Skip to main content

Analytical Chemistry

The Phosphoproteomic Landscape Speaks – What Did It Say, Again?

Medicinal chemists are extremely familiar with G-protein coupled receptors (GPCRs), and it’s a safe bet that any pretty much any neurotransmitter (for example) that can be named by the general public is a GPCR ligand, too. Serotonin, dopamine, histamine – all the classics are there, and that’s reflected in the number of marketed drugs that target this signaling family.

I started out in the industry working on dopamine receptors, and after a year or so, I thought I knew a reasonable amount about them. But that was my peak of confidence. Since then, the more I’ve learned, the less sure I am that I know anything useful about them at all. (I had a chance a few years ago to try that line out on Duke’s Robert Lefkowitz, a Nobel winner for his GPCR studies, and he said that he can report that it doesn’t get any better the further you go!) Like everything else in molecular and cellular biology, GPCR signaling and function is wildly, inhumanly detailed and complex. Every new technique we come up with to study such things only reveals new layers of ever-more-finely-grained crosstalk, feedback, and regulation. This would be a good time to reference Robert Graves’ poem “Warning to Children“!

I bring this up because of this recent paper, which gives us a look at GPCR effects that we haven’t had before. The authors (a multicenter team from the Max Planck Institute (Martinsreid), Innsbruck, Temple, and Novo Nordisk) are using high-throughput mass spec proteomics to look at phosphorylation states in response to GPCR signaling. Phosphorylation is, of course, a ubiquitous mechanism for altering the functions of proteins, and the number of such switches and signals I would very much not like to count. In this case, the group is looking at 50,000 phosphorylation sites in brain tissue via their previously described “EasyPhos” technique, a streamlined LC/MS/MS protocol. (There are plenty of other ways to study phosphorylated proteins, but this one seems to be a particularly effective wholesale method).

That’s a lot of data. Five minutes of treatment with a classic kappa-opioid agonist (U-50488H) showed plenty of changes in phosphorylation states, in the order striatum > hippocampus > cortex > medulla oblongata > cerebellum (which is pretty much the order of opioid receptor density as well). You don’t see these effect in kappa-knockout mice. Comparing that five-minute time point with a 30-minute reading across the different brain regions shows still more differences, and these don’t correlate simply with the expression levels of the kinase enzymes and substrates involved. No, these changes (which I will not attempt to summarize!) do seem to be specific kappa-opioid-mediated ones. The proteins that show up are involved in a whole range of cellular processes, from the cell membrane down to the nucleus (and those are just the ones that have some degree of useful annotation – if you think that the proteome is truly annotated for function in every case, think again, hard).

Comparing this profile with another kappa-ligand, 6′-GNTI, showed about a 30% to 50% overlap in affected proteins, depending on brain region and time point. That makes sense – it’s an agonist, but better described as a “funny partial agonist” as compared to something like U-50488. Whether you find that overlap reassuring large or terrifying small depends on your personality, I’d say, and it’s permissible to alternate between those two views.

Focusing in further on a set of five structurally distinct kappa agonists, each known to have its own behavioral profile in mice, showed that there are clusters of phosphorylation events that can be binned and apparently assigned to behavioral phenotypes. This is quite interesting, and I hope that it’s a robust result (five compounds can be described as merely a good start on that question). That’s really the state that we find ourselves in much of the time these days: looking at huge, detailed list of cellular and molecular differences, without a clear understanding of what’s going on in almost all of them, and hoping to find correlations that can (A) give useful predictions and (B) furnish clues as to what some of those mechanisms might be. Our high-throughput tools generate far more information than we know how to interpret down at the fine levels, so we have to look for broader strokes and patterns and work from there.

If someone with a fundamentalist religious outlook reads this paper – the intersection set does not have to be null, but it will not be well populated, either – they might be reminded of Paley’s “watchmaker analogy“. That’s occurred to many other people as well, but his famous formulation was of stumbling across a pocket watch while walking in a field, which would immediately suggest the existence of a watchmaker. Perhaps that last word should be capitalized?

In this case, though, as in all of molecular biology, closer inspection not only reveals a wealth of impressive detail, but huge numbers of things which do not so much appear to have been carefully crafted by a divine hand, as much as assembled by a blind lunatic with infinite time, infinite willingness to tinker (and a correspondingly infinite willingness to accept whatever works as soon as it does), and infinite supplies of duct tape, super glue, and baling wire. What’s more, the whole mechanism keeps falling apart over time in subtle (and not so subtle) ways, which just leads to casual repurposing of the altered pieces. This series of tweets illustrates some of that on a higher systems level in the brain, but the same sort of “Hey dude, what’s your problem, it worked didn’t it?” style is in operation from top to bottom. And never forget, that’s because all of those zillions of variations that were tried and didn’t work as well as what we have now are dead. Never forget, either, that the definition of “work as well” is subject to brutal, random restatement at any moment as conditions change. Supervolcano drops the Earth’s temperature for ten years? Deal. Asteroid strike? Bummer. Your savannah ecosystem is turning into an ocean again after a million-year hiatus? It happens – if you don’t know how to swim, you’d better know how to fly. And so on.

