induced senescence models. • Each candidate was evaluated using a single siRNA sequence. • Independent siRNAs and rescue experiments were not performed. • No in vivo genetic intervention demonstrated that the candidate proteins modify vascular aging. • No clinical intervention evaluated these proteins as therapeutic targets. • The study did not establish whether modifying the proteomic signature reduces cardiovascular events. 💡 Key takeaway Vascular aging can be characterized through circulating protein signatures, but identifying clinically useful biomarkers requires separating cardiovascular risk prediction from intrinsic vascular biology. This study analyzed nearly 3,000 plasma proteins in more than 42,000 participants and identified 44 proteins associated with an operational vascular aging phenotype. A combined proteomic–Framingham model achieved an internally validated AUC of 0.911, although its performance was partly influenced by overlap with the phenotype definition. Genetic analyses prioritized 11 candidate proteins, while cross-species and functional experiments highlighted five proteins associated with endothelial or pericyte senescence. The most promising mechanistic findings were: IGFBP2 knockdown worsened senescence-associated phenotypes in endothelial cells and pericytes. LAYN knockdown worsened endothelial senescence-associated phenotypes. FAM20A, ERBB2, and SMAD5 knockdown attenuated senescence-associated phenotypes in specific vascular cell models. The broader conceptual advance is: Plasma proteomics → Genetic prioritization → Cell-type-specific functional validation → Candidate vascular aging mechanisms The next step is rigorous external validation and in vivo mechanistic testing to determine which proteins are merely associated with vascular aging and which directly regulate vascular dysfunction. 📄 Liu H, Li H, Zhang W, et al. Plasma proteomics characterizes vascular aging and prioritizes functionally supported candidate proteins. npj Aging (2026) Article in Press. Accepted: October 6, 2026. DOI: doi.org/10.1038/s41514-026-0… Proteomic cohort: UK Biobank Pharma Proteomics Project Primate arterial scRNA-seq: GEO GSE117715 Analysis code: zenodo.org/records/22131690 Research institutions: Beijing Tiantan Hospital, Capital Medical University; Changhai Hospital, Naval Medical University; Fudan University, China. #VascularAging #AgingBiology #PlasmaProteomics #CardiovascularDisease #IGFBP2 #LAYN #FAM20A #ERBB2 #SMAD5 #EndothelialCells #Pericytes #CellularSenescence #ArterialStiffness #MendelianRandomization #MachineLearning #PrecisionMedicine #npjAging
3
12
671