Induced Phosphorylation
Part 1: Induced Phosphorylation as the next frontier in druggability
This post is part of a whole chapter of “Thought Experiments” revisiting how we think about target “druggability”. One of the druggability concepts I’m most excited about is “Induced Phosphorylation”. I had a lot to say about it, so I decided to split the post into two parts.
In PART 1 (here), we’ll cover:
An overview of protein post-translational modifications (focused on phosphorylation)
A thought experiment: Programming induced phosphorylation at scale
If any of this is well-known to you, feel free to skip and start at the appropriate section.
In PART 2, we’ll cover an idea that I think should become a drug discovery company1:
An overview of ✨my favorite protein✨ (Yes-Associated Protein, or YAP)
Another thought experiment: Programmable Induced Phosphorylation of YAP
1. Preamble: “Thanks, it has pockets!”
Whenever someone compliments me on a dress I’m wearing, I reflexively respond:
“Thanks, it has pockets! 💁♀️”
Am I super awkward? Sure 😅 . But there’s something deeply satisfying about having something functional and valuable. I imagine that’s how medicinal chemists feel when they are presented with a protein that has a druggable pocket: "[Thank goodness], it has pockets! 🤩" You can slip in a molecule, toggle a function and - with the right therapeutic window and thoughtful Target Product Profile - maybe even bring to market an effective treatment with blockbuster potential. Your bosses will love you, even if you’re producing a line of the “same dresses” every year for decades.
Over two decades ago, scientists tried to draw a line in the sand: which parts of the human genome were druggable, and which were not? The “druggable genome” concept, popularized in the early 2000s, narrowed our collective focus to proteins that were (1) small, (2) well-structured, and (3) carried a neat little pocket for a small molecule to bind. That framework gave rise to an explosion of enzyme inhibitors, GPCR drugs, ion channel modulators, and more, but it also left a lot of biology on the sartorial cutting room floor.2
Any biochemist will tell you that the world of proteins is far messier than the tidy concept of “pocket-fitting.” Many of biology’s most powerful actors, like intrinsically disordered proteins (transcription factors!), scaffolding complexes, dynamically regulated switches, etc., don’t have accessible pockets at all. And yet, we’ve built decades of pharmacology and drug discovery infrastructure around hunting for them.
A generation later, the definition of “druggable” has started to bend and expand. It’s been one of the joys and immense wonders of my scientific career to see our appreciation of what can be drugged evolve from lock-and-key models to entire new frameworks of thinking. Today, the realm of “druggability” includes at least the following:
The manufacturing of biologics that replace missing factors in the body (insulin, GLP-1s, etc.) ushered in the magnificent world of biotechnology3, where we treat disease with recombinant proteins rather than just chemicals.
The advent of monoclonal antibodies (mAbs) and other biologics expanded our therapeutic reach to include receptors, extracellular targets, and secreted proteins, many of which were once considered inaccessible to traditional small molecules.
Covalent and irreversible inhibitors changed how we think about binding by forming durable chemical bonds with their targets. This allowed access to proteins with weak or transient pockets and enabled longer-lasting effects with greater selectivity. [More on this in another blog post!]
Peptidomimetics and protein mimetics gave medicinal chemistry new tools to go after flat, featureless surfaces like protein–protein interfaces. By mimicking the structure or function of natural peptides, they made it possible to disrupt interactions once thought undruggable, without relying on full-length proteins or antibodies.
RNA therapeutics including siRNAs, antisense oligonucleotides, and (hopefully soon) microRNAs opened a new era of sequence-based precision, allowing us to suppress or modulate targets at the transcript level before the protein is ever made.
Gene editing and epigenetic reprogramming, with tools like CRISPR/Cas9 and base editors, now allow us to write directly into the genome, redefining "targeting" as not just drugging but rewriting a cell’s instructions altogether.
Cell therapies and living medicines like CAR-T cells or engineered microbiomes remind us that sometimes the drug is the cell. In these cases, druggability doesn’t apply to a molecule or protein, but to the entire biological system being engineered to act as a therapeutic agent.
Targeted delivery systems (like lipid nanoparticles and antibody–drug conjugates) blur the line between modality and logistics. These innovations don’t expand the universe of targets themselves, but they unlock access to previously unreachable tissues, compartments, and cells, and to wider therapeutic windows.
Last but not least: Induced proximity techniques, like PROTACs and molecular glues, changed the game by reframing drugging as recruitment rather than inhibition. Instead of blocking function, these strategies redirect it, bringing two proteins together in space to degrade, stabilize, activate, or even modify each other. This is the “biochemical geography” I want to discuss today.