So studies like this phosphoproteomics one are a look at what evolution has left us with after all this time. It’s a mess! But somehow it works, and since no instruction manual is provided, we have to write our own. And here’s where that stands, as of the latest publications. . .

27 comments on “The Phosphoproteomic Landscape Speaks – What Did It Say, Again?”

  1. Isidore says:

    Like Matthias Mann
    We want to become famous
    But we never will.
    – J.S. Richard

  2. Thoryke says:

    It’s not turtles all the way down; it’s kludges. Some more elegant than others….

  3. secret sauce says:

    Analogous investigations of changes in the ubiquitination and acetylation profiles of the proteome would be interesting.

    p.s. Derek – thanks for the comments and the watchmaker metaphor, or lack thereof

  4. Anon says:

    I get so fed up with scientists these days telling me what will or won’t work based on their “great wisdom and understanding” without ever having tried it, instead of just giving it a shot to see what happens, like nature itself does. It seems the more we learn *about* nature, the less we are willing to learn from it!

    We need less design, more trial and error!

    1. Peter Kenny says:

      Trial and error can be formalized in a Design of Experiments framework and I’ve linked a blog post that may be relevant to this point as the URL for this comment.

    2. Billy says:

      Much of this originates from the guys writing the checks. Scientists, generally, would like nothing more than to try anything and everything.

    3. tlp says:

      It’s funny because I have an opposite feeling. It seems to me that recent trend in science is departing from hypotheses and theories and leaning more to ‘big data’ approach. I wouldn’t say what approach is more ‘natural’. But without synthesizing ideas into theories we’d probably still work on the best design for steam engine.

  5. Chairman Mao says:

    DOE is the scientific equivalent of having a “safe space” for scientific snowflakes.

    1. John Wayne says:

      This comment made me laugh out loud.

      When I first learned about DOE my first through was, “Wait, we aren’t all doing that all the time?”

      It turns out that we are not.

  6. luysii says:

    ” as much as assembled by a blind lunatic with infinite time, infinite willingness to tinker (and a correspondingly infinite willingness to accept whatever works as soon as it does), and infinite supplies of duct tape, super glue, and baling wire. What’s more, the whole mechanism keeps falling apart over time in subtle (and not so subtle) ways, which just leads to casual repurposing of the altered pieces.”

    Well, we don’t have infinite time, nor does the earth. Nor does the earth have infinite mass. Suppose the whole earth were not mostly metal, but made of carbon, nitrogen, oxygen, sulfur and hydrogen in the proportions you’d need to make proteins. Now let’s just make 1 molecule of each possible protein starting with the 400 2 amino acid ‘proteins’. When do you think we’d run out of mass? How long would that final set of proteins protein actually be?

    1. Another Guy says:

      If each of the 41-aa proteins could be somehow tested for biological usefulness very quickly (say 1second), then hydrolyzed back to their constituent amino acids, and then reused their create new proteins not previously synthesized, what would be the upper protein size limit? Maybe this is what evolution does – continuously create, test, then recycle? Why bother archive all the duds for future humans to discover and ponder?

      1. luysii says:

        Another Guy — good point, and someone who read the post back then said basically the same thing. So I put this in at the end of the post

        Consider the following: Let us suppose there is a super-industrious post-doc who can make a new protein every nanosecond (reusing the atoms). There are 60 * 60 * 24 * 365 = 31,536,000 ~ 10^7 seconds in a year and 10^10 years (more or less) since the big bang. This is 10^9 * 10^7 * 10 ^10 = 10^26 different 41 amino acid proteins he could make since the dawn of time. But there are 20^41 = 2^41 * 10^41 proteins of length 41 amino acids. 2^41 = 2,199,023,255,552 = 10^12. So he has only tested 10^26 of 10^53 possible 41 amino acid proteins in all this time.

        As per your suggestion, this is making one protein at a time. However, even if the hapless post-doc was able to use the entire mass of the earth (6 x 10^27 grams) every nanosecond to make a different set of proteins (one molecule of each), he would never have made all the possibilities for a protein of length of one of the two chains of hemoglobin (141 or 146 amino acids) since time began. Hemoglobin just isn’t that big as proteins go (the gene mutated in cystic fibrosis has well over 1000).

        1. Another Guy says:

          Fascinating thought experiment! So even when ridiculous assumptions are made (entire earth made of C,N,O,H,S), each protein is tested and reused within a nanosecond (probably faster than the diffusion limit for most proteins to find their way to a target and bind), and have almost 14 billion years to work with, we can’t get past evolving hemoglobin. But nature did.

