Wednesday, January 30, 2008

Physiology behind the Magic Light Helmet for Alzheimer's

There have been a number of different blogs about the Magic Light Helmet for Alzheimer’s. Respectful Insolence and again! (Must be Orac's self-identification with blinking LED based intelligence boosting schemes ;), Science Based Medicine.

This is a more elaborate explanation of the quick comment I made at Science Based Medicine. My explanation is based on my understanding of the cause of Alzheimer's (which is not one of the common explanations (all of which I am quite skeptical of (in other words I think they are wrong)). I see Alzheimer's as being strictly due to low NO which then causes low ATP (via sGC), and invokes ischemic preconditioning (IP). The low perfusion is secondary to the low NO. Ischemic preconditioning can be mediated by oxidative stress in multiple ways. Normally IP is thought of as a good thing because it does reduce damage during ischemia. IP is invoked by brief periods of ischemia and after it is invoked organisms can survive periods of ischemia that would have been fatal. IP is a low ATP state, and it is also a low ATP consuming state. If organisms could be in the low ATP consuming status of the IP state long term, they would have evolved to do so. Spending less ATP on basal metabolism means more ATP for reproduction. Organisms haven't evolved to do so, which means there must be something incompatible with life and/or reproduction from being in the IP state long term. Neurons don't divide, so there must be something incompatible with long term survival for neurons to be on the IP state long term. I discuss more details of why accumulation of amyloid is a major outcome in Alzheimer’s below.

I am assuming that this Magic Light Helmet actually has physiological effects beyond the placebo effect. It might not. The placebo effect is mediated through NO, and all disorders associated with low NO are very susceptible to being improved by the placebo effect. That would include Alzheimer's. I will use the analogy of an internal combustion engine a lot. Not because it is an especially good analogy, but because it is easy to understand how bypassing the very simple controls of an engine to get “better performance” could easily shorten its lifetime by orders of magnitude. What does bypassing the extremely complex control system of the brain do? Most of which is completely unknown (but is likely already disrupted for that brain to have Alzheimer’s in the first place)? It is extremely unlikely (virtually inconceivable) that bypassing the normal control system of the brain would only have benign effects.

We know that Alzheimer's is not caused by a "deficiency" of light at 1072 nM. This Magic Light Helmet is not replacing some physiological need that is missing in Alzheimer's. The fundamental problem of Alzheimer's is low ATP supply (discussed below). Since the brain does not have photosynthetic centers to capture the energy of light at 1072 nM and convert that energy into ATP, any change in ATP supply can occur only via changing the regulation of existing ATP production systems via mechanisms susceptible to light at 1072 nM. Since physiology didn't evolve to be physiologically regulated by an external source of 1072 nM light these lights must be doing something else, something that is non-physiologic. Even if there are physiological receptors of 1072 nM light, they would be for sensory purposes (maybe setting day/night biorhythms?) and the levels used in the magic light helmet are many orders of magnitude greater, and narrow band rather than broad band natural sources (sunlight or fire).

If this NIR light does do "something", it pretty much has to be via modulation of kinetics of either free radical reactions, decay of "excited transition states" in activated molecules, or in photodissociation of things from other things (such as NO from cytochrome c oxidase). There could easily be hundreds of things that are being affected, none of which are well understood. Or it might just be placebo.

There are at least 4 main regions which might be important and which I will discuss, by far the most important is ATP production in mitochondria and I spend most of my time on that. Synthesis of compounds by the cytochrome P450 enzymes, catecholamine metabolism and acetylcholinesterase activity might be important too. There might be others, but these are 4 important ones, changes in any of them could conceivably be consistent with the "data" that a handful of non-blinded observers noticed improvement in 6 weeks of use. Although I will be discussing these somewhat separately, they are not separate. Physiology doesn't divide what it does into neat little orthogonal and independent boxes. In physiology everything is coupled to everything else. There is no evolutionary driving force to keep things separate or modular. There are only evolutionary driving forces for survival, and for economy (so more resources are available for reproduction).

