A new mechanism of action for Selective Serotonin Reuptake Inhibitors (SSRIs)?: Pharm 551A, Baudry et al., 2010

This is the last paper for the first section of the class and, as such, it serves as a transition from basic drug discovery work to the next section of the class, the pharmacology of gene expression. The paper, MiR-16 Targets the Serotonin Transporter: A New Facet for Adaptive Responses to Antidepressants, Baudry et al., 2010 Science, is a last minute addition for the class. I had another miRNA and pharmacology paper in this spot before but, when I saw this paper, I couldn’t help myself and switched the papers. Here’s why:

When I joined the Department of Pharmacology at UTHSCSA way back in the late 1990’s to start my PhD I did so for one reason — I wanted to study the mechanism of action of anti-depressants. In my undergrad days at UT Dallas I took a course called “Neuropharmacology” that was my first introduction to pharmacology. For a then physics major, this course was a real revelation and the chapter on serotonin (5HT) in the book for the course — Basic Neurochemistry — was hands down my favorite. That chapter was written by Alan Frazer and Julie Hensler, both of whom were (and still are) at UTHSCSA (Frazer is the chair of the pharmacology department). Hence, I decided that I was going to do my PhD in that department and my hope was to work in Alan Frazer’s lab. While I did do a rotation in Alan’s lab, I ultimately decided that I was more interested in cannabinoid pharmacology and ended up doing my PhD work in a different lab; however, I retained interest in the mechanism of action of antidepressants and I continued to follow the work of people like Alan, Julie and David Morilak (also in the same department).

Sometime between 2002 and 2003 (I can’t remember) I heard a rumor that someone had figured out that SSRIs (drugs that selectively inhibit the 5HT reuptake transporter — also called SERT) caused an increase in expression for a non-coding RNA that decreased expression of SERT. This was a very exciting rumor because even though SSRIs were used widely for the treatment of depression, no one had been able to solve a particularly perplexing feature of SSRI pharmacology. The mystery was (is) that even though SSRIs were potent and efficacious inhibitors of SERT both in vivo and in vitro, it takes them a couple of weeks to start working for depression. This suggests that in addition to their direct effect on blockade of SERT function they possess some additional effect on gene expression that is an important feature of their clinical benefit. The labs of Alan Frazer and David Morilak provided some important clues to this potential mechanism of action. First, long-term treatment with SSRIs promotes a strong downregulation of SERT protein expression (as measured by 5HT reuptake and SSRI binding sites in brain) but has no effect on SERT mRNA expression. Moreover, this downregulation of SERT protein appears to be the major mechanism through which clinically relevant benefits occur as this downregulation leads to a greater deficit in 5HT clearance than can be obtained with acute blockade of the transporter with SSRIs. The problem was that a mechanism for this downregulation of SERT protein, but not mRNA, had not been proposed. The rumor that a non-coding RNA was responsible for this effect was alluring because it had the potential to explain this differential effect on translation vs. transcription.

Well, the rumor came and no paper materialized. Year after year, in fact. Until now. A few weeks ago the paper we will discuss today became available online at Science allowing me, after all these years, to see what this potential mechanism is all about. I should say that I have no confirmation that this is the work that inspired that rumor all those years ago. I also don’t know if that rumor was even true. However, we’ll proceed under the assumption that this is the work that all the fuss was about… only because I prefer that story.

Okay, onto the paper. The authors start out with the conundrum posed by the work from the Frazer and Morilak labs in the papers linked above:

The distribution of SERT in the brain mirrors that of serotonergic neuronal cell bodies and innervating fibers (4, 5). Serotonergic raphe neurons project to most parts of the central nervous system and coordinate the physiology of the whole brain (2). Chronic SSRI antidepressant treatment promotes reductions in SERT binding and protein levels but does not affect SERT mRNA levels (6), suggesting that SSRIs may interfere with SERT translation. This control of translation could be exerted by microRNAs (miRNAs), which have emerged as crucial modulators of gene expression (7, 8). Although the roles of miRNAs in cell fate decision, differentiation, maintenance of cell identity, survival, and neuronal plasticity are being uncovered (9, 10), their targets remain largely unknown.

