iiinova — where your digital self keeps evolving

Karsten Rippe @karsten-rippe.bsky.social · 09/09/2026

🙏 9/ Take-home: heterochromatin is a mosaic of switchable nanodomains. HP1α on clustered H3K9me3 is a local barrier that weak activators fail and strong ones pass. Kudos to Robin Weinmann and the whole ChromNet team. Funded by @dfgpublic.bsky.social SPP2191, DKFZ-MOST, @hlsalliance.bsky.social.

Model scheme. Top, a silenced wild-type chromocenter with H3K9me3 nucleosomes and bound HP1α; VP16 is blocked, p65 and VPR activate it while HP1α and H3K9me3 stay. Bottom, a Suv39h dn chromocenter without HP1α, which p65 and VPR activate with acetylation and VP16 activates without it.

040

Karsten Rippe @karsten-rippe.bsky.social · 09/09/2026

🌈 8/ Three-color dSTORM makes the nanodomains visible: dCas9-VPR, nascent MSR RNA and HP1α imaged together. Transcribed nanodomains and HP1α-bound silenced nanodomains sit side by side within the same decondensed chromocenter. No chromocenter-wide (liquid-liquid) phase separation switch.

Widefield and three-color dSTORM images of a VPR-activated iMEF nucleus, dCas9 in cyan, MSR RNA in yellow, HP1α in magenta, with one zoomed region below. In the zoom the RNA cluster and the HP1α clusters occupy adjacent, largely non-overlapping areas.

140

Karsten Rippe @karsten-rippe.bsky.social · 09/09/2026

💻 7/ We tested three models: a global barrier, single nucleosomes switching alone, and nanodomains of ~5 nucleosomes. Only the nanodomain model fits. It shows how a sparse mark protects: H3K9me3 sits on 38% of nucleosomes, but 91% of 5-nucleosome units carry at least one and thus bind HP1α.

Schematic of the three models, global barrier, single nucleosome (k = 1) and nanodomain (k = 5), each shown for a condensed and a decondensed chromocenter with nucleosomes, H3K9me3 nucleosomes, HP1 and MSR RNA.

110

Karsten Rippe @karsten-rippe.bsky.social · 09/09/2026

💭 6/ Remove the barrier and weak activators get through. In Suv39h dn cells without H3K9me3 and HP1α at chromocenters, VP16 and p65 decondense and activate far more cells. VP16 does so without gaining H3K27ac or p300: chromatin can open without acetylation, possibly driven by the nascent RNA.

Left, violin plots of nuclear area occupied by chromocenters for GFP, VP16, p65 and VPR in wild-type (gray) and Suv39h dn (red) cells; VP16 and p65 decondense more in Suv39h dn cells, GFP does not. Right, H3K27ac signal at chromocenters versus decondensation per cell: VPR and p65 cells are decondensed and acetylated, VP16 cells decondense without acetylation.

120

Karsten Rippe @karsten-rippe.bsky.social · 09/09/2026

🚫 5/ Bound HP1α still represses, but only locally. At a reporter array in U2OS cells with HP1α recruited next to the activator, it silences VP16 almost completely, attenuates p65 and barely touches VPR. At 4 to 10 kb from the promoter the repression is gone.

Reporter RNA levels for VP16, p65 and VPR with the NLS control, KRAB or HP1α co-recruited. HP1α reduces VP16 about seven-fold, p65 about three-fold and VPR less than two-fold.

110

Karsten Rippe @karsten-rippe.bsky.social · 09/09/2026

🔍 4/ dSTORM reveals the structure: a condensed chromocenter is a sponge of interconnected MSR clusters. On activation, it disperses into 40-70 nm clusters linked by strands, and HP1α keeps its granular 40-70 nm distribution in both states. FRAP shows that its binding kinetics barely change.

Widefield and dSTORM images of dCas9 (cyan) and HP1α (magenta) in a condensed and a decondensed chromocenter, each with two zoomed regions. The condensed chromocenter is a sponge-like network of MSR clusters; the decondensed one breaks into small clusters linked by strands. HP1α forms granular clusters in both states.

