We're hiring at Xenium HR! Open roles in Tualatin, OR (hybrid) and remote:
🔹 Employee Relations Partner II
🔹 HR Business Partner II
🔹 Implementation Specialist
🔹 Employee Experience Rep
🔹 Benefits Specialist II
🔹 HR Account Rep
Apply today! ➡️na2.hubs.ly/H065-hg0
3/
NephroBase - the first kidney foundation AI model:
🔅 40 million cells, 1 billion parameters across human, pig, rat & mouse
🔅 Integrates snRNA, snATAC, XENIUM, OSMX spatial data
Not All Senescent β Cells Are Harmful: A Human Pancreas Atlas Reveals Adaptive and Maladaptive Senotypes
Cellular senescence is increasingly recognized as a driver of aging and age-related disease. Yet senescent cells are not a uniform population. A new human pancreas atlas now demonstrates that senescent β cells exist in at least two fundamentally different states—one adaptive and one maladaptive.
Using integrated single-cell RNA-seq, Xenium spatial transcriptomics, Visium spatial transcriptomics, CODEX spatial proteomics, secretome profiling, and functional insulin secretion assays, investigators analyzed pancreatic tissue from donors spanning ages 20–80 years.
The study identified two major senescent β-cell populations:
🟢 CDKN2A⁺ (p16⁺) senotype
🔴 CDKN1A⁺ (p21⁺) senotype
The distinction proved biologically important.
The CDKN2A⁺ senotype retained β-cell identity, preserved expression of key transcription factors such as MAFA and PDX1, maintained glucose responsiveness, and displayed a relatively low-inflammatory secretory phenotype. Functional studies showed that islets enriched for CDKN2A expression exhibited superior glucose-stimulated insulin secretion.
In contrast, the CDKN1A⁺ senotype exhibited many hallmarks of pathological aging.
These cells showed:
▪ Loss of insulin gene expression
▪ Reduced MAFA, PDX1, GK and ABCC8 expression
▪ Impaired glucose-stimulated insulin secretion
▪ Pro-inflammatory SASP programs
▪ Increased immune-cell infiltration within islets
Perhaps the most striking result came from direct functional testing.
Human islets with high CDKN1A expression failed to appropriately increase insulin secretion during both static and dynamic glucose stimulation assays. By contrast, CDKN2A-high islets remained functionally competent and even demonstrated enhanced secretory responses.
The spatial analyses revealed additional insights.
CDKN1A⁺ β cells preferentially localized near the acinar–islet border and were significantly closer to endothelial cells than non-senescent β cells. This raises the possibility of bidirectional signaling between vascular aging and β-cell senescence.
Aging was associated with progressive accumulation of both senotypes, but the consequences differed dramatically. As shown in the lifespan analyses, β-cell abundance, insulin content, and C-peptide levels declined with age, while CDKN1A and CDKN2A expression increased. However, only the CDKN1A program was consistently linked to loss of endocrine identity and dysfunction.
The implications extend beyond diabetes biology.
Current senolytic approaches often treat senescent cells as uniformly harmful. These findings suggest that indiscriminate elimination of all senescent β cells may be biologically misguided. Instead, future therapies may need to selectively target the maladaptive CDKN1A⁺ senotype while preserving or even supporting the adaptive CDKN2A⁺ population.
More broadly, the study provides compelling evidence that aging tissues are composed of distinct senescence programs with divergent physiological consequences—a concept likely relevant far beyond the endocrine pancreas.
Reference
Iwasaki K, Pan H, Dreyfuss J, et al. Distinct senescent β-cell senotypes differentially drive islet aging and dysfunction. bioRxiv (2026). DOI: 10.64898/2026.05.25.727705.
🎓 Position: Doctoral Fellow – Molecular Muscle Biology | 🇧🇪 PhD Opportunity at Ghent University
🏫 University: Ghent University
📍 Location: Ghent, Belgium 🇧🇪
🏢 Department: Movement and Sports Sciences
👨🏫 Supervisor: Prof. Wim Derave
⏰ Deadline: 30 June 2026
📅 Duration: Up to 4 years
💰 Salary: 100% net salary (AAP equivalent, tax-free fellowship)
Interested in understanding how muscles adapt to exercise at the molecular level?
This PhD project explores how skeletal muscle responds to exercise by studying transcriptional activity within and between muscle fibers. Using cutting-edge techniques like spatial transcriptomics (RNAscope & Xenium), you’ll investigate how different muscle fibers adapt—even when not directly activated.