ATP production

In neurons virtually all ATP is made by mitochondria through oxidation of substrates, lactate, ketone bodies, small acids such as acetate, aspartate. Mitochondria are small organelles that have a few thousand proteins, 13 of which are coded for by mitochondrial DNA, all the others are coded by nuclear DNA. Mitochondria have 2 lipid membranes, inside the inner one is where the DNA is and the protein manufacture stuff (the mitochondria matrix) and the enzymes of the citrate cycle and most everything that mitochondria do. Mitochondria work by generating an electrical potential and a pH gradient across that inner membrane. The different respiration complexes take electrons and protons from chemical compounds and extract energy from the chemical reactions as those electrons and protons are moved across the membrane and store that energy in the electrical and pH gradient. In the terminal enzyme, cytochrome c oxidase those electrons are gathered and 4 of them are simultaneously put onto O2 along with 4 protons making two molecules of water. Doing so many things all at once is a “tricky” and complicated thing for an enzyme to do. That energy gradient is then used to make ATP.

The 13 proteins are all parts of the respiration chain, usually the part containing the active site. All animals (except for a few invertebrates) have these same 13 proteins coded in their mitochondria. Plants have a few extra. These proteins are all large and quite hydrophobic. Why these (and only these) proteins are coded in mitochondria is not understood. I think it has to do with regulation of mitochondria, some of which has to be local to each mitochondrion and sometimes that regulation means turning off part of the respiration chain and then turning it back on. One mechanism might be destroying the protein and then making the protein again. In most cells mitochondria are close to the nucleus so that proteins can be made from DNA in the nucleus and then transported to mitochondria (in principle, whether this happens or not is unknown). In neurons that can't happen because the distance between the cell body (where the nuclear DNA is and the protein synthesis capacity) and mitochondria can be inches or even a meter in motor neurons. There simply isn't time for a signal to propagate from mitochondria to the cell body, trigger protein synthesis and then transport proteins out to mitochondria in need of them. If protein synthesis is needed for control of mitochondria, that synthesis must occur locally using locally available DNA.

If neurons are deprived of substrates, O2 and organic compounds, disruption of neuronal function (unconsciousness) occurs in seconds.

NIR does cause the photodissociation of “poisons” from cytochrome c oxidase. This is discussed in terms of NO in a recent article in Nature. They get the basic chemistry right, however they get the implications of how NO fits in physiology wrong. The problem is not too much NO, the problem is too little. One of the papers cited by the Nature article epitomizes much of what is misunderstood in the NO field (but the author is quite senior, so I won't link to the article). NO and superoxide can form peroxynitrite, however superoxide from mitochondria is confined to the mitochondria matrix, there are 2 lipid membranes the superoxide has to go through before it can get to the cytosol. Both of those lipid membranes have ~10x higher NO levels than the cytoplasm (NO partitions into lipid ~10x over aqueous). NO reacts with superoxide at near diffusion limited kinetics. It is completely implausible to me that superoxide could get out of the mitochondria matrix. Particularly when the mitochondrial potential (~140 mV) is going to keep the negatively charged superoxide inside. When peroxynitrite forms in the mitochondria inner matrix, it is consumed by cytochrome c oxidase, or decomposes generating NO2 (which is also a signaling molecule).

NO and superoxide only form peroxynitrite at near stoichiometric ratios. When NO or superoxide is in excess that doesn't happen. A much better conceptualization of what happens when NO and superoxide are generated in cells is by Wink et al. They make a very important (and correct) statement

"We demonstrate that the primary consequence of [superoxide] O2- generation concomitant with NO production is not the toxicity associated with the formation of higher nitrogen oxides, but rather the resultant phenotypic cellular changes that occur because of limiting the bioavailability of NO and H2O2."

This is precisely correct. The problem of oxidative stress is not the presence of nitrating NOx species, the problem is not enough NO. The problem of nitrated proteins in neurodegenerative diseases is from not enough NO, not from too much.

Of course what is a "poison" in one circumstance may be an absolutely necessary regulatory pathway in another. Normally cytochrome c oxidase is tonally inhibited by NO, which blocks O2 from binding and is the major regulatory pathway by which mitochondria regulate their O2 consumption. If you look at figure 1, you can see that in the absence of NO, mitochondria consume O2 down to ~5 microMolar. The solubility of O2 in plasma (assume same as in cytoplasm) at 1 atmosphere is 23 cm3/L (at 38 C) or about 1.1 mM/L. 5 micromolar O2 is then a partial pressure of about 0.0045 atm or 3.42 Torr.

