Skip to main content

Chemical Biology

Probes For Everything

In case you don’t know, there’s officially an effort to try to develop chemical probes for basically every protein in the human proteome. The “Target 2035” initiative has been looking through the literature and finding what you’d expect: power-law distributions that have most people working on proteins that other people have worked on. And that’s natural enough, since many of those have evidence of their importance (in human disease, most of the time), and these are the areas mostly likely to lead to real-world applications, not to mention real-world funding.

But it would be a good thing if research were able to branch out more and explore proteins that we don’t know so much about, given the general level of our ignorance. And the belief is that a good way to jump-start that is to provide some pharmacological tools that people can use, which seems like a reasonable assumption to me. As the authors note, “For almost every protein for which a cell-active pharmacological modulator has been made available, the paper describing the modulator is among the top-cited papers on that protein in the entire literature. . .” That said, asking people to go work on proteins of unknown or unclear function is hard enough, and telling them that they will have to develop their own tools if they even want to get started makes the barrier to entry even harder. That’s what we do in the drug industry a lot of the time, and it is no stroll through the petunia patch. Again, the paper:

Regrettably, pharmacological modulators of the required potency and selectivity to support interpretable and reproducible science are both challenging and expensive to invent, and also require skills commonly found in industry.

So I applaud the idea behind this effort. It’s not going to involve strolling past many petunias itself, though, particularly if we’re seriously trying to come up with those probes before 2035. But the Structural Genomics Consortium has had some success developing new probes (and in persuading drug companies to make some of theirs available), so they’re going for it. It’s recognized as an aspirational goal, and also that the only way to realize it will be to take advantage of further advances in technology – if you tried to generate thousands of new reagents the way we’ve been coming up with them to date, the project is simply not realistic at all. The first phase (out to about 2025) is enough of a tall order as it stands:

 (i) collecting, characterizing, and distributing existing pharmacological modulators for key representatives from all protein families in the current druggable genome and generation of chemical probes for additional family members; (ii) developing the crucial and centralized infrastructure to facilitate data collection, curation, dissemination, and mining that will empower the scientific community worldwide; and (iii) creating centralized facilities to provide quantitative genome-scale biochemical and cell-based profiling assays to the federated community.

It’s important to note that this effort will be targeting both chemical and biological probes – antibodies or nanobodies that can affect protein function are part of the picture, too, as are protein degraders, allosteric modulators, whatever works and can provide insights. The plan is to try to get the existing let’s-make-better-probes efforts to consolidate under this umbrella and be ready to jump on new ideas and techniques as they become available. And the good news is that this isn’t one of those efforts that has to make it to the end to be really valuable (as pointed out in this Nature writeup on the effort). Right now, the coverage of the proteome with really useful probes is way down in the single per cent range at best, so there’s a vast amount of room for improvement. And similarly, there’s plenty of room for improvement in how we find, assay, and characterize such reagents, and such techniques will have very broad applicability.

So while I think the goal is aspirational indeed, I still think that trying for it is a good idea, and I’ll be watching with interest over the next few years to see what progress is being made. And I have my own suggestion, if funding can be found along the way. Let’s take all the existing physical supplies of the worst chemical probes available, the ones that are still being sold in the catalogs even though they don’t do anything like the label says, the ones that are still appearing in the literature and cluttering up science with unreliable results. . .and put them deep within the most inaccessible locations we can possibly find. Maybe a nuclear waste repository. Maybe if we just round up the world supply of rottlerin and the like, people will finally stop using them.

15 comments on “Probes For Everything”

  1. Marcus Theory says:

    “Maybe if we just round up the world supply of rottlerin and the like, people will finally stop using them.”

    What is dead may never die….

  2. SP says:

    Didn’t Schreiber propose this like 20 years ago?

    1. John Wayne says:

      The NIH had an initiative called MLPCN to make good quality probes that was performed by the Broad Institute and other similar institutions. I think the reports are in the public domain, and some potentially good stuff is in there.

      1. DrZZ says:

        The probe reports from the Molecular Library project can be found at

  3. Another Guy says:

    I thought high throughput NMR or MS or whatever was going to achieve the holy grail of ” a ligand for every protein”. Where are we with that?

