How do antidepressants actually work?

February 13, 2013

The Abraham in the genealogy of antidepressants is a synthetic compound called imipramine. As shown below, imipramine belongs to an ancient line of compounds that dates back to the dawn of psychopharmacology, when neurotransmitters ruled the Earth:



A conserved ethylamine side chain (in red) pervades the psychopharmacopoeia


Discovered in the early 1950s, imipramine is the original tricyclic antidepressant, and it begat second- and third-generation antidepressants with fewer side effects but not necessarily greater efficacy, including the selective-serotonin reuptake inhibitors (SSRIs): Zoloft, Prozac, and Paxil, among others.


So how do antidepressants work at a molecular level? Which cellular bits do they bind to? You’d think we’d know the answer to that question after more than 50 years of antidepressant use. As suggested by the name SSRI, antidepressants inhibit with different degrees of selectivity the reuptake of monoamine neurotransmitter transporter proteins at synapses in the brain. Neurotransmitter transporter proteins include SERT, which is specific for serotonin, or 5-hydroxytryptamine (5-HT). Evolution has conserved the SERT gene in mammals, flies and even nematode worms. The human version of SERT is thought to be the primary drug target of SSRI antidepressants. In other words, SSRIs bind to SERT expressed on the surface of serotonin-producing neurons and then inhibit serotonin reuptake into cells.


But there’s a catch. The serotonin-centric model has an exposed, soft underbelly. In the mid-90s, psychopharmacologists Eric Nestler and Ronald Duman and others showed that antidepressants induce neurogenesis in laboratory rodents after chronic but not acute treatment. The chronic vs acute part is key. Inhibition of SERT by antidepressants occurs within minutes to hours, yet people who respond to antidepressants don’t feel better until they’ve been taking them for weeks or months. So the rodent observation meshes with long-standing clinical observations in people taking antidepressants who experience a “therapeutic lag.”


For the record, acute is defined as 30 minutes after injection of antidepressant. One of the behavioral assays employed in the antidepressant R&D pipeline is the tail suspension test. Briefly, mice are suspended upside down by their tails. As you or I would if subjected to a comparable stress position, inverted mice struggle to right themselves but eventually submit to gravity. Turns out mice treated with antidepressants fidget longer than their placebo-treated compatriots. This difference between treated and untreated mice has traditionally been the green light to move a candidate antidepressant onto human clinical trials.


In 2011, Randy Blakely’s lab at Vanderbilt published an insightful paper in PNAS that informs any serious discussion of antidepressant pharmacology, which is clearly complex. The Blakely Lab has a unique place in the history of antidepressant research. Blakely led the team that first cloned SERT from rats in 1991, validating the crude extract-based neurotransmitter reuptake assays that were originally employed by the pharmaceutical industry in the 1960s and 1970s in R&D and that ultimately yielded the SSRIs. With the 2011 PNAS paper, Blakely’s efforts came full circle. He and his group found that acute behavioral effects of antidepressants did not occur in a special SERT mutant mouse they genetically engineered.


This mutant mouse’s SERT gene was mutated at a single amino acid position in the SERT protein. As a result, this one mutation rendered the mouse SERT protein unable to be bound by many antidepressants but still able to bind and transport serotonin, its natural substrate. As such, this mSERT mutant mouse is a clean genetic test of the role of serotonin reuptake in the complex pharmacology of antidepressants.


As shown in Figure 1A of their paper, mutant SERT (M172) interacts with and transports serotonin, or 5-HT, just as well as wildtype (I172). However, several SSRI antidepressants (but interestingly not paroxetine/Paxil), as well as cocaine, must be used at higher concentrations to block serotonin transport by mutant SERT. In one case (E),  up to 1000-times more citalopram/Celexa is required:


fig 1


The single amino acid change, which swapped an isoleucine for a methionine at position 172, was not made by serendipity but on the basis of data in a 2007 Nature paper by Eric Gouaux, whose lab is at Oregon Health & Science University. The Gouaux Lab studies neurotransmitter transporter proteins. However, the Gouaux Lab couldn’t crystallize a SERT protein of mammalian origin, but they were able to crystallize an old relative of mammalian SERT – from bacteria! Bacteria don’t produce serotonin, but they express a distantly related yet structurally similar transporter that recognizes the amino acid leucine (see featured image).


Blakely’s group has demonstrated that SERT is required for the acute behavioral effects of antidepressants in laboratory rodents. So here’s the $64,000 question: is SERT required for neurogenesis in laboratory rodents after chronic antidepressant treatment?


Two years later this experiment still hasn’t been performed to my knowledge — what are we waiting for?



  • DanielHorowitz

    I don’t think Big pharma wants us to know the answer

    • Ethan Perlstein

      I don’t think we need their permission. :)

  • bill_foster

    As a scientist suffering from depression this is quite interesting. But what is more interesting is your $64,000 question. Easy enough experiments and 2 years is plenty of time. Whatever the reasons I think this is an example of the inefficiency of academic science in its current state. If science were more open and not so competitive between investigators, I could see the following scenario: Perhaps such a study is currently being worked on or currently funding is being sought. But only the PIs pursuing such work and the study sections would know this. If science were more open and collaborative, I could see perhaps a blog, or a forum, where it is known that such a study is taking place, where the investigators show results of their experiments and elicit feedback from the community. There may be many people who would have no interest in competing for doing this experiment, however there may be many who see the importance of such research and could help propose ideas for experiments, interpret experiments, troubleshoot, etc. Science needs to become more open and perhaps some sort of crowdsourcing could be beneficial.

