Weekly reads 13/07/26

From neural memory to microbial manipulation: how sex, epigenetics, and bacteria redefine resilience and disease

Jul 19, 2026

This week’s papers uncover fascinating intersections between memory, resilience, and disease: from sex-specific neural mechanisms shaping pancreatic tissue recovery to epigenetic strategies targeting leukemia. Ferreira et al. dig deep into the differences between male and female mice after pancreatic injury. Males hang onto long-term plasticity through CGRP-dependent signaling, but females shut it down using neutrophils. This sharp distinction helps explain why cancer risk varies between the sexes. Amos and team shift the focus to acute leukemia; they pinpoint TRIM28 as a key chromatin adaptor that leukemia cells rely on. Blocking TRIM28, either genetically or with a targeted small molecule, pushes these cells to turn into terminally differentiated, neutrophil-like cells, cutting their cancerous potential. On the tech side, Cohen’s et al. introduce RT&T-AMP-MERFISH. The method lets scientists image the entire transcriptome at isoform-level detail within intact tissues. The payoff? A whole new view of molecular complexity in the brain. Shifting gears, Xie et al.’s research shows how Streptococcus anginosus, a gastric bacterium, manipulates host epigenetics to keep PD-L1 stable and dodge immune defenses. At the same time, Wen and colleagues find that antibiotics can actually drive bacteria to cooperate, boosting persister cell survival through horizontal protein transfer. But there’s a cautionary note from Ishaque et al.: cell segmentation is still a huge hurdle in spatial transcriptomics. Get it wrong, and you risk flawed data and misleading downstream results.

Preprints/articles that I managed to read this week

Sex-Dimorphic Neural Memory Shapes Pancreatic Tissue Resilience

Ferreira et al., bioRxiv (2026). 10.64898/2026.06.15.732370

The paper in one sentence

This study reveals that male mice retain long-term neural and epithelial memory after pancreatic injury via CGRP-dependent sensory neuron activation, while female mice suppress this neural memory through neutrophil-mediated inhibition, highlighting sex-specific mechanisms of tissue resilience and cancer susceptibility.

Summary

Ferreira et al. investigate how pancreatic tissue retains a “memory” of injury, enabling faster regeneration upon recurrent damage but also increasing susceptibility to cancer. Using mouse models of acute pancreatitis (AP), they demonstrate that male mice exhibit sustained epithelial plasticity and neural activation long after tissue repair, driven by a CGRP-dependent neuron–epithelial axis. This neural memory is maintained by sensory neurons and supports long-term epithelial progenitor-like states. In contrast, female mice show neutrophil-mediated suppression of neural activation during injury, decoupling neural memory from epithelial plasticity. The study identifies IL-6 and G-CSF as key immune-derived drivers of sensory neuron activation in males, while female neutrophils actively inhibit this process. Mechanistically, CGRP stimulation of epithelial cells reduces chromatin accessibility at acinar identity genes (e.g., Ptf1a, Bhlha15), promoting a less differentiated, more plastic state. Disrupting sensory neuron activity in males abrogates tissue memory, demonstrating its necessity for sustained epithelial plasticity.

Personal highlights

Sex-dimorphic neural memory: Males retain long-term sensory neuron activation (CGRP+) and epithelial plasticity after pancreatitis, while females do not, due to immune-mediated suppression.
CGRP as a key mediator: Sensory neuron-derived CGRP directly promotes epithelial plasticity by reducing acinar cell lineage commitment, as shown by ATAC-seq and RNA-seq analyses.
Neutrophils as gatekeepers: Female-specific neutrophil infiltration during AP actively suppresses sensory neuron activation, preventing neural memory formation.
Neural-epithelial axis: A bidirectional relationship exists where neural memory sustains epithelial progenitor states, and epithelial plasticity is dependent on ongoing sensory input.
Therapeutic implications: Targeting neural memory (e.g., via CGRP inhibition or sensory neuron modulation) could disrupt pathological tissue plasticity, potentially reducing cancer predisposition in high-risk individuals.

Why should we care?

