SAD and Inositol.

Over a month ago I mentioned I was treated for SIBO.   If it helped the digestive issues, it was subtle.  I am however pretty sure it gave me Seasonal Affective Disorder.

Let me give you some context.   I grew up in Rochester, NY (165 sunny days/year).  I went to college in Ithaca, NY (152 sunny days/year, but less snow than Rochester).  I’ve lived for Seattle (152 sunny days/year) for almost 8 years.    Winter weather might keep me indoors more, and I do need vitamin supplements year round, but that’s because wet socks are unpleasant and I’m bad at metabolism.  I didn’t have any SAD symptoms leading up to starting treatment for SIBO, which happened to be the day after winter solstice.

The treatment for SIBO for me was two antibiotics, erythromycin and xifaxan.   Two or three days after I started, I felt fine during the day, but as soon as the sun went down it felt like the world was ending.  It felt Late as soon as it was dark, which was 4:30 PM at the time.  As time went on, I got more and more emotionally distraught and depressed.  Everything felt awful.

1.5 weeks in, I noticed this, upped my vitamin D and started using a sun lamp, and that helped.  2 weeks in the treatment naturally ended, and I felt better still.  But not all the way better.

Finally, almost six weeks after I’d stopped antibiotics, I remembered a friend telling me about inositol, which is a carbohydrate used for intercellular communication.  The conventional wisdom is that your body can naturally manufacture enough inositol from glucose that nutritional sources are irrelevant: however, there’s some evidence that it’s either made or affected by your intestinal flora.*  I’d tried it when he suggested it and found it had no effect, but kept the bottle just in case.  I gave it another shot, and felt better the next day.  2 weeks in, I feel like the SAD is completely gone.

Right on the web page, there’s a warning that Xifaxan can cause an overgrowth of Clostridium difficile.  In the study my friend described but did not give me a proper citation for, he said the researchers had isolated six different bacteria that competed with C. difficile, one of which produced inositol.  I cannot find this study, or even a news article, anywhere.

It’s hardly proven, but I have a strong hypothesis that the antibiotics screwed up my intestinal flora (which is, in fact, what they were supposed to do, we were just hoping to localize the effects to the small intestine), leading to an inositol deficiency, leading to SAD.  In many ways my digestive system feels like it’s been bumped back to earlier stages of treatment (the HCl supplements and removing some food groups), which makes me think that some of the things I experienced were second order effects of changing intestinal flora, rather than my diet directly.

What does it mean to say a gene causes something?

When writing about the genetics of sensory processing sensitivity, I’ve tried to be careful to use the phrase “associated with” rather than “causes.”  I do this because genes don’t directly code for anything.  To a first approximation, genes either code for proteins or for regulation of when those protein coding genes are expressed.*  Different alleles of the same gene code for different versions of the gene’s protein**.

Sometimes, the link between a change in the DNA and the phenotypic outcome is obvious.  To return to perennial favorite sickle cell anemia:  there are several different alleles that can cause sickle cell anemia (which appear to have arisen independently, adding even more weight to the hypothesis that the alleles are adaptive), but they all affect the hemoglobin protein.  A change in one nucleotide leads to a single amino acid change in the protein. Under certain conditions, this leads red blood cells to take the characteristic sickle shape, which causes a pile up in the blood vessels, which causes oxygen deprivation.  All of these things are easy to observe (comparatively) and easy to track the chain of causation.  And even with that, it’s not quite fair to call it “the sickle cell allele”, because the same allele also codes for malaria resistance.

The genes associated with high sensitivity don’t do anything nearly so obvious.
DRD4 7R (the mutation associated with high sensory sensitivity, ADHD, greater susceptibility to maternal trauma, and nomadicism) occurs in the DRD4 gene, which produce dopamine receptor D4.  It does not lead to an amino acid substitution: rather, there’s a string of amino acids that may be repeated somewhere between 2 and 11 times: the 7R allele codes for 7 of them.  I am unable to find any evidence we know what this does to the tertiary (three dimensional) shape of the protein or even where the D4 receptor is typically found, much less how these mutations affect metabolism.  There’s a lot of high level correlational studies about various mutations and various phenotypic pathologies, like schizophrenia and parkinsons, but there isn’t a nice neat chain of causation like we see in sickle cell.

The other gene I talked about,  5-HTTLPR, isn’t even a real gene.  It’s a promoter region for SLC6A4which codes for a protein that transports serotonin.  That means that mutations in 5-HTTLPR can affect when, where, and how much serotonin transporter is produced, but not the amino acid sequence of the transporter.