    • Ethan Perlstein

      Thanks for the great comment, Bill!

      I really dig the idea of an online clearinghouse for experimental suggestions that most experts in the field would agree is a good idea, but that never get done because of hyperspecialization or limited resources or because the idea is for whatever reason dismissed out of hand.

      In fairness to Randy Blakely, when I spoke to him on the phone last year I brought up this very point about the requirement of SERT in neurogenesis. He sounded amenable. So maybe the experiments are in progress, or better still the manuscript is being shopped around as we speak.

      I was thinking of crowdfunding the experiment and using a site like Science Exchange to outsource the actual experimentation. I think it could be done for $64,000 or less!

  • Andrew G

    How is PK/PD of these molecules established? Toward PK, how long after injection or ingestion does it take for the drug to reach the synapse? Toward PD, how long must the drug be present for the SERT dysfunction to have a physiological or behavioral consequence? Unrelated question: are you certain the $64,000 question has not already been conducted?

    • Ethan Perlstein

      SSRIs inhibit SERT right away, that’s why you can see acute behavioral effects within in 30 minutes of administration.

      If the role of SERT in neurogenesis has been tested in using Blakely’s M172 mouse, then I haven’t seen the paper. By all means point me to it!

  • Randy Bakely

    Hi guys

    Time for me to weigh in with a few comments. First, thanks Ethan for taking the time to read over and comment (favorably!) on our work. All scientists like to know that someone cares! Here are a few clarifications and response to points raised about the use of the animals for stem cell studies.

    1) I too consider the delayed aspects of SSR action to be something worth working hard on, chiefly because of the the pain and suffering and ultimately withdrawal from treatment for those who either do not respond or where the response isn’t quick enough to deter tragic consequences. I am the child of such an event and it motivates me to push as hard as I can. However, we need to be careful in connecting the lack of a delay with mechanisms that are necessarily independent of serotonin action. Depression is a disorder, we believe, where cellular and circuit level perturbations have arisen, even since childhood, and the actions of serotonin to alter these perturbations may simply take constant action for a period to time to induce critical changes in brain networks that can ultimately relieve symptoms. We hope our new transgenic mouse model can shine enough light on this issue to steer the discourse in a more informed way going forward, to the degree that mouse studies inform the argument (I believe they can).

    2) The identification of the I172M substitution that eliminates SSRI sensitivity at SERT was not found on the basis of the crystal structure of LeuT. Actually, our work the year before the LeuT structure was published predicted that I172 and another residue Y95 (that also impacts SSRi interactions) would be located at the binding pocket for serotonin. When the structure of LeuT was determined (a watershed moment in our field), modeling SERT on that structure confirmed our hypothesis and provided the best evidence that the LeuT structure has relevance for mammalian transporters (though not cited by the authors. Such as it goes in Science, spilt milk, still a lovely, groundbreaking piece of work). Here is the citation to our study and a link where the paper can be obtained.

    Tyr-95 and Ile-172 in transmembrane segments 1 and 3 of human serotonin transporters interact to establish high affinity recognition of antidepressants. Henry LK, Field JR, Adkins EM, Parnas ML, Vaughan RA, Zou MF, Newman AH, Blakely RD.
    J Biol Chem. 2006 Jan 27;281(4):2012-23. Epub 2005 Nov 3.

    PMID: 16272152

    3) As to moving forward with stem cell studies, I agree wholeheartedly that this is a key direction for future efforts with the mice and actually is embedded in a proposal to the NIH of my graduate student Alex Nackenoff. Alex has already initiated this work and hopefully we can present the fruits of his labors in a public forum soon (controls, controls, controls). There have been some delays (its just been over a year though since we published the mice). First, I had to recruit a terrific graduate student to chase these questions for his thesis studies. Done. Second, the mouse model in our 2011 paper has a genetic background (129S6/S4 for the aficionados) that does not perform as well in the standard tests of SSRI action (we note this in our paper). Therefore, we had to cross these mice with wild type C57BL/6 mice so that the I172M SERT substitution is on a more hospitable genetic background. That typically takes 10 generations of breeding. A generation is 3weeks of gestation plus 8 weeks to be able to breed again. As fast as you can go with this strategy then takes 11×10 weeks, >2 yrs. There is a more efficient process now that allows one to go faster, which we used, and this cuts the time in half, but costs ~$10,000 through our service. Having now crossed the mice, we now had to validate the model hadn’t changed in critical ways and that meant retesting the animals for SSRI insensitivity. Now that has been done and our studies are underway. I wish we could have moved faster, but it just was not possible.

    As it was, it cost us $$$ we did not have. I would love crowd source based funding for our efforts. Donations to accelerate our efforts or prevent us from shutting it all down if we go over the cliff can be sent to me at Vanderbilt University (tongue in cheek, sort of).

    Stay tuned. Thanks


    • Ethan Perlstein

      Randy, thanks for the generous reply! It’s of the quality one typically sees in closed peer review, not on open comment threads. I’m honored that you put such time and thought into it. Chime in any time on my site. :)

      Thanks for setting the record straight on LeuT (sorry to hear you weren’t cited, but it happens all the time). I also appreciate the explanation for why it’s taken over two years just to set up the conditions to do the neurogenesis experiments the right way.

      As far as crowdfunding goes, I’d love to collaborate on antidepressant pharmacology! The exciting albeit enormous challenge of integrating all the cellular effects of antidepressants on the brain requires more funding and collaboration, to be sure.