To investigate whether miRNAs provide a mechanism for adaptive changes in SERT expression in monoaminergic neurons, we first exploited the 1C11 neuroectodermal cell line, which can differentiate into either serotonergic (1C115-HT) or noradrenergic (1C11NE) neuronal cells (fig. S1A) (11). 1C11 neuroectodermal cells express transcripts encoding SERT and neurotransmitter-related markers before their choice of cell fate (Fig. 1A and fig. S1B). Because these transcripts remain at roughly constant levels during serotonergic or noradrenergic differentiation, miRNAs may participate in the posttranscriptional mechanisms that prevent illegitimate mRNA translation according to each program.

The authors perform some pretty clever experiments with these 1C11 cells to kick off the paper. In a nutshell, they show that miR-16 suppresses a 5HT phenotype in 1C11 cells that take on a norepinephrine (NE) fate. In cells that become noradrenergic, if miR-16 function is blocked by an anti-miR-16 construct, the cells then take on a 5HT phenotype revealed by increased 5HT content, expression of SERT, ability to degrade 5HT and expression of 5HT2B receptors. This, in and of itself, is a really fascinating finding insofar as it suggests that NE neurons in the brain may be able to switch to a 5HT phenotype if miR-16 expression is modified. Showing that this happens in vivo is the major focus for the rest of the paper.

To show this, the authors focus on the major 5HT and NE areas of the brain, the raphe nucleus (5HT) and the locus ceruleus (LC). They demonstrate that SSRI administration into directly into the raphe increases miR-16 expression and decreases SERT binding sites. This effect is recapitulated by direct administration of miR-16 and blocked by anti-miR-16 constructs. Additionally they do several experiments that suggest that SSRIs suppress WNT function and that this leads to the maturation of miR-16. This suggest a model, in the raphe, where tonic WNT activity prevents the maturation of pre-existing pre/pri miR-16. Through suppression of the WNT pathway (the mechanism for this is unclear) SSRIs lead to a maturation of miR-16 which in turn decreases SERT protein expression.

That’s what happens in the raphe, how about the locus ceruleus? This is where the story really gets interesting. SSRI administration into the raphe decreases miR-16 expression in the locus ceruleus, increases SERT expression, induces 5HT synthesizing capacity and increases 5HT2B receptor expression. These findings indicate that SSRI administration into the raphe is capable of switching the phenotype of locus ceruleus neurons toward a 5HT phenotype. How does this happen?

