Autumn is coming. Hopefully bringing changes for the better.

As luck would have it, my PhD advisor Mike Greenberg happened to be in Barcelona at the same time as me.

Like a science-y version of a Whit Stillman movie, we chatted about engrams and enhancers while dining over tapas and strolling through the city.

Was fun celebrating the recent Nobel with many of the pioneers in optogenetics at Optogen 2026 in beautiful Barcelona — a great place to reunite with old friends.

The conference brought together wide range of approaches, from optical methods to new indicators to clinical trials

Celebration at Stanford Bioengineering for Karl’s Nobel Prize. His commitment to teaching came up, and Karl said something very interesting: that one of the best feelings is finishing a classroom presentation and realizing you had taught students something new. Completely agree!

Photos from this morning's impromptu celebration taken by my postdoc Daesun Song (I missed it unfortunately)

The name badge is a nice touch. "I'm Karl Deisseroth and it's okay for me to be here"

Sunset over Stanford

Third reviewer just commented on variability and the degree of sleep promotion by activating these cells

Peer review file: media.springernature.com/original/sprin…

Another reviewer

But I don't see a necessity experiment asked at all in the referee evaluations. Instead we get suggestions to measure correlations. Correlation does not equal causation.

One reviewer:

Exciting to see how quickly the frontiers of neuroscience are advancing. With investigators Jung Ho Hyun, TK Kim, Dohoung Kim, Changchun Ryu, Jihye Seong, Seong-Eon Roh, Sang Jeong Kim, and Yulong Li

Photos from the journey

Had the privilege of attending K-BRAIN and the IBS engram satellite symposium in 대전/大田/Daejeon, South Korea

Presented unpublished results on local control of gene editing in the brain and voltage imaging with ASAP6c/7y — opposite ends of the temporal spectrum that we work on

My office Sunday morning

Yes that’s tomorrow for some of us!

Today’s office

However, oftentimes, discussions about crosstalk forget that wavelength orthogonality is not necessary when different cells express opsin and voltage indicator, and they can be excited in their separate locations by 2-photon microscopy, as done in our current study.

In contrast, when performing field stimulation near an ASAP5-expressing neuron, eliciting electrical transients in multiple spines at the same time, then somatic depolarization can be detected.

There's a lot of interest in performing combined optogenetic stimulation (via opsins) and voltage imaging, i.e. all-optical input and output.

Victor did that here, too, exciting axonal ChRmine to elicit ASAP5 events in nearby dendritic spines. Again there were no somatic events

But, unlike our finding in vitro, the new work shows that synaptic ASAP5 events in vivo do not result in somatic depolarization measurable by ASAP5 or e-phys.

A likely explanation is inhibitory neurotransmission (GABA-induced Cl influx) attenuating passive electrical propagation

This finding extends unitary synaptic event detection by ASAP5 beyond our previous demonstration in cultured neurons (e.g. the below movie from the 2024 Neuron paper describing ASAP5, cell.com/neuron/fulltex…) to the natural situation in vivo.

First, Victor established that ASAP5 can detect unitary events in sensory cortex pyramidal neurons. Whisker stimulation increases ASAP5 event frequency (e) but not amplitude (f) in spines, consistent with the known binary nature of synaptic vesicle release per bouton (0 or 1)

I'll let the results speak for themselves. Conveniently they are all listed in the abstract

We and Promega together discovered and characterized fluorofurimazine as an in vivo substrate for NanoLuc, reporting it in 2020 in Nature Methods. We have been cited >200 times.

But if you search google scholar for fluorofurimazine, the top hit is a 2025 paper with 6 citations

@KarlDeisseroth @smoolee85 @yusterafa @lisa_giocomo All made possible by @StanfordBrain
@NIH_NINDS NIMH NIGMS @NIH_Innovates and @USBrainAlliance (#BRAINInitiative) supporting "new technologies to explore how the brain’s cells and circuits interact at the speed of thought"

ASAP7y also enables fiber-optic and whole-brain mesoscale recordings of electrical oscillations in neuronal populations in behaving mice. Thanks to the Schnitzer Lab and @KarlDeisseroth!
12/14

In mammalian systems, ASAP7y enables single-trial subthreshold recordings of APs and dendritic SPs with 1035-nm 2P excitation, using scanless holography and SLAP2 microscopy. Thanks to @PodgorskiLab and @EmilianiLab!
11/14

Pairing ASAP7y recordings with analysis of the EM connectome of 717 neuron types in the fly brain, we show how branching geometry acts as a physical determinant of neuronal calculations. In some cases, branches can even function like independent parallel circuits.
10/14

ASAP7y overcomes this limit. It expands its excitation spectrum to utilize 1030–1035 nm lasers while achieving a 2x boost in subthreshold responsivity over ASAP5 (>2% per mV near rest). This enables SP visualiation along fine processes of the smallest neurons even in vivo.
9/14

But what if you need to look deeper into the tissue, run ultrafast parallelized scanning, or track millivolt-scale subthreshold dynamics? That's where the negatively tuned ASAP7y comes in.
7/14
biorxiv.org/content/10.648…

We even stacked ASAP6c against the state-of-the-art calcium indicator jGCaMP8f in fast-spiking cortical interneurons. While calcium sensors missed the majority of rapid, isolated spikes, ASAP6c detected individual APs cleanly in a binary, all-or-none fashion.
6/14