This work challenges the traditional view of tissue resilience as an intrinsic, cell-autonomous property of epithelial cells. Instead, it demonstrates that tissue memory is a coordinated, niche-dependent phenomenon, where sensory neurons and immune cells play active roles in sustaining or suppressing regenerative plasticity. The sex dimorphism observed, with males retaining neural memory and females suppressing it, may explain known epidemiological differences in pancreatic cancer incidence, which is higher in men. Critically, the study suggests that neural memory is not just a bystander but a driver of long-term tissue adaptation, with implications for both regeneration and oncogenesis.

Epigenetic Reactivation of Lineage Differentiation to Target Leukemia

Amos et al., bioRxiv (2026). 10.64898/2026.07.08.737260

The paper in one sentence

This study identifies TRIM28 as a chromatin adaptor and selective dependency in acute leukemia, whose inhibition, genetically or pharmacologically, drives terminal neutrophil-like differentiation of leukemia cells, reducing their proliferative and leukemic potential.

Summary

Using focused CRISPR screening in mouse and human models of acute leukemia, the authors discovered that TRIM28 (a multi-domain chromatin scaffold) is essential for leukemia cell survival. TRIM28 suppresses neutrophil differentiation programs via non-canonical, H3K9me3-independent mechanisms, instead collaborating with PRC2 to enforce H3K27me3-mediated repression at key neutrophil transcription factor loci (e.g., CEBPA, GFI1, SPI1). Genetic knockdown of TRIM28 induces irreversible, functional neutrophil-like differentiation in leukemia cells, with reduced leukemic potential in vitro and in vivo. To translate these findings, the team developed KI-T28-03, a first-in-class small-molecule inhibitor targeting the TRIM28 PHD-bromodomain. KI-T28-03 phenocopies genetic TRIM28 loss, impairs leukemia proliferation (IC₅₀ ~5 µM in AML cells), and synergizes with Menin-MLL inhibition to enhance anti-leukemia effects. Proteomics and dependency map analyses reveal that TRIM28 assembles leukemia-specific chromatin complexes, distinct from its roles in non-malignant cells, highlighting its context-dependent function.

Personal highlights

TRIM28 as a leukemia-specific vulnerability: CRISPR screens and DepMap analyses identify TRIM28 as a selective dependency in acute myeloid and lymphoid leukemias, with high expression correlating with poor patient survival, positioning it as a potential therapeutic target.
Non-canonical repression of neutrophil differentiation: TRIM28 represses neutrophil gene programs independently of its canonical SETDB1-H3K9me3 axis, instead cooperating with PRC2 (EZH2) to deposit H3K27me3 at promoters of neutrophil transcription factors, revealing a novel mechanism of lineage suppression in leukemia.
Terminal differentiation as a therapeutic strategy: TRIM28 loss drives leukemia cells into irreversible, functionally mature neutrophil-like states (e.g., oxidative burst, NET formation), a therapeutically desirable outcome given the short lifespan of neutrophils and their rapid clearance.
First-in-class TRIM28 inhibitor: KI-T28-03 selectively binds the TRIM28 PHD-bromodomain, stabilizes the protein, and recapitulates genetic knockdown effects, including anti-proliferative activity and neutrophil differentiation in AML cells, with synergy when combined with Menin inhibitors.
Context-dependent chromatin scaffolding: TRIM28 assembles distinct protein complexes in leukemia vs. non-leukemia cells, shifting from RNA biogenesis roles to repression of myeloid differentiation, underscoring its adaptability as a chromatin adaptor.

Why should we care?

This work expands the therapeutic landscape for epigenetic therapies in leukemia by targeting a non-enzymatic chromatin adaptor (TRIM28) rather than classical epigenetic enzymes. The study demonstrates that forcing leukemia cells to differentiate into short-lived neutrophils, a strategy previously limited to specific subtypes like APL, could be broadly applicable, offering a potential route to reduce relapse and improve outcomes in acute leukemias, which often develop resistance to current treatments. By demonstrating that differentiation therapy can be extended beyond genetically defined cases, this work suggests a paradigm shift: targeting lineage plasticity itself, rather than just proliferation, could be a viable strategy for treating aggressive cancers.