Among these expanding frontiers, I’m intrigued by different subsets of induced proximity. One of the most elegant (and surprisingly under-explored) is induced phosphorylation. It sits at the intersection of induced proximity and post-translational control. If degradation was version 1.0 of induced proximity, phosphorylation might be version 2.0, a writable molecular switch rather than a delete key.
2. Phosphorylation as a key member of the Post-Translational Modification family
Before we talk about inducing phosphorylation, it would be helpful to step back and review all post-translational modification “codes”. IMHO, they’re one of the more sophisticated programming languages of biology.
Most people think of proteins as a 3D-structured string of amino-acids, instructed to exist via the standard DNA→mRNA→Protein dogma. That’s partially true. If DNA is the source code and RNA is the compiler, then proteins are the machinery. But that machinery is modular, dynamic, and profoundly context-dependent, because proteins are constantly modified after they’re made.
No eukaryotic protein, as far as I know4, is just a string of translated amino-acids alone. Post-translational modifications (PTMs) to a protein are molecular add-ons: small chemical groups, sugars, lipids, or even other proteins that get attached to amino acids after translation to make the “complete” protein. To borrow the anecdote from early on, proteins are less like “dresses that Vera can afford” and more like “Met Gala situations”.
These modifications control nearly every aspect of a protein’s fate, including:
Structure: What does the protein look like (and does its conformation change)?
Activity: Is the protein turned on or off?
Location: Is it sent to the nucleus, membrane, or secreted?
Stability: Should it be degraded or protected?
Interactions: What partners can it bind?
Here’s a handy (albeit not comprehensive) cheat sheet of most popular PTMs:
Among these, phosphorylation is arguably one of biology’s most versatile and powerful molecular switches. By attaching a phosphate group, typically to a serine, threonine, or tyrosine residue, cells can dramatically change a protein’s behavior: turning it on or off, altering its shape, shifting its location, or flagging it for destruction. The process is driven by enzymes called kinases (which help add the phosphate group) and enzymes called phosphatases (which help remove it). The process is fast, reversible, and highly tunable, making it ideal for dynamic control in signaling networks. In many ways, phosphorylation is the cell’s native language for decision-making. It’s the foundation of signal transduction, metabolic control, cell cycle regulation, you name it.
I’m particularly intrigued by phosphorylation as a form of cellular signaling because it resists being reduced to a simple on/off switch. Depending on the site and the cellular context, phosphorylation can:
Deactivate a protein, for example by blocking its catalytic site
Change its location, such as triggering cytoplasmic-to-nuclear translocation
Promote degradation, by exposing phospho-degrons for E3 ligase recognition
Prevent degradation, by masking recognition motifs or stabilizing the structure
Alter protein–protein interactions, reshaping the interaction network
Rewire signaling cascades, flipping feedback loops or rerouting pathway flow
Phosphorylation, in this sense, is less like a binary toggle and more like a molecular decision tree. It encodes cellular logic not through sequence, but through site-specific, context-dependent chemical changes.
We can thus think of phosphorylation as a programmable instruction set that can work with incredible precision. Which begs the question:
If we understand phosphorylation so well and the signal is so critical to biology, why aren’t we using it as a programmable drug modality?
3. Induced Phosphorylation: Can we write code in phosphate?
Traditionally, drug developers have tried to block kinase activity, especially in cancer, where overactive kinases drive growth signals. The resulting class of kinase inhibitors has produced many successful therapies. But these drugs are blunt instruments. They shut down whole cascades, not just the nodes we care about. They don’t write new logic into the system.
Instead of blocking a kinase, can we coax it to do its job in a programmable way? What if we purposefully recruited it to a specific protein to phosphorylate a specific site and change that protein’s function in a specific way?
That’s the idea behind induced phosphorylation: a next-generation drug modality that borrows from the logic of induced proximity.
You’ve probably heard of Proteolysis-targeted chimeras (PROTACs): bifunctional molecules that recruit a target protein to an E3 ubiquitin ligase, marking it for degradation. This innovation made a lot more possible: we didn’t need to inhibit a protein directly. We just needed to bring it close to a machine that could target it for destruction.
Much of what we know about induced proximity as a therapeutic strategy (PROTACs, molecular glues, and now induced phosphorylation) can be traced back to the pioneering work of the Crews lab at Yale, whose innovations have helped reshape how we think about druggability.
The Crews lab and others (including commercial entities like Arvinas, Monte Rosa Therapeutics, General Proximity, Vant AI, Photys therapeutics, UbiquiTX (edited to add this one as one of the founders highlighted it after reading this article), and many more.) have expanded on the idea, to include:
Molecular glues stabilized weak protein–protein interactions
DUBTACs recruited deubiquitinases to protect proteins from degradation
RIBOTACs targeted RNA for cleavage by recruiting RNases
And more, including any drug discovery platform that allows for large-scale discovery in this space
For more on induced proximity, I’ve added a number of papers to the reference section at the end of this article, and am borrowing a schematic here from a great review paper by Liu and Ciulli.