What are the phenotypic consequences of slight variations in the expression of this gene?  That’s a really good question.  It could theoretically affect anything that is affected by serotonin, whose  clients include “the digestive system” and “the central nervous system.” And if it’s affecting your digestive system, it could affect your nutrition. So pretty much anything, ever, could plausibly be affected by mutations in this area.

The DRD4 gene doesn’t code for parkinson’s disease.  It codes for a protein.  Any variation in that protein will have a multitude of consequences, one of which might be parksinson’s.  There is no one gene for high sensitivity: there’s a number of genes that influence a number of traits, one of which may be sensory sensitivity.  So please remember that if someone tells you gene foo codes for happiness, they are not your friend.

*There are exceptions, but they’re very complicated and don’t change what I’m about to say.

**This is a simplification.  It’s possible to have two different DNA sequences code for identical proteins and yet lead to slightly different outcomes, because is slower or more error prone to transcribe.

If high sensitivity is so costly, why is it still around?

Evolution is often sold as a species/population converging on a single ideal solution for a population.  This is incorrect.  For one, the ideal solution in 2013 is not necessarily the ideal for 2014.  Peter and Rosemary Grant demonstrated that ground finch weight and beak size varied from year to year, that this corresponded to shifts in food availability due to weather, and that the change was driven by genetics, not starvation.  Beyond that, there is often more than one good strategy at a time.  Wal-Mart and Rodeo Drive both technically sell clothes, but they’re not in competition.  When you combine temporal variation with the existence of multiple viable strategies, you get population diversity.

What this means is that even if a gene is very, very bad, it’s probably there for a reason.  At worst, it was a local optimum for a problem that no longer exists.  See: sickle cell anemia.  Sickle cell is a very nasty disease caused by a single gene.  But being heterozygnous for the sickle-cell allele gives you some resistance to malaria.  Which is useless if you live in the USA in the 2000s, but highly relevant if you’re in Africa now, and even more so Africa in the past.  That is why the allele stuck around.

Moreover, the returns to a particular strategy depend on other people’s strategies.  You’ll make more money selling cheap, mass-market clothes if there’s no Wal-Mart in your city.  This is known as frequency dependent selection.

The moral of this story is: any time you see a genetic trait that looks negative but is widespread, look around.  It’s almost certainly either linked to an advantage you aren’t seeing, or is advantageous under a different set of circumstances.

As we talked about previously, this appears to be the case for the trait known as sensory processing sensitivity .   Belsky et al have an exhaustive synthesis of various ways high sensitivity has been shown to correlative with both unusually positive and unusually negative outcomes, depending on environment.  You remember the DRD4 7R mutation, that was associated with ADHD and susceptibility to the individual’s mother’s trauma?  More prevalent in nomadic and nomadic-descended populations than sedentary.  And carriers do better than non-carriers when nomadic but worse when sedentary.

I tripped a little bit reading this, because something associated with ADHD and nomadicism seems like a novelty seeking gene, which is associated with extroversion.  High sensitivity is highly correlated with introversion, and even the extroverts among them have very different brain patterns than low sensitivity extroverts.  I got this from Quiet, which the library has unfortunately taken back so I can’t look up the specifics.  I wonder if the extraverts were the ones who had good childhood environment.  And it turns out this isn’t the only allele linked with both high sensitivity and ADHD.  5-HTTLPR short is too.

Quantifying the advantages of high sensitivity in humans is hard.  Luckily, there’s a number of very good animal models.  That’s a severe understatement.  A more accurate statement would be “you can’t throw a rock without hitting a member of a species that demonstrates a shy-bold continuum.”  This may mean we haven’t defined our terms well enough, but it could also mean that almost all species demonstrate frequency dependent selection in sensory sensitivity.  The theme seems to be that higher sensitivity animals are more cautious, and thus get eaten less, but every once in a while a low sensitivity animal blunders its way into something amazing, like a new source of food, or founding a start up.

What we talk about when we talk about heritability

There’s some behavioral evolution stuff I want to talk about soon.  This is great, because it’s what I studied in college and I will be able to contribute something beyond wikipedia synthesis.  In order to really dig into this, I have to explain how we measure heritability.