Whole-transcriptome-scale isoform-resolved spatial imaging of single cells in tissues

Cohen et al., Cell (2026). 10.1016/j.cell.2026.06.027

The paper in one sentence

RT&T-AMP-MERFISH integrates in situ RNA amplification with multiplexed error-robust fluorescence in situ hybridization (MERFISH) to achieve whole-transcriptome-scale, isoform-resolved spatial transcriptomics of single cells in intact mouse brain tissue.

Summary

This study introduces RT&T-AMP-MERFISH, a novel method combining in situ RNA amplification (RT&T-AMP) with MERFISH to image ~23,000 genes and ~10,000 isoforms across ~130,000 single cells in mouse brain sections. The approach enables systematic analysis of spatial gene programs, cell-cell communications via ligand-receptor interactions, and reveals widespread region- and cell-type-specific isoform usage. Notably, brain structures like the choroid plexus and hippocampus exhibit particularly rich isoform diversity. The method reduces probe requirements and costs by nearly an order of magnitude compared to unamplified MERFISH, while maintaining high spatial resolution and detection efficiency.

Personal highlights

First whole-transcriptome, isoform-resolved spatial imaging in intact tissues: RT&T-AMP-MERFISH detects ~33,000 distinct RNAs (23,000 genes + 10,000 isoforms) simultaneously, achieving transcriptome-wide coverage with isoform resolution for the first time in spatial transcriptomics.
Spatial gene programs and cell-cell communication mapped: The method identifies gene modules associated with anatomical structures and reveals ligand-receptor interactions (e.g., Notch1 signaling) with distinct spatial patterns across brain regions.
Widespread isoform diversity uncovered: Many genes exhibit region- and cell-type-specific isoform usage, with structures like the choroid plexus and hippocampus showing particularly rich isoform specificity, suggesting specialized functional roles.
Technical advancement in probe efficiency: RT&T-AMP reduces the number of encoding probes needed per transcript by nearly an order of magnitude, lowering costs and enabling detection of short sequences and isoforms differing by minimal sequence variations.
Strong cross-validation with existing datasets: Results show high correlation with bulk RNA-seq data and spatial patterns from the Allen Brain Atlas, confirming the method’s accuracy and reliability.

Why should we care?

This work represents a major technical leap in imaging-based spatial transcriptomics, moving beyond gene-level analysis to capture the full complexity of RNA isoforms, within their native tissue context. The discovery that isoform usage varies dramatically across brain regions and cell types reveals a previously underappreciated layer of molecular specialization that likely underpins cellular identity, signaling, and disease mechanisms. However, the study has important limitations: it is currently restricted to mouse brain sections, a subset of isoforms (~10,000 out of ~43,000 annotated), and a limited number of tissue slices, so whole-brain or cross-species applications remain to be demonstrated. Additionally, while the method’s reduced cost and probe requirements make it promising for broader adoption, the biological significance of many observed isoform patterns still requires functional validation.

A gastric microbial chromatin remodeler drives gastric cancer progression and immune evasion by reprogramming the host epigenome

Xie et al. bioRxiv (2025). 10.64898/2025.12.26.696631

The paper in one sentence

Streptococcus anginosus promotes gastric cancer progression and immune evasion by delivering a bacterial chromatin remodeler (saSNF2) via extracellular vesicles, which reprograms host cell transcription to stabilize PD-L1 and suppress anti-tumor immunity.

Summary

This study identifies Streptococcus anginosus (SA) as a tumor-resident bacterium enriched in gastric cancer (GC) tissues, particularly in advanced and metastatic disease. Using mouse models and in vitro experiments, the authors demonstrate that SA accelerates tumor growth and reshapes the tumor immune microenvironment. Mechanistically, SA secretes extracellular vesicles (saEVs) containing the bacterial chromatin remodeler saSNF2, which is internalized by gastric cancer cells via dynamin-dependent endocytosis. Inside host cells, saSNF2 partners with the transcription factor TEAD1, through its ATPase activity and integration into the host BAF chromatin-remodeling complex, to upregulate the palmitoyltransferase ZDHHC11. This enzyme stabilizes PD-L1 via palmitoylation, reducing its degradation and enhancing immune evasion by suppressing CD8⁺ T-cell infiltration and activity. The study further shows that pharmacological inhibition of ZDHHC11 (using 2-BP) reverses PD-L1 stabilization, restores CD8⁺ T-cell function, and, when combined with anti-PD-1 therapy, synergistically suppresses tumor growth in preclinical models. Clinical data from 139 patients suggest that SA and saSNF2 expression correlate with poor prognosis in GC, supporting the translational potential of these findings.

