Max Haase

No, DeepMind has not solved the protein folding problem.

#Alphafold predictions are valuable hypotheses and accelerate but do not replace experimental structure determination.

M18BP1 valency and a distributed interaction footprint determine epigenetic centromere specification in humans - The EMBO Journal The histone H3 variant CENP-A is considered an epigenetic landmark of centromeres. Its deposition reflects cell-cycle-regulated assembly of M18BP1, HJURP, and PLK1 on a divalent MIS18α/β scaffold. The...

It’s out! 🥳 Excited to share our new paper (with Kai Walstein, @andrea-musacchio.bsky.social and all others) on the role of M18BP1 in CENP-A loading!

“M18BP1 valency and a distributed interaction footprint determine epigenetic centromere specification in humans”
link.springer.com/article/10.1...

Two new papers from the lab published in The EMBO Journal!

link.springer.com/article/10.1...
By Kai Walstein, @louisa-hill.bsky.social and others – On role of M18BP1 in CENP-A loading

link.springer.com/article/10.1...
By Arianna Esposito Verza and others – On mechanism of activation of PLK1

Molecular requirements for PLK1 activation by T-loop phosphorylation - The EMBO Journal Activation of PLK1, a master mitotic kinase, requires phosphorylation of its activation segment on Thr210, within a basic consensus sequence for Aurora kinases. Aurora B-dependent phosphorylation of T...

Arianna Esposito Verza, @andrea-musacchio.bsky.social et al describe the long-sought-after mechanism of Bora dependency of PLK1 activation by Aurora A kinase during mitotic entry
Another #RefereedPreprint ℅ @reviewcommons.org
link.springer.com/article/10.1...

Discovery of additional ancient genome duplications in yeasts Whole-genome duplication (WGD) has had profound macroevolutionary impacts on diverse lineages,1,2 preceding adaptive radiations in vertebrates,3,4,5 t…

Excited to announce our latest publication in reporting evidence for three (3!) new whole genome duplications (WGDs) in yeasts. Scientists have often wondered why WGD is so rare in fungi, it turns out we may just not have been looking hard enough! 🧪 🍄 🧬
www.sciencedirect.com/science/arti...

We warmly welcome our new independent @maxplanck.de Group Leader Andrija Sente. His group will investigate the molecular mechanisms of neuronal communication.

Learn more about Andrija and his work and click on the link🔗
www.mpi-dortmund.mpg.de/news/new-max...

LetA defines a structurally distinct transporter family - Nature The distinct architecture of the Escherichia coli membrane transporter LetA mediates lipid trafficking across the bacterial envelope in partnership with the tunnel-like complex LetB.

Excited to share a new paper led by Cristina Santarossa (Bhabha & Ekiert groups) on LetA, a bacterial phospholipid transporter that defines a structurally distinct transporter family. Really enjoyed working with Cristina on their deep mutational scanning work.
www.nature.com/articles/s41...

Sex without crossing over in the yeast Saccharomycodes ludwigii - Genome Biology Background Intermixing of genomes through meiotic reassortment and recombination of homologous chromosomes is a unifying theme of sexual reproduction in eukaryotic organisms and is considered crucial ...

Very interesting study. You might also consider citing Saccharomycodes, which undergoes meiosis without crossovers, since it seems closely related to the topic here.
link.springer.com/article/10.1...

Quick intuition: a crowd is strongly correlated with a concert being loud. But it’s not logical to conclude “the crowd drives the music volume”—the concert (or the same upstream event: “show started”) generates both.

the key, "RNAPII cluster formation depends on transcription initiation"

Absolutely thrilled to share the latest work from my lab focused on the variation and evolution of human centromeres among global populations! We assembled 2,110 human centromeres, identifying 226 new major haplotypes and 1,870 α-satellite HOR variants. www.biorxiv.org/content/10.6...

I've never understood why people want an ecosystem where AI writes papers for AI audiences. I think us humans have lost the plot.

Fellow yeast enthusiasts, does anyone have the genomic or amino-acid substitution for the cdc20-1 ts mutant - thanks!