The only reason mitochondria can regulate their O2 consumption is because NO "poisons" cytochrome c oxidase and inhibits O2 consumption. Remove that inhibition and mitochondria consume O2 to very low partial pressure, well below what is the "normal" basal O2 level at the location of the mitochondria. At "rest", the O2 flux to a mitochondrion in the heart is 1. The O2 consumption by that mitochondrion can increase by 10x. The flux of O2 from the blood vessel to the mitochondrion is purely passive down a concentration gradient. For the flux to go up 10x, either the gradient has to go up 10x, or the distance has to go down by 10x because the concentration at the blood vessel stays the same. For the gradient to go up by 10x, the concentration at the mitochondrion has to go down, and go down a lot, by a factor of 10x. It has to go down while the mitochondrion is increasing its O2 consumption by 10x. The specific O2 consumption by that mitochondrion, moles O2/mg protein/Torr O2 has to go up by a factor of ~100. This is only achieved by removing the "poisoning" of cytochrome c oxidase by NO during the “at rest” condition.

With NO blocking cytochrome c oxidase, the electron respiration chain becomes fully reduced and O2 can pick up electrons from complex III, forming superoxide. This superoxide is vectorally produced in the inner matrix, where MnSOD dismutates it at near diffusion controlled kinetics. NO also reacts with superoxide at near diffusion controlled kinetics and it is the destruction of NO by superoxide generated by too great a reduction of the respiration chain that lowers the NO level and disinhibits cytochrome c oxidase so that O2 can consume electrons (and be reduced to water).

If this light does dissociate NO from cytochrome c oxidase it would cause the respiration chain to become more oxidized and would reduce the quantity of superoxide formed. This would increase NO levels and upregulate ATP levels via sGC.

An analogy by looking at the regulation of an engine would be to consider that the “throttle” “poisons” the engine so that the engine does not run at 100% full throttle 100% of the time. What the Magic Light Helmet might be doing is bypassing the normal mitochondria regulation, an effect analogous to jury rigging a separate wire to the carburetor to bypass the normal controls. If you had only one mitochondrion to control, perhaps external control might work. Each cell has many thousands, if not millions of mitochondria. If they are not working together and sharing the load, some are working harder than others, consuming O2 that others could consume, generating non-physiological O2 gradients perhaps generating non-physiological gradients in ATP, or in other substrates consumed by mitochondria. A better analogy would be to think of it as a power grid, with millions of engines connected together. Removing the throttles on some of them might improve the total power production of the grid, until those engines running at 100% full power 100% of the time start to fail, or until the idle engines (those that are not doing anything) get shut down. Of course if the engines running at 100% full throttle were uniformly distributed it might work out. If they were only the ones within 50 miles of the ocean (analogous to limited light transmission in the brain), the non-uniform distribution might set up instabilities in power flow leading to roving brownouts or blackouts or even grid failures. Maybe not too bad for a carbureted engine, but what if they are fuel injected? What if your jury-rigged control bypass only delivered extra fuel to some cylinders and not others? Do you start getting bad vibration? Do you start getting bad fuel-air mixtures? Too rich in some and too lean in others? If you were as ignorant about engines as we are about physiology it would be easy to do serious damage due to ignorance. Of course the more ignorant you are, the more difficult it is to tell just how ignorant you are.

A figure that shows the mitochondria respiration chain is here. There is control mediated all along the respiration chain. Exactly how the respiration chain is regulated under what conditions in which tissue compartments under what normal and abnormal circumstances is a good question that all of the senior researchers working in the field would like to understand. This particular paper is about nitration of proteins in mitochondria under conditions of hypoxia and denitration when that hypoxia is removed. This nitration and denitration occurs rapidly and reproducibly. It is likely some sort of control mechanism for controlling each of the proteins that get nitrated and denitrated. The nitration is likely mediated via NO reacting with superoxide and forming peroxynitrite, or forming NO2, both of which can nitrate proteins. Precisely how that happens is unknown. Virtually all the nitration is on tyrosines which are in specific parts of the enzymes that are so regulated. Tyrosine gets nitrated more than other things because it is aromatic (has a benzene ring in it) and the pendant groups on that benzene ring direct agents that do nitration to specific carbons on that benzene ring. Tyrosine also tends to form tyrosyl radicals, that has an unpaired electron distributed on the aromatic ring (which stabilizes it somewhat). Tyrosyl radicals are much more reactive than tyrosine. It may be (actually is quite likely) that formation of tyrosyl radicals and then quenching of that radical by something else is likely part of a regulatory system.