Personal highlights

SA as a tumor-resident oncobacterium: Streptococcus anginosus is enriched in gastric cancer tissues, particularly in advanced stages, and promotes tumor progression in mouse models.
Cross-kingdom epigenetic reprogramming: SA-derived extracellular vesicles deliver the bacterial chromatin remodeler saSNF2 into host cells, where it modulates host gene transcription.
Novel PD-L1 stabilization mechanism: saSNF2 cooperates with TEAD1 to upregulate ZDHHC11, which palmitoylates PD-L1 at Cys272, preventing its ubiquitination and degradation.
BAF complex modulation: saSNF2 integrates into host BAF chromatin-remodeling complexes via its ATPase activity, amplifying oncogenic transcription (e.g., of AXL, CTGF, CYR61).
Therapeutic synergy: Inhibiting ZDHHC11 (with 2-BP) reverses immune evasion and synergizes with anti-PD-1 therapy in mouse models, suggesting a potential combination strategy for SA-positive gastric cancers.

Why should we care?

This work provides a compelling example of how the microbiome can actively drive cancer progression through direct epigenetic reprogramming of host cells, rather than merely responding to tumor-associated changes. The discovery of the saSNF2-ZDHHC11-PD-L1 axis offers a mechanistic link between microbial colonization, immune evasion, and tumor growth. While the findings are robust in preclinical models, their clinical translation will require validation in human trials. The complexity of targeting microbial-derived factors (e.g., saSNF2) and the potential off-target effects of palmitoylation inhibitors remain open questions. Nevertheless, this work highlights the therapeutic promise of disrupting microbial-host epigenetic crosstalk in cancer.

Antibiotics stimulate protein transfer to persister cells

Wen et al., Science (2026). 10.1126/science.adx3972

The paper in one sentence

Antibiotic stress induces bacteria to differentiate into vesicle-producing donors and protein-receiving recipients, enabling horizontal protein transfer that enhances the survival of metabolically dormant persister cells.

Summary

This study provides direct evidence for antibiotic-induced horizontal protein transfer (HPT) between bacterial cells via membrane vesicles. Using a Cre-loxP genetic reporter system in Escherichia coli, the authors demonstrate that sub-inhibitory antibiotic concentrations (e.g., ciprofloxacin, carbenicillin) stimulate an isogenic bacterial population to split into two distinct states: donor cells that activate the phage shock protein (Psp) membrane stress response to release protein-containing vesicles, and recipient cells that suppress Psp activity to uptake proteins. Single-cell transcriptomics and genetic analyses reveal that protein-competent recipients express high levels of translation inhibitors (e.g., HipA), which promote antibiotic persistence by inducing metabolic dormancy via (p)ppGpp signaling. Protein uptake further enhances recipient survival, linking vesicle-mediated HPT to antibiotic tolerance. Notably, this transfer occurs not only within E. coli but also between different Gram-negative species, suggesting a conserved mechanism of bacterial cooperation under stress.

Personal highlights

Direct detection of horizontal protein transfer: A Cre-loxP reporter system enables sensitive and specific tracking of protein transfer events between bacterial cells, ruling out DNA or mRNA as the transferred material.
Antibiotic-induced cell state differentiation: Sub-minimum inhibitory concentrations (sub-MICs) of antibiotics trigger bacteria to adopt either a vesicle-producing (donor) or protein-uptake (recipient) state within a genetically identical population.
Psp response drives vesicle formation: The conserved Psp membrane stress response, including the ESCRT-III-like protein PspA, promotes the production of transfer-competent vesicles in donors, with vesicle integrity and lipid composition (e.g., enriched phosphatidylglycerol) supporting efficient transfer.
HipA links protein uptake to persistence: Recipient cells suppress Psp activity through HipA, a kinase that promotes persistence via (p)ppGpp-mediated metabolic shutdown, thereby enhancing protein uptake and survival under lethal antibiotic doses.
Cross-species relevance: Protein transfer extends beyond E. coli to other Gram-negative bacteria (e.g., Pseudomonas putida), indicating that HPT may be a widespread survival strategy in microbial communities.