The research combines animal and human studies, aiming to uncover fundamental mechanisms behind muscle plasticity, with implications for health, performance, and chronic disease prevention. You’ll collaborate with international experts in bioinformatics, molecular biology, and neurobiology across leading research groups.
👤 Ideal Candidate:
• Master’s in biomedical sciences, biology, bioinformatics, medicine, or related fields
• Strong interest in muscle physiology and molecular biology
• Experience with lab techniques (e.g., transcriptomics, immunohistochemistry)
• Basic programming/data analysis skills (R/Python)
• Independent, detail-oriented, and collaborative mindset
• Proficient in English
🌟 Why Apply:
• Work on high-impact research in exercise biology and health
• Access state-of-the-art facilities (genomics, imaging, bioinformatics)
• Join an internationally recognized research lab (Ghent Muscle Lab)
• Strong collaboration across disciplines and countries
• Structured PhD training and career development support
• Excellent work-life balance with 36 days leave benefits
🌍 About Ghent:
A vibrant student city in Belgium, Ghent offers a rich cultural scene, historic charm, and a strong international research environment—making it an ideal place to live and study.
🔗 More Info:
phdscanner.com/opportunities…#PhDOpportunity#BiomedicalResearch#MuscleBiology#PhDPositions#GhentUniversity#Belgium#LifeSciences#Bioinformatics#ResearchCareers
A recent independent study published in Nature evaluated imaging-based #spatialtranscriptomics platforms in inflammatory bowel disease (IBD) tissues, CosMx SMI demonstrated higher detection efficiency than Xenium across commercially available panels. 🔗 nature.com/articles/s41467-0…
🧫 PDAC Has Entered the Era of Integrated Tumor Microenvironment Atlases
Pancreatic cancer spatial biology is rapidly evolving from descriptive single-cell profiling into multi-modal ecosystem mapping capable of defining actionable therapeutic niches.
A new generation of integrated PDAC atlases is now combining:
• scRNA-seq
• Visium spatial transcriptomics
• Xenium high-resolution imaging
• bulk RNA deconvolution
• paired primary/metastatic spatial datasets
into unified tumor microenvironment frameworks. (PMC)
The scale is becoming enormous.
Recent PDAC atlas efforts collectively span:
🔹 >180,000–700,000 single cells
🔹 multi-cohort spatial transcriptomics
🔹 metastatic paired sampling
🔹 and increasingly subcellular-resolution imaging platforms.
But the most important shift is conceptual.
The field is moving away from:
“Which cell types exist?”
toward:
“Which spatially organized cellular programs drive progression, immune collapse, fibrosis, and therapy resistance?”
One emerging axis appears repeatedly across datasets:
🧱 POSTN fibroblasts
×
🧫 SPP1 macrophage programs
These stromal–myeloid ecosystems are consistently associated with:
• ECM remodeling
• EMT activation
• invasive phenotypes
• immune suppression
• and poor prognosis.
POSTN signaling in particular is becoming one of the dominant stromal drivers in PDAC biology. Integrated single-cell and spatial analyses show that POSTN-enriched fibroblasts interact with tumor cells through integrin signaling pathways including ITGAV/ITGB5, activating PI3K/AKT/β-catenin programs linked to aggressive disease. (IJBS)
At the opposite end of the spectrum, immune-active niches enriched for:
• CCL4 T cells
• plasma-cell programs
• antigen presentation signatures
appear associated with more immunologically permissive microenvironments.
This suggests that future PDAC therapy may require simultaneous remodeling of both stromal and immune ecotypes.
Not simply “kill tumor cells.”
But spatially re-engineer the ecosystem.
The implications are enormous.
With Xenium and MERFISH-scale imaging now approaching subcellular and junction-level resolution, PDAC atlases are evolving into functional maps for:
• theranostic imaging
• spatial biomarker selection
• resistance prediction
• and ecotype-guided trial design.
PDAC research is no longer entering the atlas era.
It is entering the ecosystem engineering era.
#PDAC#SpatialTranscriptomics#SingleCell#TumorMicroenvironment#Xenium#MERFISH#CancerAtlas
The era of static tumor atlases is ending.
We are entering the 11th dimension of oncology: Non-Invasive cfDNA-based Spatial Ecotypes. 🩸🔬
For years, decoding the tumor microenvironment (TME) required invasive spatial profiling platforms like Xenium, CosMx, Visium, and Stereo-seq. These technologies transformed cancer biology, but they remained expensive, tissue-dependent, and difficult to repeat longitudinally.
Now, two landmark studies have fundamentally shifted the field.