Low ATP caused by low NO hypothesis of Alzheimer's

One of the most consistent symptoms of Alzheimer's is a reduction in brain metabolism. Amyloid (and all the other protein aggregates of the neurodegenerative disorders) is normally cleared by ATP powered proteases in the proteasome, or during autophagy (which requires an ATP powered pH gradient). My hypothesis of Alzheimer's is that the buildup of amyloid is a natural compensatory mechanism to cope with there being not enough ATP. There are simply more important things to use ATP for (when there is not enough) than to use it to get rid of amyloid (or to make new mitochondria). If the ATP setpoint is too low, my hypothesis is that ATP conservation pathways are invoked (which include not getting rid of amyloid) as in ischemic preconditioning. I discuss this in my blog on fevers and autism . Getting rid of garbage is something that can always be put off "a little bit longer" if there is something better to do. I think that Alzheimer's happens when that heuristic is taken a bit too far.

Just about every other physiological function inside neurons is more important than getting rid of amyloid. Nerve conduction is a lot more important. If your nerves are not conducting properly, a bear could catch and eat you. Keeping nerves alive is more important too. Amyloid buildup causes essentially no problems for quite a while, months, even years. My mother and both her parents died with advanced Alzheimer's, so I am not being flip, I know what it can do. But that the brain can function so well and for so long with such a serious decline in metabolism is quite remarkable to me. One of the things that has helped me in my research is that I inherited my mother's low nitric oxide physiology, so I experienced quite dramatic changes when I corrected it (I appreciate that my experiences are anecdotes).

The Magic Light Helmet might work in the short term for people who already have low ATP. I would be very concerned how it modifies the normal regulatory pathway(s) by which the number of mitochondria per neuron is regulated. My research indicates that regulation involves formation of long lived NOx species which accumulate in mitochondria and which release NO during autophagy and then trigger appropriate mitochondria biogenesis. I think those long lived NOx species include nitrated proteins, which derive only from the combination of NO and superoxide. The formation of superoxide by mitochondria under metabolic “stress” is then the signal by which the cell “measures” the metabolic stress of the mitochondria it has, and then infers how many it needs to make. In a neuron that number can be different by 3 or more orders of magnitude depending on the length of the axons.

In the rat CNS, mitochondria lifetime is on the order of a month. How long it is in humans is unknown. It might be somewhat longer, but likely not more than an order of magnitude. The turnover time for mitochondria might be inferable from the progression rates of some neurodegenerative diseases. If we assume that a neurodegenerative disease affects mitochondria biogenesis, and is not acutely toxic to mitochondria, then the minimum course of that disease might reflect mitochondria turnover. ALS tends to have a course longer than a certain minimum length. A more stringent criteria might be the woman who received an acute toxic dose of dimethyl mercury but had no symptoms for 5 months and was dead at 7 months post exposure. Her symptoms were consistent with neurodegeneration. If the mercury had acutely poisoned mitochondria her death would have been much sooner. If it only poisoned mitochondria biogenesis, then her survival to 5 and 7 months might reflect how long her inventory of mitochondria at exposure could sustain neuronal activity. In any case, a treatment for Alzheimer’s has to extend beyond several mitochondria lifetimes to be considered an effective treatment.

The Magic Light Helmet might slow the progression of Alzheimer’s for a while by operating existing mitochondria in a way that generates less superoxide. But that would likely compromise the regulation of mitochondria by superoxide. Mitochondria regulation by superoxide includes regulation of O2 consumption under normoxia and hypoxia, and very likely includes regulation of mitochondria number. The progression of Alzheimer’s could greatly accelerate at some future time when mitochondria wear out faster than they are replaced.

A (poor) analogy would be to suppose that one had a car with 100,000 miles on it, and it is starting to run not as well as it used to. A salesman says he can "fix" the car by installing nitrous oxide injection and by adding nitromethane to the fuel. Sure enough when you do that, the car "runs" better than it ever has. But if the car had been nursed along gently, it could have easily gone 150,000 miles, but with nitromethane it only goes 100,500. That is why long term trials with the proper endpoints (i.e. death) are most appropriate for terminal conditions like Alzheimer's. Four weeks is too short. I imagine that the right dose of cocaine or amphetamine might produce "improvement" in Alzheimer's for 4 weeks but would accelerate decline and hasten death.

On the Science Based Medicine blog, neurocritic made a comment about the result reported on the very rapid improvement of Alzheimer's on enterocept injection (into the spine). I see this as consistent with the low NO hypothesis of Alzheimer's with NO being what regulates the acute functional connectivity in the brain. I replied to neurocritic in this comment.