The question then arises of how the response of serotonergic neurons to fluoxetine treatment is relayed to noradrenergic neurons in vivo. Reciprocal connections exist between these two brainstem monoaminergic nuclei, thus supporting communication between the two systems (16). Recently, the expression of miR-16 in monocytes was shown to be down-regulated by S100β (17), a neurotrophic protein that is up-regulated by fluoxetine treatment (18). We therefore hypothesized that the secretion of S100β increases upon exposure of raphe to fluoxetine and that this protein acts as a paracrine factor to promote the reduction in miR-16 in the locus coeruleus, in turn unlocking the expression of serotonergic functions. We first exposed 1C115-HT cells to fluoxetine and observed an accumulation of S100β in the culture medium (Fig. 4A). Although the addition of S100β slightly decreased miR-16 levels in these serotonergic cells (Fig. 4B), it did not affect SERT expression (Fig. 4C), which is in agreement with the lack of impact of miR-16 silencing on SERT in 1C115-HT cells (Fig. 1B). A larger decrease (43% of control level) of miR-16 was seen in 1C11NE cells exposed to S100β (Fig. 4B), which correlated with the appearance of SERT (Fig. 4C). In addition, after S100β treatment, 1C11NE cells acquired the ability to synthesize and store 5-HT (Fig. 4, D and E) and to express 5-HT2B receptors (Fig. 4F). These data thus validate our working hypothesis on an in vitro level. We then measured the level of S100β in raphe upon infusion of fluoxetine. Fluoxetine up-regulated S100β levels in serotonergic nuclei (133% versus control) (Fig. 5A). Further, injection of S100β into the locus coeruleus decreased (by 22.4%) miR-16 levels and turned on the expression of SERT (Fig. 5, B and C). Finally, antibody-mediated neutralization of S100β in the locus coeruleus prevented the decrease in miR-16 levels observed upon infusion of fluoxetine in raphe (Fig. 5D). In addition, the decrease in miR-16 and the onset of SERT expression observed in the locus coeruleus, upon systemic fluoxetine treatment, were both eliminated by small interfering RNA–mediated knockdown of S100β in raphe (fig. S8, A and B). The data from 1C115-HT cells (Fig. 4A) and the innervation of the locus coeruleus by raphe fibers (16) strengthen the hypothesis that secretion of S100β by serotonergic neurons, at the locus coeruleus, mediates the action of fluoxetine. Secretion of S100β by glial cells in the raphe is less likely to promote a long-range action on the locus coeruleus.

So, in their 1C11 cells, an SSRI increases expression of S100beta . S100beta decreases miR-16 expression in both 1C11 5HT and NE cells but S100beta has no effect on SERT expression in 1C11 5HT cells while it induces a 5HT phenotype in 1C11 NE cells. In vivo, SSRI administration into the raphe causes an increase in S100beta expression. S100beta administration directly into the locus ceruleus decreases miR-16 and increases SERT expression suggesting that S100beta is responsible for the switch to a 5HT phenotype in the locus ceruleus.

Does all of this have an impact on a behavioral output relevant to the mechanism of action of antidepressants? To show this the authors turn to the unpredictable chronic mild stress (UCMS) paradigm. Using four measures (coat quality, body weight loss, sucrose preference and locomotor activity), the authors show that three treatments have an antidepressant effect: 1) SSRI into the raphe (no surprise), 2) direct miR-16 administration into the raphe and 3) anti-miR16 administration into the locus ceruleus.

This leaves us with the following model for a novel mechanism of action for SSRIs:
1) SSRIs suppress WNT signaling in the raphe leading to a maturation of miR-16. This causes an increase in expression of S100beta and a decrease in SERT expression in these 5HT neurons.
2) SSRIs increase S100beta release in the locus ceruleus causing a decrease in miR-16 expression in these NE neurons. That decrease in miR-16 expression releases a suppression of 5HT function in those neurons leading to an increase in 5HT synthesis, increased SERT expression and 5HT2B receptor expression.
3) Presumably this causes a sort of double whammy for brain 5HT levels. Less SERT expression in the raphe leads to greater 5HT levels because reuptake mechanisms are decreased. In combination with this, locus ceruleus neurons take on a 5HT phenotype thereby increasing 5HT levels through the activation of a novel 5HT pathway in the CNS.

I have to admit, in some ways this is almost too amazing to be true so it will be very interesting to see what lines of research emerge from this incredible story. On the other hand, the data makes really goes a long way towards explaining some of the more perplexing clinical features of SSRI function. The story fits very well with the lag in clinical benefit for SSRIs, for instance. We’ll see where this goes but I can guarantee one thing, I’ll be watching closely as I continue to think that this is one of the most fascinating stories (SSRI mechanism of action, that is) in the neuroscience arena. I just wish I could have Alan, Julie and David in class with me today. I would love to hear their comments!

Baudry, A., Mouillet-Richard, S., Schneider, B., Launay, J., & Kellermann, O. (2010). MiR-16 Targets the Serotonin Transporter: A New Facet for Adaptive Responses to Antidepressants Science, 329 (5998), 1537-1541 DOI: 10.1126/science.1193692