A massive Stanford/Harvard consortium study published in Nature (2026; DOI: 10.1038/s41586-026-10452-4) integrated over 10 million spatial transcriptomic spots and single cells across carcinomas and melanomas. Using machine learning, the team identified 9 conserved Spatial Ecotypes (SEs)—multicellular “neighborhoods” defined by distinct cellular compositions, signaling programs, and spatial topology.
These SEs directly correlated with:
• overall survival (OS)
• progression-free survival (PFS)
• immune checkpoint inhibitor (ICI) response
• invasive tumor fronts and immune niches
But the real breakthrough came next.
Each spatial ecotype carried a unique DNA methylation signature. Using deep learning on plasma cfDNA, investigators reconstructed the tumor’s spatial ecosystem directly from blood. In melanoma patients, a simple liquid biopsy predicted immunotherapy response by recovering SE composition non-invasively.
At nearly the same time, the Wang Lab’s Nature Cancer 2025 “TabulaTIME Pan-Cancer” atlas (DOI: 10.1038/s43018-025-01039-5) established the pan-cancer structural backbone:
• 4.48 million integrated cells
• 36 cancer types
• 103 harmonized studies
• conserved immunosuppressive barriers including $CTHRC1^ $ CAFs and $SLPI^ $ macrophages
Together, these studies redefine spatial oncology:
Tissue atlas → Pan-cancer ecotypes → cfDNA spatial recovery
We are no longer limited to watching tumors through a single biopsy snapshot.
The tumor microenvironment can now be serially monitored through blood.
The future of spatial biology is not only on glass slides.
It is circulating through our veins. 🩸✨
#SpatialBiology#LiquidBiopsy#SingleCell#CancerResearch#Oncology#Bioinformatics#MachineLearning
Excited to present the first pre-print from our group, an investigation the human bone marrow microenvironments in patients with myelodysplastic syndromes (MDS) and normal age-matched subjects using Xenium genotype-informed spatial transcriptomics:
biorxiv.org/content/10.64898…
Phoenix may be a major inflection point for spatial biology
A new bioRxiv preprint introduces Phoenix, a generative AI system that predicts single-cell spatial transcriptomics directly from routine H&E pathology slides — potentially transforming how we study cancer ecosystems at population scale.
The scale is striking.
Phoenix was trained on:
22.2 million cell-image/expression pairs
16 organ systems
79 Xenium slides 924 tissue microarray cores
10,000 GPU hours on a pre-exascale supercomputer
using high-resolution Xenium spatial datasets rather than lower-resolution Visium data.
The key advance is not simply “predicting genes from images.”
Many previous methods failed to generalize across:
institutions,
tissues,
stain variability,
unseen cohorts,
and batch effects.
Phoenix instead demonstrates:
zero-shot prediction,
transfer across unseen organs,
pan-cancer spatial inference,
and clinically relevant biomarker discovery.
Importantly, this is not just benchmark engineering.
The authors used Phoenix to recover meaningful biology across multiple systems.
Examples include:
breast cancer subtype ecosystems,
colorectal tumor budding evolution,
pancreatic PanIN progression,
immune-fibroblast spatial interactions,
chemotherapy/radiotherapy response,
sarcoma immune remodeling,
and KRAS-driven spatial states in mouse PDAC.
One of the most interesting concepts is the emergence of pan-cancer spatial ecotypes.
Across 9,544 TCGA patients, Phoenix identified recurrent multicellular ecosystems:
EC1 → dysfunctional inflammatory state
EC2 → immune competent state
EC3 → fibrotic/remodeling state
with distinct prognostic implications.
The EC3 “stromal remodeling” ecosystem is especially notable because it echoes growing evidence across:
fibrosis,
aging,
CAF biology,
and chronic wound repair
that maladaptive stromal niches may actively sustain disease progression.
Another major implication:
Phoenix enables virtual spatial transcriptomics on archived pathology collections.
Instead of running expensive Xenium/Visium workflows on thousands of samples, existing H&E archives could potentially become:
spatial atlases,
biomarker discovery engines,
therapy-response maps,
and longitudinal ecosystem datasets.
The paper is also a reminder that the future of pathology may not be “AI replaces pathology,” but rather:
AI converts morphology into systems biology.
There are still important limitations:
dependence on Xenium-quality training data,
restricted gene panels,
uncertainty across platforms,
and the need for prospective clinical validation.
But conceptually, this feels important.
Spatial omics may be transitioning from a boutique assay into a scalable computational layer on top of routine histology.
bioRxiv DOI: 10.64898/2026.04.25.720812