Open Peer Review Of Ethan Perlstein’s Genetic Studies Of Sertraline Toxicity In Yeast

Setraline is the chemical name for Zoloft, a commonly prescribed antidepressant. An independent scientist currently unaffiliated with any academic or industrial research entity named Ethan Perlstein has been recently soliciting funds via “crowdsourcing” to continue his studies on the mechanisms by which sertraline is toxic to yeast, with the goal of shedding light on antidepressant function in the human brain.

The experimental foundation for his research plan that he has been soliciting funds to pursue is a paper he published in 2007 in the journal Genetics. This paper describes the isolation and characterization of yeast mutants that are resistant to sertraline-induced toxicity.

In this post, my goal is to enumerate various experimental findings that may be considered germane to the likelihood that his research will succeed at its stated goal of shedding light on antidepressant function in the human brain.

(1) Sertraline is, as pointed out in the Genetics paper, a “cationic amphiphile”. That means that it has both a hydrophobic component and a hydrophilic component that is positively charged under physiological conditions. The paper also points out that cationic amphiphiles “[have] long been recognized [to] interact with phospholipid membranes”. This is, of course, not at all surprising, since cationic amphiphiles are basically detergents.

(2) In human patients, the total concentration of sertraline in plasma at therapeutic doses has been estimated as 80-165 nM, that is bioavailable in the plasma (i.e., unbound to albumin and other proteins) has been estimated as 1.5-3 nM, and that is bioavailable in brain has been estimated as similar to plasma 2-4 nM. This means that in human patients taking sertraline, their brain neurons are probably exposed to a concentration of sertraline somewhere between 1.5 and 4 nM. (These estimates come from this paper.)

(3) According to the Genetics paper, the concentration of sertraline employed in the screen for resistant yeast mutants is “∼45 μm sertraline, which is approximately three times the IC50 of sertraline as determined by dose–response experiments in liquid culture”. The paper also notes that “lowering the selection concentration from ∼45 to ∼42 μm resulted, on average, in a 10-fold increase in the number of resistant colonies”. This indicates that the dose employed is nowhere near saturating with respect to sertraline’s physiological target in yeast cells, and is rather near the lower bound for having any physiological effect.

(4) Since the yeast are plated on a growth medium (YPD) that contains no intact proteins or long peptides, the yeast cells are likely exposed to close to 45 μm sertraline. This means that the yeast cells are likely being exposed to somewhere between 11,250 and 30,000-fold greater concentrations of sertraline than the brain neurons of human patients being treated with sertraline.

(5) As discussed in detail in the Genetics paper, the yeast clones resistant to this dose of sertraline have spontaneous mutations/polymorphisms in genes that encode proteins important for cellular lipid membrane homeostasis and trafficking.

(6) As discussed in detail in the Genetics paper, the dose of sertraline used for the resistance screen induces severe defects in the intracellular membrane structures within the yeast cells.

Based on all of the above, these conclusions follow:

(1) The most parsimonious interpretation of the results of the yeast screen is that mutants resistant to detergent disruption of intracellular membranes have been isolated, and this likely has nothing to do with specific effects of sertraline versus other detergents.

(2) The astronomical difference in effective dose for human therapy versus influencing yeast cellular physiology suggests that the effects of sertraline on yeast cells most likely have absolutely nothing whatsoever to do with its effects on neurons.

(3) Continued study of the effects of sertraline and other amphiphilic compounds at astronomical doses on yeast cells is likely a waste of time and effort vis a vis the goal of elucidating antidepressant actions in the human brain.


  1. DrugMonkey says

    This indicates that the dose employed is nowhere near saturating with respect to sertraline’s physiological target in yeast cells, and is rather near the lower bound for having any physiological effect.

    I’m thinking this is the thing that needs to be answered next. I hate to be all dose-response but really, it there is anything useful in this model….