          Perhaps the blind, lunatic watchmaker can still use the information gained from each protein synthesis and “bioassay” to guide the next synthesis to prevent going down totally useless paths?

          1. AVS-600 says:

            Evolution goes down useless or harmful paths all the time; that’s what deleterious mutations are, after all. The resolution to your question is actually just that nature gives no guarantees that it came up with the best solution possible. We are pretty sure that hemoglobin is the best heme-based O2 carrier that nature has ever come up with, but we have no idea whether there are other proteins that would be even better at the job that have just never been mutated for.

          2. Believer says:

            Or perhaps the blind lunatic watchmaker has the entire cosmos to work with and not just earth. I personally believe microscopic life to be common and broadly distributed across inumerable worlds. I hope I live long enough for this belief to be put to the test on Europa and perhaps elsewhere.

          3. AI says:

            Hemoglobin and chlorophyl are suspiciously similar. Evolution probably just flailed around to find the best solar power plant it could, and then repurposed it with a different metal ion when the time came to move oxygen around.

    2. Anonymous says:

      Molecular Evolution has been tried in the lab, too. One of the earliest examples I know of is due to Duane Venton (UI-Chicago). “Nonspecific protease‐catalyzed hydrolysis/synthesis of a mixture of peptides: Product diversity and ligand amplification by a molecular trap.” Peptide Science, 1996, 40(6), 617-625. That was, basically, automated in vitro protein evolution by synthesis, binding, hydrolysis of the weaker binders, re-synthesis, recycle, repeat in order to get the tightest binder.

      An Old Chemist recently (2-July-2018; Complex Organics Are Out There) mentioned Steve Benner who is also a big name in molecular evolution, e.g., at the Foundation for Applied Molecular Evolution. Others have made contributions to the field, as well.

      1. Derek Lowe says:

        David Liu’s phage-driven protein evolution work is some of the current state of the art.

    3. Roger Moore says:

      All this argument suggests, though, is that evolution doesn’t work by exhaustively trying all possible combinations of amino acids to see what works best. Instead, it works by taking something that kinda sorta works and tinkering with it to see if it can be made to work better. But that’s exactly what Derek was saying, with minor poetic license by substituting “infinite” for “so vast it beggars human imagination”.

      1. luysii says:

        Sure. Just start tinkering to produce the C* complex of the yeast spliceosome which contains 46 proteins, 3 snRNAs (U2, U5 and U6), a free 5′ exon and an intron lariat. The combined molecular mass is 1.7 megaDaltons. Cell vol. 169 pp. 918 – 929 ’17.

  7. steve says:

    This study is the equivalent of physicists trying to understand two-body gravitational interactions by mapping the motions of every star in the milky way. Why not start with something simple – try EasyPhos on C. elegans (only 302 neurons!) and once you understand that system then maybe move up the ladder to mammals.

  8. Scott says:

    The human brain is *very* adept at seeing patterns. Sometimes a little too adept, and sometimes way too adept (to the point of mental illness, conspiracy theorist-style).

  9. David Edwards says:

    Ah, in vitro evolution. I have some interesting papers covering this topic in my collection.

    One of these covers the use of in vitro evolution to produce a variant of the HIV virus, that can be “turned on” by tetracycline antibiotics, and “turned off” once it’s done its job of activating the immune system, by withdrawing the antibiotics from the recipient of the proposed vaccine. The paper in question can be downloaded from PNAS here.

    A related paper is featured in PubMed.

    1. Anonymous says:

      Just gave a cursory read of the tetracycline regulated viral expression. One problem that popped into my head was the extent of antibiotic contamination in the environment, including drinking water (depending on where you live and the water treatment methods used). Before leaving the research lab, make sure the regulator is not a common pollutant or a ubiquitous but unintended promoter. Although tetracycline control is dose dependent, and environmental levels are very low, I wouldn’t want to see someone have their vaccine turned on or off by accident.

      (Researchers are finding odd ball pollutants spreading all over the globe, including the Arctic and Antarctic where there is no direct exposure or emission.)

      1. David Edwards says:

        I owe you a big “thank you” for pointing that out … an issue I was completely unaware of! Which is going to have a potentially nasty impact on a research programme that I really wanted to see bear fruit.

  10. anon says:

    We used to classify disease based on visual inspection of histological slides. Now we classify them by molecule defects. All the omics will allow us to classify disease in a even finer manner. Beyond that, we still need human intellect to penetrate and reduce complexity to simple elements and to find laws that govern cause-effect and then put them back to re-synthesize the complexity and to create new things. This process produces marvels in physics and chemistry.

  11. RicoTorpe says:

    From a programmer’s perspective, all of these processes — as well as DNA — seem like an enormous program which has been poorly maintained, poorly designed, and the ultimate in “spaghetti code” featuring an uncountable number of global variables which are used in multiple, seemingly-unrelated processes.

Comments are closed.