110

Karsten Rippe @karsten-rippe.bsky.social · 09/09/2026

💥 3/ The surprise: even when VPR fully decondenses chromocenters and drives transcription, HP1α and H3K9me3 stay put. Textbook logic says the repressive marks must go before activation. Here, both persist for the 24 to 36 h we followed, on the same chromocenters that are being transcribed.

Confocal images with zooms of a condensed and a decondensed chromocenter, rows DNA, dCas9, H3K9me3, HP1α. H3K9me3 and HP1α remain enriched on the decondensed chromocenter chain, with the HP1α signal outlined by dashed lines. Scale bar: 5 µm.

110

Karsten Rippe @karsten-rippe.bsky.social · 09/09/2026

🔬 2/ We recruited activators of increasing strength (VP16 < p65 < VPR) with dCas9 to mouse major satellite repeats, which assemble into dense foci called chromocenters. Decondensation, H3K27ac, p300, and MSR transcription all scale with activator strength. VP16 does nothing in wild-type cells.

Confocal images of wild-type iMEF nuclei, columns ctrl, VP16, p65 repressed, p65 active, VPR; rows DNA, dCas9, H3K27ac and MSR RNA. Chromocenters decondense, gain H3K27ac and produce RNA from p65 onward; VP16 nuclei look like the control. Scale bar 5 µm.

110

Karsten Rippe @karsten-rippe.bsky.social · 09/09/2026

🧬 1/ HP1α marks silenced heterochromatin, but how does it repress, and what happens when transcription is switched on? On mouse fibroblast chromocenters, HP1α stays bound, its barrier is local, and the domain reorganizes nanodomain by nanodomain. 👉 @cp-molcell.bsky.social doi.org/10.1016/j.mo...

Schematic of the two experimental systems. Left, activators recruited to chromocenters of wild-type and Suv39h dn mouse fibroblasts, with DAPI and anti-HP1α images showing HP1α enrichment and loss. Right, the U2OS 2-6-3 reporter array with an activator recruited alone or together with HP1α.

12810

Karsten Rippe @karsten-rippe.bsky.social · 20/05/2026

The calculation is something like this 100 runs/year x 4 slides x 2 samples per slide = 800 samples. Sample number obviously will depend on size (500 mm2 per slide).

110

Karsten Rippe @karsten-rippe.bsky.social · 19/05/2026

For our analysis of cancer tissues it will be worth it. But if you do not need full transcriptome, e.g., only need a few hundred genes, other approaches will be less expensive.

000

Karsten Rippe @karsten-rippe.bsky.social · 19/05/2026

Not sure whether you can grow cells. I have not seen data but I would expect that the z-stack range is similar to Xenium. It will start with the full transcriptome panel of 18,000 genes and you have to ask 10x Genomics about their plans for the rollout of other probe sets.

110

Karsten Rippe @karsten-rippe.bsky.social · 19/05/2026

DM should work. Our runs were done at 10x Genomics in California. It is imaging-based.

000

Karsten Rippe @karsten-rippe.bsky.social · 18/05/2026

Sure, as long as you do not expect tons of answers 😉

110

Karsten Rippe @karsten-rippe.bsky.social · 14/05/2026

6/ This strand-segregated topology constrains the ALT mechanism. The t-loop resolution at the PML shell could generate a C-circle template, driving rolling-circle amplification into G-rich repeats inside. Mechanistic ALT models can now be tested against the actual APB structure. Read the preprint!

040

Karsten Rippe @karsten-rippe.bsky.social · 14/05/2026

5/ Mapping all 4 components on a single radial axis (r=1 , shell midpoint) shows that ssC-rich and ssG-rich repeats occupy distinct compartments in an APB: the ssC-rich repeats overlapped with the PML shell, TRF1-marked telomeres occupy the interior, and ssG-rich repeats spread broadly throughout.