  2. says

    My general caveat with yeast is that they are remarkably impermeable to most drugs. In my old field of topoisomerase-targeted chemotherapy, one would have to expose yeast to 1,000- to 10,000-fold higher concentrations than in human plasma to observe the same magnitude of enzyme inhibition in cells. So, I’d never equate in vitro yeast with in vivo human. For this reason, folks generally use permeability mutants as the starting point (John Nitiss did this in my field). So, no surprise that permeability mutants were elected here. My question is what would happen in yeast already preamble to sertraline.

  3. says

    I really wish this person had carefully read the followup study in our peer-reviewed PLOS ONE ( before making the same tired concentration arguments I’ve been dismantling for years.

    I should add that this individual has savagely trolled me in the past, here (, and more decisively here (

    Here’s my favorite:

    “Perlstein is obviously going through some variant amalgamation of Kubler-Ross’s five stages of grief. Right now he’s in a hybrid anger-bargaining stage, telling himself “Fucke those assholes in academia and at NIH! I’m totally gonna fund a sustained wet-bench scientific career with a paypal tip jar! YEAH!”

    So any impartial scientific evaluation he/she — who am I kidding, only a man can be this much of an asshat — makes about my open proposal, behind a veil of pseudonymity no less, should be considered in that light.

    Anyway, I made painfully clear in our peer-reviewed Genetics paper that Zoloft overdose occurred at supraphysiological concentrations, and that concerned us. So that’s why we obtained radiolabeled Zoloft in order to perform uptake assays at nanomolar concentrations, much closer to the Kd of ADs determined in synaptosomal reuptake assays, where SSRIs were born in the 1970s and 1980s. We found that overdose-resistant mutants exhibit differences from wildtype in nanomolar Zoloft uptake. I repeat — NANOmolar concentrations. So below the critical micelle concentration. Way, way, way below.

    So if the V-ATPase and clathrin pathway genes only played a role under unphysiological drug concentrations, then why did we see striking differences in uptake at nanomolar/”physiological” concentrations? In fact, why do the swa2 and chc1 overdose-resistant mutant actually exhibit more uptake than wildtype if these genes are just involved in some boring surfactant response?

    What’s more, why does Zoloft exhibit a cytoprotective affects on the swa2 and chc1 mutants at sub-lethal concentrations (0.5uM-10uM) if these genes are only involved in some boring surfactant response? So to be 100% clear: Zoloft exhibits biphasic effects. Genotype-specific cytoprotection at sub-lethal doses, and overdose above a catastrophic concentration of around 22.5uM, as measured in single- cell dose response experiments.

    I wish this person had actually read the entire Genetics paper (from 2010 btw, not 2007, but who cares about the details, right?) and thought about it for more than like 20 minutes before basically trashing my research program.

    Please check out the 16 comments on our PLOS ONE paper for serious review, especially the comments of Prof Jeff Krise, who actually knows something about pharmacology, and who hasn’t trashed me by name on comment threads.

    I’m happy to debate anyone, anytime about the relevance of simple model organisms to neuropharmacology. And if you want non-yeast examples, like worms on Prozac, check out the references of both Zoloft papers. The concentration straw man gets some nice takedowns from Bob Horvitz and Ed Buttner, among many other examples.

  4. leigh says

    i am not making the connection between “look, this drug interacts with lipid membranes” (there are many drugs that may interact with membrane lipids, so yippee) and “this interaction dictates the pharmacological therapeutic mechanism of antidepressants and warrants further study” – large pieces of the narrative from the observation to the fundamental research questions posed are missing, and i do not see a science-based line of logic here. the only defenses i am reading invoke The Importance Of This Work Is Self Evident and/or The Fact That I Have To Sell This Means Pharmacology As A Field Is Floundering.

    i am interested to see whether scientific replies are posted here or if it will be more of the same.

  5. says

    IANAPh, but the piece of evidence that has always convinced me that Ethan may be onto something very interesting is that if you make a large collection of yeast recombinants, screen them against a large panel of drugs, and then cluster the drugs by their profiles of sensitivity across the recombinants, you recover a drug cluster of structurally unrelated psychoactive drugs. I.e. antipsychotics, tricyclic antidepressants and SSRIs cluster together. To me, this is strong evidence that they is some pathway/process/thing happening in yeast which is a common target of drugs that have similar effects in organisms with neurons.