Dysregulation of cytochrome P450 metabolism

Cytochrome c oxidase is not the only heme enzyme inhibited by NO. All the cytochrome P450 enzyme are too. Many of them are active in the brain. How is changing the operating point of all of those enzymes going to change things? Very complexly, and very likely not in only benign ways.

There are about 60 cytochrome P450 enzymes, which synthesize such things as steroids, cholesterol via a reaction cycle that generates superoxide. Normally the P450 enzymes generate significant superoxide which is vectorally produced to the inside of the microsomes the enzyme is active in. Microsomes normally have superoxide dismutase and catalase inside them. Testosterone synthesis is known to be inhibited by NO.

Dysregulation of catecholamine metabolism

I bring up catecholamine metabolism because many aspects of it are also quite involved with free radical chemistry and so would likely be susceptible to disruption by NIR. Parkinson’s disease has some similarities to Alzheimer’s (it is characterized by buildup of protein inclusions too, but of a different composition). Dopamine pathways are highly involved in feelings of wellbeing. Parkinson’s disease-like symptoms can be reliably induced by killing certain nerves associated with the dopamine pathways. The usual experimental mechanism is with a compound that damages mitochondria (MPTP) in those neurons, causing their death via oxidative stress and ATP depletion.

Dysregulation of acetylcholinesterase metabolism

NIR does affect the activity of acetylcholinesterase enzymes in red blood cells (abstract only). Not surprisingly, the effect is complex with low levels reducing it and higher levels increasing it. Which is the Magic Light Helmet doing? Neither or some of both?

Acetylcholinesterase inhibitors are used to treat Alzheimer’s. There is a treatment effect, and the treatment effect is dose dependant, but the treatment effect is small. There does seem to be an association of increased executive function with increased inhibition of acetylcholinesterase.

If the Magic Light Helmet is inhibiting acetylcholinesterase, it is doing so in a spatially non-uniform manner. It might even be increasing activity in some parts of the brain and decreasing it in others. Because some nerve cells are larger than the region where the light has a single flux (and hence a single dose-response), some cells might experience both increased and decreased acetylcholinesterase activity but in different regions.

Expression of acetylcholinesterase is ultimately regulated in the cell body because that is where the DNA is that codes for it. If one axon of a nerve has it inhibited and one has it enhanced, how does the cell decide how much to make? What does that do to the actual operation and long term regulation of those two different axons? It would be extremely unlikely that the nerve cells most in need of having their acetylcholinesterase enzyme activity modified by the Magic Light Helmet were located in the regions of the brain where that could actually happen.

Safety

With no understanding of the mechanism there is no basis for saying if it is safe or not.

They claim it works, but offer no physiological explanation. They have a 6 week open label trial, but with no controls and no blinding of patients or investigators. All of physiology is non-linear. You can’t extrapolate from known conditions to unknown conditions when the underlying phenomena are non-linear (and unknown). The lifetime of mitochondria is likely longer than 6 weeks. Adverse effects might not show up during that time.

What concerns me most about this device is that the investigators don't seem to be asking the right questions about what it is actually doing, if it is doing anything. Rushing to human testing is (in my opinion) quite premature with something with little to no understanding behind it. This could easily be a situation where anecdotes were used instead of actual clinical trials and people become seriously injured.

Summary:

In summary, I have presented several plausible (but speculative) physiological mechanisms by which the Magic Light Helmet could have actual physiological effects. In none of these mechanisms does the Magic Light Helmet produce an effect that is representative of actual physiological needs that the brain has. The Magic Light Helmet doesn't regulate anything by any mechanism that is known or understood.

If it does any of those things it is likely to do so indiscriminately. There is nothing "magic" about 1072 nM, these investigators looked at a handful of wavelengths (where LEDs are available for cheap). Something like that might have a short term positive effect but is very likely to be very bad in the longer term. Physiology is too complex and too coupled to simply whack away at it indiscriminately and make it "better".

Monday, January 14, 2008

The Myth of Homeostasis: Implications for Neurodegeneration

Myths of physiology

This is my first attempt at a blog for the Skeptics Circle. I won't go after the obviously wrong alternate medicine cranks, quackery, woo and the delusional thinking that other skeptics handle so easily (which I see as the equivalent of fishing with dynamite in a bucket). When you are successful, the results are not entirely satisfactory because you don't end up with anything useful (just fish guts splattered everywhere). I will go after myths that are more damaging because some scientists actually believe them and waste time pursuing dead ends based on wrong ideas and clutter the literature with sloppy thinking which delays development of real treatments for real patients. I plan to write on a number of myths over time. Some may object to their favorite myths being debunked. Feel free to object and (try to) defend your myths, but be prepared to back it up with facts and logic, just as I am prepared to back up everything I say with facts and logic.