120

Karsten Rippe @karsten-rippe.bsky.social · 14/05/2026

4/ The ssC-rich repeats were enriched within the PML shell, while ssG-rich repeats were abundant in the APB interior. The ssG-rich signal comprised mostly DNA as it was RNase-resistant and proximal to POT1 (peak distance ~17 nm).

120

Karsten Rippe @karsten-rippe.bsky.social · 14/05/2026

3/ The images revealed that PML in an APB is not a filled droplet. It rather assembles into a partially open ~70 nm-thick spherical shell with an uneven protein distribution and a lower density than canonical PML bodies. This shell is the scaffold and coordinate system for everything inside.

100

Karsten Rippe @karsten-rippe.bsky.social · 14/05/2026

2/ 3D MINFLUX nanoscopy with multiplexed exchange DNA-PAINT conducted at the @bioquant.bsky.social imaging platform (tinyurl.com/2auwmbfu) localized individual PML and TRF1 proteins together with ssG-rich and ssC-rich telomere repeats at ~3 nm precision.

110

Karsten Rippe @karsten-rippe.bsky.social · 14/05/2026

1/ Some cancers can proliferate without telomerase. They recombine and extend their telomeres inside ALT-associated PML bodies (APBs). @emmakoeleman.bsky.social @dkfz.bsky.social mapped where proteins and single-stranded telomeric repeats sit in APBs: doi.org/10.64898/202... 🧵

1133

Karsten Rippe @karsten-rippe.bsky.social · 20/04/2026

8/ Enjoyed two exciting weeks with the teams of Philipp Mallm and Rene Jackstadt, diving into the Atera data, and we are looking forward to more of that. Our spatial transcriptomics work is supported by SATURN3/BMFTR and the MULTI-SPACE platform @mspacealliance.bsky.social.

020

Karsten Rippe @karsten-rippe.bsky.social · 20/04/2026

7/ Adding immune cells to the picture: fetal-like cancer stem cells and CD4+ memory T cells co-localize at the invasive front while CD8+ effectors are excluded. Stem cell plasticity coupled to immune evasion. Everything visualized on one tissue section.

140

Karsten Rippe @karsten-rippe.bsky.social · 20/04/2026

6/ The two stem cell poles sit in different compartments. Normal LGR5+ stem cells: crypt bases. Fetal-like cancer stem cells (IRC high): invasive front. At least five stem cell types could be resolved in total. This was not possible in our previous Xenium analysis.

130

Karsten Rippe @karsten-rippe.bsky.social · 20/04/2026

5/ Ezgi Sen, a great PhD student in the Rippe and Stegle labs, led the downstream analysis: Atera clearly resolved 15 main cell clusters on one CRC section. Non-malignant epithelium splits into absorptive and secretory lineages with LGR5+ stem cells at the crypt bases.

151

Karsten Rippe @karsten-rippe.bsky.social · 20/04/2026

4/ Building on Rene Jackstadt's CRC biology expertise, the interplay between two stemness poles is mapped: LGR5+ cancer stem cells expand the tumor. Fetal-like cancer stem cells are chemo-resistant, seed metastases and interconvert. Detecting them on FFPE was out of reach for us.

120

Karsten Rippe @karsten-rippe.bsky.social · 20/04/2026

3/ In our work in the SATURN3 consortium on tumor heterogeneity saturn3.org, a crucial question is emerging: how can we resolve the different normal and cancer stem cell types together with immune cells at single-cell resolution?

130

Karsten Rippe @karsten-rippe.bsky.social · 20/04/2026

2/ Atera maps ~18,000 genes in single cells on FFPE tissue with 2-3x higher sensitivity and a broader dynamic range than Xenium 5k in our hands. Having the whole transcriptome at this coverage matters when you are looking for expression signatures and low-abundance markers.