  6. leigh says

    drugs lipophilic enough to cross the blood-brain barrier also incorporate into cell membranes rather easily. i don’t see this as indicative of an underlying mechanism, Ben.

  7. DrugMonkey says

    Abel, I would be interested in your view not on the dose equivalency, per se, but rather in what seems to be PPs most devastating suggestion, i.e., that this is really just a detergent-like nonspecific disruption of the cells. Are you saying that we can’t infer this from the concentration either?

  8. Namnezia says

    Interesting that the reply to the critique begins with character attacks on the critic, even when there are no such attacks to the author of the proposal in this critique.

    Even if differences were seen between mutants and controls in accumulation in the nanomolar range, it is clear from these studies that free living organisms have very unique mechanisms of handling high concentrations of all sorts of stuff, and it is not clear whether these specializations have an relevance to whole tissue. Plus as PP says, there’s no evidence that the effects are specific to this drug and not all detergents.

    And I think PP’s point here is not that these studies are or not relevant to studying accumulation of ampiphilic compounds, but rather that they have no relevance to understanding actions of antidepressants in intact brain, and that perhaps other avenues or angles should be explored.

    Rather than becoming defensive and resorting to attack tactics, the author of the proposal would benefit more from the peer review that he solicited if he used these critiques (and many that keep getting echoed by others) to design and propose new experiments that would address these objections.

  9. Namnezia says

    So in other words, if an 11,000 – 30,000 fold greater dose is needed to see an effect in yeast, then yeast is probably not the best model system to model this effect, especially since so may other great model systems are available.

  10. says

    @leigh could be. But I suspect there are many other drugs in the 100 drug panel that have the same hydrophopbicity / charge properties, but don’t cluster with the psychoactive drugs. You’d have to ask Ethan for those details, or be a better chemist than me and look at the paper’s supplement.

  11. DaveUK says

    I’ve not read the paper, but what is the rationale for using yeast in this work? Wouldn’t a neuronal culture model work best for this?

  12. Evo cb says

    So, having been a mammalian physiology rig jockey and then switching over to Protist cell bio/ evo the major thing which has made itself clear to me is signalling is different in metazoans! If EP were truly an “evo” pharmacologist rather than someone who has dabbled with yeast he would appreciate that a three orders of mag. Difference suggests a different mech. For anybody who hasn’t considered this before yeast may be our nearest cousins but EP hasn’t shown that the detergent effects are not significant in amoebazoans or even anything further away from ophisthos. This badly thought through shite gives us evo cell biol folks a bad name in what may actually be a rare useful implementation of the new sequencing tech.

  13. says

    So after reading EP’s defense (well, counterattack) above, my thoughts:

    EP you need to learn how to deal with “trolls”. Regardless of if PP was trolling you before – and really those posts about you are written with his shtick – you should carefully consider your audience. You aren’t going to change his mind by accusing him of trolling, and the combative tone doesn’t sell to the rest of us who have no idea what’s going on science-wise. You ask a lot of “Why doesn’t X do Y” questions in there, which presumes that the rest of us have a clue what is happening. There’s also a lot of “Why didn’t you read this and this?”

    Skip all that. Just explain the concepts, calmly and clearly. In the space you’re pioneering, you have to be nonstop explaining your work from the very beginning to the very end so that new donors can get interested. This sort of thing doesn’t help that cause – because as a fellow scientist but one who probably isn’t going to read your papers, I didn’t understand what was going on in that reply to PP.

    The attitude “I’m happy to debate” should be replaced with “I’m happy to explain”.

  14. Takver says

    Herein lies a possible long-term problem. EP’s whole career plan requires the public to fund him via crowdsourcing. But the public who may want to donate to his research is by and large incapable of assessing if his work is solid or not. Experts who review grants are (even if flawed) still better positioned to figure out if this work is worth funding or not.


Leave a Reply

Your email address will not be published. Required fields are marked *