Heritability“, or h^2, is a very specific measurement in biology.  It measures how much  variation of a particular trait in a particular population is due to genetic variation, relative to the total variation.  It is not a synonym for “genetically determined.”  For example, having two arms is barely heritable, because there’s almost no variation in the number of arms people have.  Of people that have less than two arms, most of them lost them to environmental accidents, not genetics.  There’s no gene for “one arm”, although developmental accidents may have some genetic basis.  Biological sex isn’t very heritable either.  A person’s parents’ sex has very little predictive power on the person’s own sex.    Despite this, number of arms and biological sex is very, very genetically determined

Additionally, a heritability measurement is only valid for the population it was measured in.  In a population in which every individual has exact the same environment will demonstrate much higher heritability for every trait than a population with a varied environment.

Now here is the really weird thing: how sensitive a trait expression is to environmental variation can be influenced by genetics.  High Sensitivity was originally discovered in people who were having severely maladaptive responses to normal stimuli, and was assumed to have a uniform negative effect on people who carried it.  Newer research indicates that the effect is more complicated:  Highly Sensitive People with really good environments do better that Low Sensitive People in those environments.  HSPs in bad environments do worse that LSPs.  Whatever genes cause the trait we call High Sensitivity make a person more sensitive to their environment.  Ellis et al call this biological sensitivity to context.

Lee et al  investigated a particular HSP- associated allele, DRD2 Taq1A, and found that mothers with the gene were harsher parents during a recession than mothers without the gene.  Moreover, they found worsening economic conditions led to worse parenting in everyone, but the effect was more pronounced for mothers with the T allele as opposed to the CC allele.  So a child’s outcome is being affected by their own genetically-driven response to their parents, their parents’ genetically-driven response to the environment, and the actual environment.

I thought what depressed me here was that highly sensitive parents are more likely to have highly sensitive children, which will magnify the negative effects of the recession on children.  Then I read a study showing that maternal psychological trauma was associated with higher disorganization in infants, but only if they had a high-SPS associated allele.*

High Sensitivity is going to come up a lot, so remember that.  But also remember that when journalists talk about heritability, they do not mean what they think they mean.

*The same mutation, DRD4 7-repeat polymorphism, is associated with ADHD.

Sensory input and economies.

Expanding on my metaphor from yesterday, let’s posit three different types of sensory inputs/equations: Trivial (sensory input that is not necessarily effortless to process, but not taxing), Unsolvable (puts you into overload), and Solvable (requires noticeable amounts of processing power).  Unsolvable equations represent either completely  new data, or complex combinations of problems that would be Trivial on their own, but combine in difficult ways.

More concretely:  let’s say you have two sensations, A and B.  Either one on its own is a Taxing problem, but together they’re Unsolvable.  If you manage to move A into the Trivial category (by repeated exposure and processing), that moves A+B into Solvable.  And when you have solved it, you have A, B, and A+B in the trivial category, which will come in handy when you see A+B+C later.  In this way, very small differences in initial processing capabilities can compound into vastly different levels of ability.  Going into overload isn’t just painful, it cuts you off from learning the things that could prevent it next time.

This suggests a huge payoff to investing problems that are right on the border between Trivial and Solvable.  Not only do you move that one thing off your plate, and move some other problems from Unsolvable into Solvable, you create an environment where you can eventually solve the other components of that previously Unsolvable problem.  

[The following paragraphs are free-association at best]

For some reason I’m reminded of Jane Jacob’s The Economy of Cities, in which she postulates that most economic growth follows the following pattern:

  1. I’m doing a thing.
  2. Hurray, I invented a slightly more efficient way to do that thing.
  3. Wait, I could use that new invention to do this other thing.  There’s no demand for it yet because no one knows they want it, but there will be.
  4. Welp, I invited a whole new sector of the economy.

Tim Harford implicitly talked about this in Adapt.  The best economics have several interlocking industries with moderate overlap.  Similar enough that innovations in one can help enough, but dissimilar enough to cause them to approach problems in different ways.  Adapt as a whole is about trying a lot of things quickly, with the expectation that most of them will fail.  One important component of that is minimizing the cost of failure, and one important component of that is recognizing failure quickly.  Which sounds like a sensory problem.

Like I said, this is free association.    I know I’m on to something, but I’m not sure what yet.

I figured it out.

One thing that has frustrated me as I researched SPS has been that I couldn’t connect the macro with the micro.  I know damn well what the macro pattern is (too much noise -> everything is terrible, and also I bump into walls a lot), and I’ve been learning about the micro pattern (auditory, visual, or touch stimulus leads to over-activation of certain areas of the brain), but how did overactivation lead to irritability?    The Threads of Autism (which I’ve since finished) went on and on about teaching people to “organize” sensory input, but it only described the effects of this in macro language (people become more relaxed, more resilient, able to thrive under higher levels of stimulus), but not what that actually looked like in the brain.  Much less why pronating and supinating my wrists in time with my breathing or tapping along my facial nerves was going to accomplish this.

