What Is Depression, Anyway?: The Synapse Hypothesis

I.

The problem with depression research isn’t that we don’t have any leads on what causes depression. It’s that we have so many leads on what causes depression that we don’t know what to do with all of them. For example:

1. Life adversity, like getting fired or breaking up with a partner, can make people depressed. The biological correlate of this seems to be the hypothalamic–pituitary–adrenal axis (HPA), where your brain tells your adrenal glands to produce glucocorticoid stress hormones like cortisol and this does something to your brain that increases the risk of depression.

2. Inflammation and immune overactivity can make people depressed. The classic examples of this are cancer-related depression (which exceeds what you would expect just from cancer being stressful) and depression induced by administration of the immunomodulator interferon-a. Antiinflammatory drugs have a small but clinically relevant antidepressant effect. Some of the relevant chemicals here seem to be TNF-A and IL-1; these do something to your brain that increases the risk of depression.

3. Serotonin and other monoamines seem to be involved. Most existing antidepressants, like SSRIs and MAOIs, seem to work by increasing monoamine levels. There are some conditions which affect monoamine levels and also increase risk of depression, though it’s nothing like a perfect correlation.

4. The glutamate system (eg NMDA and AMPA receptors) seem to be involved. Ketamine acts on both of these receptors in different ways, and one of those actions is the source of its rapid and unprecedented antidepressant effects.

5. There’s some kind of important link between depression and folate balance. Various folate-related chemicals (eg l-methylfolate and s-adenosylmethionine) are effective antidepressants. Some studies show that people with depression sometimes have disrupted folate cycles, for example elevated homocysteine levels.

6. Electroconvulsive therapy (“shock therapy”) is very effective at treating depression if it induces a seizure in the patient, so the increased activity from seizures must be helpful somehow.

So if we wanted to know what depression really was, it might be promising to look for some process that seems to match depressive symptoms and affects/is affected by life adversity, inflammation, monoamines, glutamate, folate, and electricity.

Recently some people think they’ve found one. According to Duman’s Neurobiology of Stress, Depression, and Rapid Acting Antidepressants, it’s decreased synaptogenesis, and it’s regulated by a protein complex called mTORC1.

Neurons communicate with other neurons through branches called dendrites and connections called synapses. Healthy neurons often create new dendrites and synapses to expand their network of connections and adjust to new information. The process of making new synapses is called “synaptogenesis”, and it’s common throughout the adult brain.

As mentioned above, depressed people have decreased volume in some brain areas. But in postmortem studies, they don’t actually have fewer cells in those areas. So it looks like maybe these neurons just have less synaptogenesis going on.

Synaptogenesis is partly controlled by a protein complex called mechanistic target of rapamycin complex 1 (mTORC1 to its friends). Like every other protein, mTORC is controlled by a giant mess of receptors and second messengers and intracellular signals with names like VDCC and GSK3.

People try to make this seem simple by displaying it as a system of billiard balls and tubes in a cute cartoon, but don’t be fooled – no human being has ever remembered any of it for more than two seconds.

The factors that affect synaptogenesis and mTORC are many of the same factors that affect depression. Let me count the ways:

1. Life adversity causes chronic stress, biologically represented by upregulation of the HPA axis and increased corticosteroid production. A 2008 study finds that rats who are subjected to chronic stress develop atrophy of dendrites in their prefrontal cortex. Administering glucocorticoids directly mimicked some of these effects, suggesting that stress is a whole cocktail of things including glucocorticoids and other things. When humans take glucocorticoids (they’re a useful medicine for various diseases) they tend to develop hippocampal atrophy and “simplification of dendrites” there, which I think is the same as decreased synaptogenesis. They also tend to get depressed – in some studies of Cushing’s Syndrome (the medical name for the collection of bad things that happen when you take too much glucocorticoid medication), up to 90% of patients are depressed.

2. I didn’t find the linked paper’s attempt to link inflammation to synaptogenesis very convincing, but it looks like there’s a little bit of research that has found that systemic inflammation decreases synaptogenesis. “Morphometric analysis of dendritic spines identified a period of vulnerability, manifested as a decrease in [dendritic] spine density in response to inflammation. The density of presynaptic excitatory terminals was similarly affected. When the systemic inflammation was extended from 24h to 8 days, the negative effects on the excitatory terminals were more pronounced and suggested a reduced excitatory drive.” This seems pretty relevant.

3. Everyone used to think that traditional antidepressants like SSRIs worked by increasing serotonin (and so by extension depression must have something to do with low serotonin levels). But SSRIs increase serotonin very quickly (within hours) yet take months to work. Something longer-term must happen when serotonin levels have been increased for long enough. That something has now been pretty conclusively identified as an increase in brain-derived neurotrophic factor (BDNF) – although I can’t find any good explanation of why increased serotonin should cause increased BDNF after a month. BDNF is a nerve growth factor – its main action is activating mTORC and telling nerve cells to grow more dendrites and synapses. And it’s most active in the cortex and hippocampus.