One of the most pernicious and pervasive myths, one which seriously interferes with progress in understanding and treatment of dysfunctional physiological states is the myth of homeostasis. The term was coined by Walter Bradford Cannon in 1932 from the Greek homoios (same, like, resembling) and stasis (to stand, posture). He also coined the term "fight or flight", which is actually a useful concept and is ironically completely inconsistent with the concept of homeostasis. How can physiology change to engage in fight or flight by remaining static? It can't. There has been some attempt to fix the flawed homeostasis concept with things such as "dynamic homeostasis" which I see as a simple oxymoron, and akin to the epicycles of the geocentric solar system model. Another approach is to limit the physiological parameters that homeostasis applies to, a "homeostasis of the gaps" idea. Ad hoc complications to try and rescue a failed hypothesis.

The idea of homeostasis (not a hypothesis because it is demonstrably wrong) is that in order to maintain physiologically appropriate conditions internally and maintain itself as a living organism, organisms maintain a state of homeostasis. That is many physiological parameters are maintained in a static state, and deviations from that static state are dysfunctional and are actively countered via compensatory pathways. A disease state occurs when homeostatic mechanisms are overwhelmed and homeostasis cannot be maintained. Ok, not a bad default assumption in 1932 when you are not able to really measure anything inside an organism, let alone inside an organ, or a cell or an organelle. Before there was any real idea of how cells functioned, a simplistic assumption there was something (not quite) magical called "homeostasis" that kept everything constant helped researchers to ignore the obvious complexity they couldn't begin to measure and focus on the very simple things they could measure, temperature, pH, water consumption, urine output (which usually were controlled by physiology better than the accuracy of the measurement techniques of the time). It is a step up from vitalism (which posits an imaginary life force that does the same thing), but only a small step. When all you can measure are simple things, with low precision and only on a very long time scale, everything does look constant. If you could only measure heart rate averaged over 10,000 beats (about 3 hours), you could detect a change if someone ran a marathon, you couldn't if they only ran a mile. 5 minutes with a 3x heart rate would be lost in 3 hours of 1x. Now we know that heart rate is actually chaotic, with inter-interval spacing varying in a power-law manner. Chaotic behavior is characteristic of systems of multiple coupled non-linear parameters. Many physiological systems are known to be chaotic; it is likely that virtually all of them are (all known physiological systems comprise systems of multiple coupled non-linear parameters). What about a chaotic system is or can be "static"?

On some level for an organism to maintain itself in a living state all parameters have to be kept within a certain range. But that range is not necessarily the value at rest or a constant. Organisms do remain alive with those parameters at values different than at rest. The value that physiology calls for at any moment may be different than the value at rest if a different value achieves some physiological goal different than remaining alive at rest. Are there any parameters that are kept static? Not so far as I have been able to find any evidence for in the literature. Virtually all parameters are regulated via feedback control, but that precludes stasis. Many parameters are regulated via feed-forward control. That is, parameters are adjusted in anticipation of a future need, not in response to a past need. This is a subtle point, but feed-forward is absolutely essential. When the heart pumps blood, the blood going through the heart is needed at a future time, when that blood reaches the target organ. The heart has to pump blood in anticipation of the need of the organ for that blood. When the liver puts glucose into blood, it is for a future need, when the glucose reaches the organ that uses it. Feed forward control is necessarily less precise than feedback control, but a sacrifice of efficiency for speed is a good evolutionary trade-off.

As skeptics, our default position has to be that we don't know something, rather than to assume that something has a particular property or value. Do we know that any parameters of physiology are static? No, we don't unless we actually measure them sufficiently precisely over the time and length scales in the physiological state of concern. We do know that those parameters are highly regulated, and regulated by control system(s) that are extremely complicated, extremely robust, which have evolved over billions of years, the details of which remain mostly unknown. We do know that these control systems evolved only to preserve the life and reproductive capacity of organisms that have them. There is no evolutionary pressure to maintain any physiological parameter constant; there is only evolutionary pressure to sustain the life and reproduction of the organism. If preserving the life or reproductive capacity of an organism would be better achieved by changing a control setpoint, presumably physiology would have evolved to do so (yes, I know, duh).