ASAP6c's high contrast is especially beneficial for 1-photon imaging, as you can now see events above the background blur of out-of-focused neurons. This allows high-throughput recordings of APs across hundreds of densely labeled neurons using simple CMOS cameras.
6/14

ASAP6c's 100% ΔF/F is 3x larger than earlier positive ASAPs. This means you can achieve the same signal-to-noise ratio with 9x less light (SNR is proportional to √photons). That means you can record an order-of-magnitude more neurons at the same power density as before.
5/14

Let's start with ASAP6c, a quantum leap for high-throughput spike detection. Screened from >2000 site-directed mutants, ASAP6c (c = contrast) yields a massive ~100% fluorescence increase in response to single APs.
4/14
biorxiv.org/content/10.110…

To crack neural circuits, we need to map input-output computations of neurons in real time. But a big challenge stands in the way: recording the subthreshold potentials (SPs) that are the inputs, and the action potentials (APs) that are the outputs, across spatial scales.
2/14

Excited to announce a powerful new one-two punch for voltage imaging from our lab and collaborators! In two new preprints, we introduce ASAP6c for high-throughput population spike-recording, and ASAP7yfor deep, subthreshold 2P imaging.
🧵 1/14

Some details about the position

The Lin Lab is hiring!

We're seeking motivated postdocs for projects on voltage imaging and optogenetics in the brain.

If you're excited about using new methods to undersatand and correct brain diseases, working together with other neuroscience labs at Stanford, drop us a line!

@Coalesce_x Great question; the mutagenesis template is a bicistronic FLuc-p2a-NanoLuc construc. Mutations are placed only into NanoLuc, and assays are done for both NanoLuc and FLuc. The FLuc thus serves as a RNA abundance control, capturing any upstream differences (tfn, txn, stability)

And workflow with MIDAS:

Importantly, MIDAS is highly compatible with automation and robotics. Because the workflow is simply multiwell PCR + parallel transfer, it scales easily

That opens the door to automated protein engineering pipelines that are actually efficient

Compare: Workflow proposed in 2025

And because it was easy (well, at least for the amazing Pengli) we did it on two more substrates, finding some universal determinants of activity but also some specificity determinants. The maps are all different, quite interestingly.

We were immediately able to find an improved variant of NanoLuc on its original substrate Fz, which we didn't expect as it had been highly optimized already on Fz using classical methods.

And we also learn how each amino acid does — no need to throw away that information!

In MIDAS each variant costs just an oligo and a PCR reaction, so testing 20 is just as easy as 1.

With MIDAS, we tested all 20 amino acids at 25 active site positions in the enzyme NanoLuc.

Many want to do something similar for their protein, but didn't want to make 500 clones.

Random mutation rates are kept low to preserve overall function, so double or triple hits within the same codon are exceedingly unlikely.

You can reach saturation only by mutagenic primers directed to a site, either pooled then picked and verified, or cloned one-by-one.

MIDAS (“Microbe-Independent Deep Assembly and Screening”) deletes most of that workflow

MIDAS directly assembles complete genes by PCR and transfects the PCR products into mammalian cells

Yes you can do that!

And you can reduce a week's work into 1 day

embopress.org/doi/full/10.10…

Existing protein engineering workflows are very complex if you need to test in mammalian cells:

Plasmid assembly → transformation → colony picking → plasmid prep → verification → transfection → assay.

For 100s or 1000s of variants, this is painfully slow/expensive, e.g.

@MarkusCovertLab To test if forced separation can be therapeutic, we applied DARPin binders -/+ a bulky attachment to cells with different EGFR mutants. Sure enough, DARPins with a bulky attachment were more effective in suppressing oligomerization, ERK activity, and cell proliferation.

@MarkusCovertLab Is there a causal relationship though? To test this, Daesun first fused a non-interacting domain BICD2N to the EC side of mutant (oncogenic) EGFR chains. This reduced baseline oligomerization to wild-type levels and prevented activation by EGF.

@MarkusCovertLab So far not surprising, but efficacy of allosteric inhibitors, which overcome resistance to afatinib and osi, also correlated with suppression of EGFR oligomerization and ERK activation. Thus EGFR monomerization correlates well with efficacy.

@MarkusCovertLab He then saw that EGFR oligomerization and ERK activation correlate with drug resistance; mutations conferring drug-resistance for cancer growth (such as T790M on erlotinib, or C797S on afatinib and osimertinib), it also allows continued EGFR oligomerization and ERK activation.

@MarkusCovertLab Daesun first saw that hyperphosphorylating mutations L858R (a common early oncogenic mutation) and T790M (which increases affinity to ATP substrate) promote oligomerization and ERK activation, reproducing past work. Interestingly ERK signaling is nearly maximized even without EGF

As EGFR oligomerization induces phosphorylation and downstream effector recruitment, Daesun asked if HomoFC can reveal effects of mutations and inhibitors on this step. At the same time, he used @MarkusCovertLab's ERK-KTR reporter to assess mutational effects on Ras signaling.

Postdoc Daesun Song had previously developed an elegant GFP-based reporter of homomeric interactions called HomoFC, in which a protein or interest carries two halves of GFP, arranged so they could not assemble in cis, but could assemble in trans.

onlinelibrary.wiley.com/doi/10.1002/an…

Mutations in the receptor tyrosine kinase EGFR drive lung cancer via hyperactivation of the Ras-ERK pathway, and treatment with EGFR kinase inhibitors erlotinib, afatinib, or osimertinib can suppress cancer growth. However, additional mutations soon arise to cause resistance.