    1. John Wayne says:

      Those ideas are possible, but don’t address issues of probe usefulness. You could probably demonstrate binding to most proteins with the top 50 most promiscuous compounds we know of, but that lack of selectivity isn’t very useful for sorting out biology. Once you add in functional binding as a requirement, things get even more complex.

  4. MAEngineer says:

    Aspirational for sure, but admirable, too. An important point is that one probe will not be enough for some biological targets. Some proteins possess multiple domains with distinct functions, like separate enzymatic and structural domains. For example, many mammalian aminoacyl-tRNA synthetases (join amino acids to tRNAs) are parts of multi-protein clusters. Why? We only have seen the tips of the (functional) icebergs in this area…

  5. Anonymous says:

    From the article abstract, “One of the best ways to interrogate the function of a protein and to determine its relevance as a drug target is by using a pharmacological modulator, such as a chemical probe or an antibody.” Although I strongly support the need for small molecule R&D, I thought that the first part of that, “ways to interrogate the function of a protein and to determine its relevance as a drug target” is being studied with many of the new bio techniques. If I believe everything I read, it seems to be possible to scan through the entire genome (proteome) one by one (or ~20 bp by ~20 bp) using CRISPR, siRNA, and variations of those methods. You no longer need to make a genomic level knockout and worry about getting an uninformative lethal deletion.Just like adding a small molecule probe, you add or turn on/off a gene at any time in the cell’s development or life. (I make it sound easy, huh!)

    I think that Schreiber’s proposal to survey small molecules against many targets is what led to the establishment of the PubChem dbases (including PubChem BioAssay). From the early going, one of the problems was something that is discussed frequently In The Pipeline: reproducibility and relevance. Especially in the early going, without curation of the publications and data, you had a lot of inconsistent and contradictory results and claims. I thought that there was an effort to try to make assays more uniform; however, I think that most of us know that a uniform protocol doesn’t mean that every lab gets the same result from the same assay even using the supplied controls and reagents.

    (Previously, In The Pipeline, it was noted that assays in transformed cells sometimes change as the cell line mutates. Your HeLa cells are not the same as Helen Lane’s original tumor cells and may not be the same as those in the lab down the hall or across campus. Healthy cells or tissues are sometimes maintained in unnatural media that could mislead finding normal functions. Solubility and precipitation; the non-natural effect of co-solvent DMSO or EtOH on an assay; do you assay in kidney cells, liver cells, or leukocytes? – or all of them? – and how do you compare the outcomes?; etc..)

    I’m NOT saying NOT to do this. I am saying PLEASE do it RIGHT!

    1. HEK293 says:

      The original clinical source of the HeLa cell line was a patient called Henrietta Lacks, not Helen Lane.

    2. Yeah, I agree with this anonymous lady. do this but PLEASE do it RIGHT!

      1. Anonymous says:

        I guess I proved my point! (But not the way I wanted to. Thanks for the HeLa cell correction.)

    3. loupgarous says:

      And, as the American Type Culture Collection found out to its dismay, your HeLa cells might wind up infecting other cell lines, without anyone realizing it before working with them in their research.

  6. gippgig says:

    Off topic, but here’s another example of a reagent not being what it was supposed to be:
    doi: 10.1126/sciadv.aay5611

  7. drscoop says:

    Interesting review of this on the Nature website:

    That article suggests a pot of 58M Euros to deliver 5000 probes, by 2035. I make that roughly 11,500 Euros per probe, at a rate of ~300 probes a year. That’s quite an ask, even for more druggable proteins and contrasts with the (somewhat more realistic) $2m per probe estimate in the DDT article.

    Given the recent paper on CRISPR vs siRNA and the actual target of believed-to-be-selective targeted cancer drugs (discussed here:, it’ll also be an interesting debate as to how the target engagemetnand selectivity of these probes will be defined and categorised…

    A worthy effort, but some significant hurdles to overcome, both scientifically and tactically.

    1. Peter J Brown says:

      Agreed that there are plenty of hurdles to overcome, however, the goal is 5000 chemogenomics compounds which do not need to reach the selectivity standard of a probe. Therefore the comparison of costs to earlier probe efforts ($2 million) is not meaningful.

      The figure of 11,500 Euros per chemogenomic compound is certainly a challenge!

Comments are closed.