Why should we care?

The key takeaway of this work is that antibiotics may inadvertently fuel bacterial resilience by promoting this division of labor, where persister cells, already difficult to kill due to their metabolic inactivity, gain additional support from their active neighbors. This mechanism could explain why some infections recur after seemingly successful treatment. While the study opens avenues for new therapeutic strategies (e.g., disrupting vesicle-mediated transfer or targeting the Psp/HipA pathways), its clinical relevance remains to be fully established. The findings also underscore the complexity of bacterial survival strategies, where cooperation plays a critical yet underappreciated role in antibiotic tolerance.

The Challenge of Cell Segmentation in Spatially Resolved Transcriptomics

Ishaque et al. (2026). The Challenge of Cell Segmentation in Spatially Resolved Transcriptomics. arxiv.org/abs/2606.09675

The paper in one sentence

This perspective paper reframes cell segmentation as a critical, unresolved bottleneck in spatially resolved transcriptomics (SRT), arguing that current methods are inadequate for the field’s unique technical challenges and that errors in this step can systematically bias downstream biological interpretations.

Summary

Spatially resolved transcriptomics (SRT) enables the measurement of gene expression within the spatial context of tissues, but its potential is hindered by a lack of robust methodological guidance for one of its most fundamental steps: cell segmentation. The authors identify key technical challenges, such as sparse molecular signals, transcript displacement (due to diffusion or optical artifacts), complex 3D cellular morphologies projected onto 2D imaging planes, and the absence of universal membrane markers—that make segmentation particularly difficult in SRT compared to traditional imaging. These challenges lead to errors that propagate through downstream analyses, distorting cell-type annotations, clustering, differential expression, cell-cell interaction inference, and spatial organization studies. The paper reviews current approaches, including transcript-agnostic methods (e.g., Cellpose, StarDist, U-Net) that rely on morphological cues from staining, and transcript-informed methods (e.g., Baysor, BIDCell, Segger) that integrate transcript-level data to refine cell boundaries. However, the authors emphasize that no single method consistently outperforms across diverse tissues and technologies. They also highlight the lack of gold-standard benchmarks, appropriate evaluation metrics, and standardized reporting practices, which hinder reproducibility and progress. DDThe authors propose a community-driven path forward, calling for shared evaluation frameworks, scalable benchmark datasets, and transparent reporting standards to address segmentation’s reproducibility crisis and unlock SRT’s full potential.

Personal highlights

Segmentation as a core unresolved problem: The paper elevates cell segmentation from a routine preprocessing step to a central, unresolved challenge in SRT, demonstrating how errors can create reproducible but misleading molecular patterns (e.g., artificial cell states, spurious cell-cell interactions).
Unique technical challenges: Details SRT-specific obstacles, including transcript diffusion, optical artifacts, 3D-to-2D projection ambiguities, and background contamination, which complicate accurate cell boundary delineation and transcript assignment.
Downstream impact of errors: Shows how segmentation inaccuracies can systematically bias high-plex analyses, such as clustering, differential expression, and ligand-receptor inference, leading to false biological conclusions.
Methodological landscape and gaps: Surveys existing solutions—from classical image analysis to AI-based and transcript-informed approaches—while noting their limitations and the absence of universal benchmarks for fair comparison.
Call for community action: Advocates for coordinated efforts to develop gold-standard datasets, interpretable metrics, and scalable infrastructure to ensure robustness and reproducibility in SRT segmentation.

Other papers that peeked my interest and were added to the purgatory of my “to read” pile

Thanks for reading.

Cheers,

Seb.

Sebcentrism

Discussion about this post

Ready for more?