4. Ketamine affects the brain by either blocking NMDA receptors (boring traditional explanation), activating AMPA receptors (exciting new explanation), or possibly both (wishy-washy neoliberal compromise explanation). Duman et al are kind of ambiguous about which explanation they accept, but I think they present a theory where NMDA blockade causes AMPA activation, or something, which I’d never heard before. In any case, they present ample evidence that AMPA rapidly affects BDMF and dendritogenesis – for example, Positive AMPA Receptor Modulation Rapidly Stimulates BDNF Release And Increases Dendritic MRNA Translation. The “rapidly” part is important – the surprising thing about ketamine is how quickly it works compared to other antidepressants, so it’s exciting to find a theory that predicts this should happen.

5. I haven’t seen much attempt to fit folate into this theory, which is a shame. A quick Google search brings up a few people talking about how folate deficiency decreases neurogenesis in the hippocampus, which is sort of related.

6. Studies show that ECT increases BDNF levels and increases hippocampal volume, though I’m not sure exactly how or why giving someone a seizure should do that.

So the synapse hypothesis can unify at least five of the six lines of research into the causes of depression.

II.

My remaining skepticism is mostly based on a worry that anyone can do this with anything. The body is so interconnected, and there’s so much bad biology research out there, that I worry that if I said that the real cause of depression was, uh, thickness of the blood, I could find some way that all of those lines of research above affected blood thickness.

A quick demonstration: glucocorticoids can cause thicker blood, inflammation can cause thicker blood, SSRIs cause thinner blood, folate causes thinner blood. Huh, actually that’s kind of creepy.

My point isn’t that the (very respectable) academic research on depression is anywhere near this silly. It’s just to explain why I can hear a theory that seems to explain everything beautifully and my only reaction is “Eh, sounds like it has potential, let’s see what happens.”

Here are some of the things that confuse me, or that I hope get researched more in the future:

1. Why should decreased synaptogenesis cause depression, of all things? If you asked me, a non-neuroscientist, to guess what happens if the brain can’t create new synapses very well and loses hippocampal volume, I would say “your memory gets worse and you stop being able to learn new things”. But this doesn’t really happen in depression – even the subset of depressed people who get cognitive problems usually just have “pseudo-dementia” – they’re too depressed to put any effort into answering questions or doing intelligence tests. Why should decreased synaptogenesis in the hippocampus and prefrontal cortex cause poor mood, tiredness, and even suicidality? All that the Duman et al paper has to say about this is:

This reduction in dendrite complexity and synaptic connections could contribute to the decreased volume of PFC and hippocampus observed in depressed patients. Moreover, loss of synaptic connections could contribute to a functional disconnection and loss of normal control of mood and emotion in depression (Fig. 1). In particular, the medial PFC exerts top down control over other brain regions that regulate emotion and mood, most notably the amygdala, and loss of synaptic connections from PFC to this and other brain regions could thereby result in more labile mood and emotion, as well as cognitive deficits.

…which sounds more like an IOU for a theory than anything really fleshed out.

2. Why can’t we just give people BDNF for depression? I’ve been looking into this and it seems like the answer is something like “this works great if you cut open someone’s skull and inject it directly into their brain, but most people aren’t up for it” (the relevant studies were done in rats). But why can’t it be given peripherally? Some studies suggest it’s stable on injection and crosses the blood-brain barrier. Some people tried this in mice and got modest results, but why aren’t people looking into it more?

3. Why does the body have so many “decrease synaptogenesis” knobs? That is, why go through the trouble to evolve all these chemicals and systems whose job is to tell your brain to decrease synapse formation so much that you end up depressed? Is there some huge problem with having too much synapse formation which the brain is desperately trying to avoid? For that matter, what is it like to have too much synapse formation? If it’s the opposite of depression, it sounds kind of fun. If I got someone to open up my skull and inject a lot of BDNF, could I be really happy and energetic all the time? How come all the good stuff is always reserved for rats?

4. Why is depression an episodic disease? That is, how come so many people get depressed for no reason, stay depressed for a few months to a few years, and then get better – only to relapse back into depression a few years later? If people get depressed because of some life stressor like a divorce, how come they don’t get un-depressed once the life stressor goes away? Is depression some kind of attractor state? If so, why?

5. Why doesn’t rapamycin cause depression? Remember, mTORC is “mechanistic target of rapamycin”, so named because the drug rapamycin inhibits it. But we give people rapamycin for various things all the time, and depression isn’t really known as a major side effect (even though IIRC it crosses the blood-brain barrier). If depression is really under the immediate control of mTORC, rapamycin should be the most depressive thing. Instead it’s not obviously depressive at all.

6. How does bipolar disorder fit into all of this? Is mania the answer to my “what is it like to have too many synapses?” question from point (3)? If so, why do some people go back and forth between that and depression?

A lot of these questions could be answered in one stroke if we had a good evolutionary theory of depression. I’m skeptical that this exists – depression just seems too fitness-decreasing, and the various just-so stories people have come up with for why it might increase fitness in certain weird situations seem a little too convoluted. So it’s not that I’m expecting some sort of evolutionary story to work out. Just noticing that, even if the synapse theory of pathophysiology turns out to be right, there’s still a lot more that needs to be explained.