I think I’ve figured it out.  Before I tell you, I want to make it very clear that this is my own personal metaphor, and not something that’s been tested scientifically.  I’m not even sure what it would mean to test it.  But it makes sense to me.

Your/your body’s ultimate goal is to figure out what it should be doing at any given second, and especially if it should be running away from or trying to breed with something.  In order to do that, it tries to work backwards from the sensory input to derive a model of what is actually happening*.  It can then decide how it wants to respond to that thing that is happening.  For example, if you’re wandering through the jungle and hear a twig snap, you would like to know if it’s a delicious herbivore, a hungry tiger, or a human being, and if so, is it from  your tribe or the one you’re at war with.  Past sensory experience is really helpful in this interpretation.  For example, if you’ve heard lots of tigers walking in the jungle, you can pattern match the current sounds against the ones you remember and see how close they are.

But what if it were more basic than that?  We think we just know when and where and how someone is touching us, or how we’re oriented in space, but that is actually something you have to learn.  We don’t notice because it’s mostly done when when we’re very very young, and because it’s done in parts of our brain that we’re not  consciously aware of.  But human brains are actually very plastic, and we devote an extraordinary amount of time and energy to learning how to translate “nerve 43b is firing” to “something happened on my left ring finger.”

My metaphor is as follows:  people with SPS either don’t have the same bank of experiences to pattern match against, or are worse at matching.  So given the same amount of sensory input, it takes them a lot more energy to correctly model the source.  It’s sort of like simplifying a mathematical equation.  If you have something awful with lots of terms you can solve it by hand, but it’s error prone and time consuming.  If you take that same equation and simplify it by removing terms that cancel and grouping like terms together, that same equation can be trivial.

I think the point of the sensory integration exercises is to build up either the database of experiences and/or your skill at simplifying equations.  You can’t give a person every single experience, but you can teach them “this is what stimulus from the trigeminal nerve feels like deep inside your brain.”  This makes it marginally easier to identify, or at least not freak out about, novel stimulus.  It’s like teaching someone about a 3-4-5 triangle.  They’ll not only recognize other 3-4-5 triangles faster, but eventually 6-8-10 and 4.5-6-7.5 as well.

When I started in computational biology, I was all about complex computer simulations.  By my senior year of college, I’d learned to appreciate mathematical models.  They left out details, but that was what let you see the general patterns.  If I’m right, I’m about to undergo the same process for basic sensory data.  And I know I can do it.

*Exception: reflexes

Possibly fictitious diseases of the endolymphatic sac

Last week I asked: what could go wrong with the vestibular system.  I was hoping to have a more satisfying answer by now, but I haven’t.  So here are some random wanderings.

The scientifically suspect book my scientifically suspect first sensory integration therapist gave me suggests “underinflation of the endolymphatic sac”, without anything so droll as a definition of the endolymphatic sac.  No problem, I have the internet.  Wikipedia’s entry on the subject is… weak.  I originally misread it as saying the endolymphatic sac was connected to only the saccule (one of the two linear-motion detecting mini-organs).   The Pictorial Guide To Cochlear Fluids sets me straight on this: all of the various vestibular systems are interconnected, and they all connect to the endolymphatic sac.

An underinflated endolymphatic sac would imply an insufficient amount of endolymph- either because you’re not producing enough, you’re somehow leaking it, or you need extra for some reason  (extraordinarly large semicircular canals?  I don’t know).  The various vestibular structures can’t have their fluid levels too tightly coupled, because that would ruin their specificity, but they are connected via endolymph ducts.  It would (famous last words) make sense if they all had some ideal fluid level, and overflow went into the endolymphatic sac.*

Dr. Internet has pointed me to many descriptions of how an excess of endolymph is bad for you.  And it’s not hard to imagine how a deficit could hurt you.  But is it possible to have enough for the vestibular system, and yet the endolymphatic sac is undesirably empty?  Unscientific Book suggests that because the sac is directly touch with brain fluid, information is reported to the brain through it, and that an underinflated sac can’t do this.  I can’t prove that’s not true, but it strikes me as unlikely to be a large effect.  If I had more faith in the book I’d poke around more, but for now I’m going to leave it.

*Last week I implied the ortoliths (linear motion detecting systems) contained only gel, not endolymph.  In my defense, I thought it was true.  I was wrong.