130

Karsten Rippe @karsten-rippe.bsky.social · 20/04/2026

1/ Philipp Mallm from the @dkfz.bsky.social Single-cell Open Lab just broke the news at #AACR26: the first results on colorectal cancer (CRC) tissue analysis with the new Atera spatial transcriptomics platform from @10xgenomics.bsky.social. Is it "making the impossible possible?" 🧵

32610

Karsten Rippe @karsten-rippe.bsky.social · 24/12/2025

8/ 🤘 For those about to rock (around the Christmas tree), we salute you 🎸🎄: Isabelle Seufert & our team from @dkfz.bsky.social & BioQuant & @akispapantonis.bsky.social, Philipp Mallm, @s-anders.bsky.social & Petros Kolovos labs with support from @spp2202.bsky.social & @mspacealliance.bsky.social.

132

Karsten Rippe @karsten-rippe.bsky.social · 24/12/2025

7/ 🌟 The broad applicability of our RWireX framework was shown in different cellular systems. In human blood cells, cell-type-specific regulatory programs could be distinguished, e.g., the DCs at the CD22 and the TBX21 loci had highest activity in B cells and mature CD8+ T cells, respectively.

110

Karsten Rippe @karsten-rippe.bsky.social · 24/12/2025

6/ 🎯 Our AC/DC framework reveals how cells achieve precise gene control. TRGs with TF promoter binding show fastest induction. DCs enable coordinated burst-size increases, while ACs allow fine-tuned gene-specific regulation—working together to support complex programs.

231

Karsten Rippe @karsten-rippe.bsky.social · 24/12/2025

5/ ⛰️ DCs are larger contiguous regions of simultaneously accessible chromatin sites. These domains exhibit locally increased transcription factor binding activity and may represent the genomic footprints of TF hubs.

120

Karsten Rippe @karsten-rippe.bsky.social · 24/12/2025

4/ 🔗 ACs denote interactions between separated regulatory sites at ATAC peaks. We observed differences between rare and frequently occurring ACs, suggesting distinct roles of stochastic vs. architectural links.

120

Karsten Rippe @karsten-rippe.bsky.social · 24/12/2025

3/ 💻 Using our new RWireX software (github.com/RippeLab/RWi...), we distinguish two regulatory chromatin modules: ACs (autonomous links of co-accessibility between separated sites of 1 kb size) and DCs (domains of contiguous co-accessibility of ~200 kb) that both can co-exist.

120

Karsten Rippe @karsten-rippe.bsky.social · 24/12/2025

2/ 🔍 We investigated the proinflammatory response to TNF in primary human cells using #scRNA and #scATAC sequencing and fluorescence microscopy. We identified ~1,500 TNF-regulated genes (TRGs) that often clustered in the genome and exhibited induced co-expression.

130

Karsten Rippe @karsten-rippe.bsky.social · 24/12/2025

1/ 🎄 We got our Christmas present today: "Two distinct chromatin modules regulate proinflammatory gene expression" is now published @natcellbio.nature.com doi.org/10.1038/s415.... Our study introduces a scATAC-seq-based framework for genome-wide analysis of gene regulation features.

37725

Karsten Rippe @karsten-rippe.bsky.social · 21/08/2025

Happy to report that the Xenium system for spatial transcriptomics in the Single Cell Open Lab www.dkfz.de/en/single-ce... @dkfz.bsky.social was the first one worldwide to pass the mark of 100 runs according to @10xgenomics.bsky.social. And it keeps on running with another one, so stay tuned...

dkfz.de

Single-cell Open Lab - German Cancer Research Center

041

Karsten Rippe @karsten-rippe.bsky.social · 13/08/2025

These features allow it to predict immunotherapy response and point to potential universal T cell based therapies. Great multi-group effort led by Mirco Friedrich together with @michael-platten.bsky.social, @raabms.bsky.social, @stefaneichmueller.bsky.social, and many more collaborators.

020

Karsten Rippe @karsten-rippe.bsky.social · 13/08/2025

Glad that the preprint on identifying and characterizing rare tumor-reactive T cells in multiple myeloma & AML patients is out now doi.org/10.1101/2025.... Single-cell TCR and transcriptome profiling & immunopeptidomics revealed conserved programs and shared non-canonical antigens.