PhosTACs and kinase-recruiting chimeras that bring a kinase into proximity with a protein of interest can be the next step in this family tree. I’ve seen a number of papers exploring this space (e.g., here, here, and here) and maybe some emerging biotechs like the ones mentioned above that are trying to bring drugs in this category to market. It’s exciting!
How does Induced Phosphorylation work?
The basic concept of a Phos-TAC (and induced phosphorylation more generally) falls in line with the induced proximity approach.
Create a bifunctional small molecule simultaneously binds:
A kinase (the effector)
A target protein (the substrate)
A linker of some sort (that can bind to the kinase on one end and the target protein on the other)
By bringing the kinase into close proximity with the substrate, the molecule enables or enhances phosphorylation of the target at a specific site.
Step-by-step, the mechanism of action would be as follows:
1) Binding domains: A PhosTAC is designed with two functional ends:
One end binds a specific kinase (e.g., PKA, CK1, CaMKII, LATS1)
The other binds a target protein of interest
2) Ternary complex formation: When the PhosTAC enters the cell, it binds both partners, forming a ternary complex of kinase + PhosTAC + substrate.
3) Phosphorylation: The kinase is now colocalized with the target, allowing it to phosphorylate a residue on the substrate protein, typically a serine, threonine, or tyrosine.
4) Biological outcome: The phosphorylation event alters the target protein's function in a programmable way.
Thus, induced phosphorylation changes the kinetics of the whole process. The kinase doesn’t need to find its target downstream of some complex signaling, it was brought there by this linker.
4. Thought Experiment: Programming Induced Phosphorylation at scale
“What if, much like engineers program logic gates into their software, we could build a universal toolkit that lets us add, remove, or redirect phosphorylation across the proteome? What if we could do it on-demand, with cell-type, time, and site specificity?”
Targeting ANY single undruggable protein with any working Phos-TAC would be a wonderful feat, and the basis of a whole biotech company’s program. In Part 2, we’ll focus on what this could look like for my favorite protein, YAP. But here, let’s go way more ambitious…
Let’s imagine a world where AI-assisted drug discovery is the norm, and where induced phosphorylation is a generalizable modality, as mature as PROTACs are today. And let’s take it further, where we can create a modular system to control phosphorylation states of arbitrary proteins at selected sites, in living cells, using programmable proximity tools.
1) The modular design
Every PhosTAC system could be thought of as three plug-and-play components:
Effector Module: A binding moiety to recruit the kinase (e.g., PKA, CK1, CDK1, CaMKII, LATS1).
Target Module: A binding domain specific to the substrate protein (ideally site-directed).
Linker Module: A tunable linker that sets distance, geometry, and potentially even kinetics of phosphorylation.
We could then add additional programmable components:
Logic Gates: AND/OR/NOT behaviors to integrate multiple inputs.
Sensor domains: Domains that are responsive to calcium, pH, redox, or light.
Feedback elements: Built-in phosphatase recruitment after a given time window to ensure reversibility.
2) Use case: Discovery and research
Yes, therapeutics are the holy grail. But what if we could use these as a research tool to map kinase–substrate relationships in all living cells, under different conditions? Lots of drug targeting ideas could come out of that. You could imagine:
A high-throughput platform that screens libraries of PhosTACs for functional phosphorylation across thousands of targets.
A barcode-linked delivery system in cells that traces which phosphorylation event led to which phenotypic outcome.
3) Integration with synthetic biology
Pair this with the growing library of synthetic biology tools, like:
Using CRISPR-based synthetic promoters to transiently express kinases only under defined circuit conditions.
Layering in optogenetic or chemogenetic control so phosphorylation happens only in response to light, drug, or metabolic status.
Embedding this system in designer T cells, regenerative organoids, or engineered microbiomes, where logic-gated phosphorylation controls differentiation, activation, or immune synapse formation.
4) Layer in even more logic!
Spatial controls (PhosTAC system only works at the synapse, or only in T-cells for example)
Temporal controls (reversible phosphorylation within minutes)
More combinatorial logic (phosphorylation only if another protein is also bound, or recruiting a kinase AND inhibiting a phosphatase
5) Applications and use cases
Let’s go wild:
Regenerative medicine: Recruit GSK3β to selectively phosphorylate β-catenin ONLY in senescent fibroblasts to suppress Wnt signaling and rejuvenate tissue niches.