Very cool to see Jana’s epic on point centromere evolution published. Congratulations to all!

Although, it makes one wonder how these strange centromeres evolved in the first place?

Heterochromatin epimutations impose mitochondrial dysfunction to confer antifungal resistance | The EMBO Journal imageimageHeterochromatin-island epimutations can provide resistance to caffeine and antifungal drugs in the fission yeast Schizosaccharomyces pombe. This study reveals that some epimutations cause re...

www.embopress.org/doi/full/10....

Our latest now online at EMBO Journal. Read if you are interested in how epimutations mediate antifungal resistance & how this might result in heteroresistance in human & cereal crop fungal pathogens. Big Thx & congrats to Andreas Fellas, Pin Tong & Alison Pidoux

we are considering an organelle with a phospholipid monolayer as a membraneless compartment now? If so, then I must concede to you good sir

Woof summary 🐕‍🦺

Hyman-style scaffold: main dog bed + visitors that mysteriously never change it

Musacchio: physics says visitors always change the pile

Assay: too toy-like, ignores the whole house

Conclusion: that brand of LLPS ≈ story first, physics later → pseudoscience

Put together, the argument is:

The definition of the magic bed (scaffold) is physically inconsistent

The assay that “finds” them ignores half the system (the real solvent)

Yet we now claim there are lots of these magic beds in cells

Musacchio’s complaint:

You can’t go from
“one dog curled up on a cushion in an empty room”

to

“this is a universal, self-organizing bed that runs all nap-piles in all houses.”

That’s a huge leap.

Problem: real cells are not tiny clean rooms.

The solvent (cytosol) = the entire noisy house:
other dogs, toys, furniture, kids, treats, mess.

The assay basically ignores the whole house and only studies:
one dog + one cushion in a white box.

“But wait, people found lots of scaffolds!” you say.

Musacchio’s answer: they used a too-simple test:

Take one purified “dog” (protein)

Put it alone in a tiny, clean room (buffer)

Watch it curl up into a little lump

Declare: “Aha! Self-piling magic bed!”

So, either:

no one really “binds” to the bed → it’s not a binding scaffold

or

they do bind → then the bed’s behavior must change

In this view, the official “scaffold that drives droplets but is unaffected by clients” is a unicorn dog bed. 🦄🛏️

In BOTH cases, bed + newcomers = new dog pile.

No “untouchable” magic bed.

Once there is a pile (“above C_sat”):

Now you already have a dog heap. More dogs + toys squeeze in.
➡️ The shape, tightness, and edges of the pile change.

Before there’s a pile (“below C_sat”):

Dogs are wandering. A few lie on the bed → fewer free dogs, different bumping, new connections.
➡️ That changes when a big cuddle-pile even starts.

Musacchio’s core claim:

If more dogs jump on the bed, the bed has to squish, bend, or change how the pile forms.

You can’t have:

a bed everyone piles onto,

and that never changes anything about the pile.

1/ 🐶💤

In Tony Hyman–style LLPS land:

Scaffold = the dog bed

Clients = extra dogs + toys that hop on

Rule of the game:

The bed decides where the nap-pile forms.
Other dogs can jump on, but somehow they don’t change how the bed behaves.

I don't deny that oil and vinegar is LLPS, and I would agree with you that LLPS is a real phenomenon.

I just find it absurd that it is used as an example to suggest that since we see the phenomenon in our salad bowls, then we probably see it in our cells.

How do new centromeres evolve while staying compatible with the division machinery?

Discover it in our new Nature paper! We show centromeres transition gradually via a mix of drift, selection, and sex, reaching new states that still work with the kinetochore.

👉 doi.org/10.1038/s41586-025-09779-1

just because your salad phase separates doesn't imply biogenesis of membraneless compartments by LLPS. This analogy needs to die. I would wager that the liquid behavior of membraneless compartments is just a spandrel of the underlying multivalent interactions, and in of itself, not so important.

If there are site specific multivalent interactions then what purpose is LLPS serving? Current models explain and account for multivalent interactions.