130

Karsten Rippe @karsten-rippe.bsky.social · 11/07/2025

Liquid-liquid phase separation of HP1? Not in our mouse fibroblasts… This paper by the Fabian Erdel group offers a fresh perspective on how this can be reconciled with seemingly conflicting reports: The phase separation propensity of HP1 decreases from yeast → fly → mouse. doi.org/10.1038/s414...

13313

Karsten Rippe @karsten-rippe.bsky.social · 20/06/2025

3/ Great joint effort with Philipp Mallm and the Single Cell Open Lab @dkfz.bsky.social, led by Anne Rademacher, Alik Huseynov, and Michele Bortolomeazzi, with key contributions from our colleagues at KiTZ Heidelberg, supported by the @mspacealliance.bsky.social program of @hlsalliance.bsky.social.

020

Karsten Rippe @karsten-rippe.bsky.social · 20/06/2025

2/ Using medulloblastoma cryosections, we analyzed sensitivity, specificity, and spatial signal patterns to distinguish background from rare-cell signals. We also show how reimaging slides improves cell segmentation—and enables additional transcript and protein readouts from the same section.

110

Karsten Rippe @karsten-rippe.bsky.social · 20/06/2025

1/ Spatial transcriptomics technologies are rapidly evolving. We compared five methods—Visium, RNAscope HiPlex, Molecular Cartography, Merscope, and Xenium—and developed technology-agnostic quality control metrics to guide their use: doi.org/10.1186/s130...

doi.org

Comparison of spatial transcriptomics technologies using tumor cryosections - Genome Biology

Background Spatial transcriptomics technologies are revolutionizing our understanding of intra-tumor heterogeneity and the tumor microenvironment by revealing single-cell molecular profiles within the...

1163

Karsten Rippe @karsten-rippe.bsky.social · 19/06/2025

How do transcription compartments form — and does phase separation drive gene expression? I enjoyed discussing these questions with @akispapantonis.bsky.social a lot, and we put our thoughts together for @naturerevgenet.bsky.social, now out at rdcu.be/erP1u

15314

Karsten Rippe @karsten-rippe.bsky.social · 12/04/2025

From enhancer hijacking to ecDNA and onco-condensates – we review how changes in nuclear architecture drive oncogenic gene expression by disrupting enhancer–promoter communication. doi.org/10.1002/ijc..... Kudos to Isabelle Seufert, @claire-vrgs.bsky.social and Sina Wille 👏

1238

Karsten Rippe @karsten-rippe.bsky.social · 31/03/2025

“We all rely on science.” U.S. scientists are speaking out for scientific integrity and academic freedom. What they’re doing matters. We owe them our solidarity and a clear stand beside them. www.nytimes.com/2025/03/31/s...

nytimes.com

Trump Administration Has Begun a War on Science, Researchers Say (Gift Article)

Nearly 2,000 scientists urged that Congress restore funding to federal agencies decimated by recent cuts.

1136

Karsten Rippe @karsten-rippe.bsky.social · 25/01/2025

A mechanism for maintaining and spreading H3K9me3 in heterochromatin from the Fejes Toth and Aravin labs that depends on the local H3K9me3 density: HP1 dimers recruit SetDB1 to chromatin by simultaneously binding H3K9me3 on histone H3 and auto-methylated SetDB1. www.biorxiv.org/content/10.1...

0388

Karsten Rippe @karsten-rippe.bsky.social · 22/01/2025

Fakten sind das Rückgrat unserer Demokratie Facts are the backbone of our democracy When facts are treated as opinions and replaced by rhetoric or propaganda, meaningful discourse suffers, threatening democracy and leaving it vulnerable to manipulation. www.openpetition.de/!ychtb

openpetition.de

Fakten sind das Rückgrat unserer Demokratie - Online-Petition

Liebe Mitbürgerinnen und Mitbürger, als Wissenschaftlerinnen und Wissenschaftler arbeiten wir täglich daran, die Grundlagen für fundierte Entscheidungen und Innovationen zu schaffen – basierend auf üb...

040