Neurobiology: Build a system where CaMKII is recruited only to PSD-95 in active synapses, strengthening circuits in real time as learning occurs.
Immuno-oncology: Induce phosphorylation of PD-L1 in tumor cells, marking it for internalization without destroying the cell, turning off immune evasion temporarily.
Cardiology: Deliver an LNP-packaged PhosTAC that reactivates dormant calcium channels only in hypoxic zones post-myocardial infarction, restoring contraction locally.
I asked a friendly AI chatbot near you to name such a program. It suggested calling it the “Kinome Compiler”. I rather like it.
5. What’s Next in Part 2: YAP, phosphorylation, and you
In Part 2, we’ll zoom in on a case study of ✨ my favorite protein ✨, YAP to explore how we might design an induced phosphorylation system to control its activity.
YAP is a transcriptional co-activator involved in organ size, regeneration, and cancer. It’s exquisitely regulated by phosphorylation, and yet rarely drugged directly.
We'll walk through a thought experiment:
Which kinases might we recruit?
Which sites on YAP should we target?
What cell types would we test this in?
How might this intersect with synthetic biology circuits?
Spoiler: there will be choices, trade-offs, and some creative protein engineering.
💭 That’s all, for now.
6. Selected references 🤓
Hopkins AL, Groom CR.
The druggable genome.
Nat Rev Drug Discov. 2002;1(9):727–730.
https://doi.org/10.1038/nrd892
The seminal paper that defined the idea of the “druggable genome” framing drug discovery around structural tractability, and inadvertently highlighting all the biology it left out.Liao X et al.
Recent advances in targeting the “undruggable” proteins: from drug discovery to clinical trials.
Signal Transduct Target Ther. 2023;8:187.
https://doi.org/10.1038/s41392-023-01589-z
Broad review of the growing arsenal against “undruggable” targets, spanning covalent inhibitors, allosteric modulators, PPI disruptors, nucleic acid therapies, targeted degradation, and immunotherapeutics. Also highlights how many of these tools have now reached trials.
Sakamoto KM et al.
Protacs: Chimeric molecules that target proteins to the Skp1–Cullin–F box complex for ubiquitination and degradation.
PNAS. 2001;98(15):8554–8559.https://doi.org/10.1073/pnas.141230798
The first proof-of-concept for PROTACs, showing that small molecules could induce degradation by recruiting E3 ligases to a target protein.
Bondeson DP et al.
Catalytic in vivo protein knockdown by small-molecule PROTACs.
Nat Chem Biol. 2015;11(8):611–617.
https://doi.org/10.1038/nchembio.1858
The breakthrough that demonstrated PROTACs could work robustly in living cells. This took PROTACs from a theoretical tool to a drug development platform.
Chen et al.
Modulation of Phosphoprotein Activity by Phosphorylation Targeting Chimeras (PhosTACs)
ACS Chem Biol. 2021 Nov 15;16(12):2808–2815.
doi: 10.1021/acschembio.1c00693
The first demonstration of induced phosphorylation as a modality, using bifunctional molecules to bring kinases to new substrates, essentially the phosphorylation equivalent of a PROTAC.Paiva SL, Crews CM.
Targeted protein degradation: elements of PROTAC design.
Curr Opin Chem Biol. 2019;50:111–119.
https://doi.org/10.1016/j.cbpa.2019.02.022
Review that provides deep insights into linker chemistry, E3 ligase selection, and the modular logic of PROTACs. This paper highlights principles that directly inform how we might design phosphorylation-inducing chimeras.Churcher I.
PROTAC-induced protein degradation in drug discovery: breaking the rules or just making new ones?
J Med Chem. 2018;61(2):444–452.
https://doi.org/10.1021/acs.jmedchem.7b01272
An industry-focused perspective on how PROTACs challenged classical drug discovery dogma, and what that means for new modalities like induced phosphorylation.Liu X, Ciulli A.
Proximity-Based Modalities for Biology and Medicine.
ACS Cent Sci. 2023;9(7):1269–1284.
https://doi.org/10.1021/acscentsci.3c00395
A comprehensive and accessible review covering PROTACs, molecular glues, and emerging proximity tools like PhosTACs and DUBTACs.
Any takers? :)
FWIW, it really made sense at the time. We had just barely sequenced v1 of the human genome. A lot of the innovations that came in the last two decades built on the shoulders of giant innovations in molecular and cellular biology, which did not exist in 2002.
I feel like I can’t do justice to the biotech world in this article. For more, I recommend you read Tim Harris’ recent book “In Pursuit of Unicorns: A Journey Through 50 Years of Biotechnology”
Maybe some small or rapidly-degraded proteins don’t undergo PTM? Curious if anyone has more detail on this.