Some might argue that even though nothing is actually static, that homeostasis is a useful concept because it makes "intuitive sense". That is a wrong idea. The most dangerous and pernicious bad ideas are those that are wrong but which do make "intuitive sense". Most people's intuition is poorly suited to sorting out bad ideas. All woo and quackery is accepted by some because to them it makes "intuitive sense". As skeptics we must have zero tolerance for demonstrably false ideas, especially when they make intuitive sense. The problem is with our intuition. If our intuition is wrong, we must compensate for that wrong intuition and reject compelling ideas whenever they are wrong.

Researchers may not be comfortable without a backdrop of something "constant". A century ago, physics had the concept of absolute time and space. Relativity showed that concept to be wrong and reality could be better understood without it. Some physicists had great difficulty abandoning absolute time and space and adopting relativity.

Good control of any physiological parameter requires a setpoint and regulatory mechanisms to restore deviations from that setpoint. Physiological control is fabulously good. Far better than most actually appreciate. Bad control can either be good regulation around a bad setpoint, or bad regulation around a good setpoint (or some of both). Homeostasis posits that the "setpoint" is constant and fixed, and that any dysfunction is due to bad regulation about that fixed "homeostatic setpoint". A skeptical position is that any setpoint might not be fixed, it may well be a control parameter, and a deviation from the at rest setpoint might actually be a compensatory response to correct a dysfunctional state. Since we don't know every single physiological role of each and every single physiological parameter, we don't know if physiology is best served by maintaining that parameter constant or not. If that particular parameter is used as a signal (and it must be if anything is controlled by it including the parameter itself), then a deviation of that parameter may be the absolutely necessary trigger to activate a downstream pathway necessary for compensation. With a homeostatic perspective, any deviations of physiology from the "at rest" state are considered pathological or dysfunctional. But this ignores what is cause and what is effect. If physiological parameters are coupled (as we know they are), different parameters cannot be regulated independently. If multiple parameter(s) change, what is or should be the treatment goal? Which parameter(s) should be kept fixed so as to maintain homeostasis? The vast majority of physiological parameters cannot be measured and are regulated in each cell somewhat independently of all other cells. How do we choose which one(s) are important and are the "homeostatic" ones? Only through actual measurement.

I will focus on ATP concentration in cells which relates to my NO research. Don't worry, this is the only paragraph that I discuss NO in. NO regulates the ATP setpoint via sGC, and is the diffusible signal that communicates ischemic preconditioning (discussed below). Superoxide generation triggers ischemic preconditioning, consumption of NO by that superoxide turns cells generating superoxide into NO sinks, and the fall in NO propagates the ischemic preconditioning signal (as spreading long term depression in the brain) between cells. I mention it here so I can avoid mentioning NO later and confine NO physiology to this one paragraph. Some of the specifics relate to my previous blog on how fevers affect energy status in the brain and how low ATP shows up as white matter hyperintensities. I focus a lot on ATP in nerves because it is somewhat simpler there and relates to my work in autism, Alzheimer's and Amyotrophic Lateral Sclerosis (ALS). In many ways these neurodegenerative disorders are all "the same", in that they are characterized by low NO and low ATP (which I think is the ultimate cause). Just the details are different. But of course there are always a lot of details, and it is easy to get lost in those details and get caught up in the differences (which are mostly unimportant regarding treatment possibilities) and ignore the similarities (where treatment possibilities lie). The forest for the trees problem.

I will try to get into glucose regulation in a later blog. Regulation of glucose level is important under different conditions, and standard practices may be (actually are) detrimental, for example what glucose level is best to artificially impose in the ICU. Those who seek to engage in Evidence Based Medicine may want to consider the actual evidence for the optimum blood sugar under conditions such as sepsis. For example, why would we expect it to be the same as in non-sepsis when the immune system mostly functions by glycolysis and in experimental animals LPS always causes an increase in blood glucose? Maybe the hyperglycemia of sepsis is functional. I appreciate that a default of an "at rest" glucose level might be reasonable in the absence of any studies showing higher levels to be superior. But we don't know if higher levels would be superior unless we actually test them.

In the case of glucose, what glucose parameter is important? Carbohydrate consumption? Carbohydrate assimilation? Glucose level in bulk blood? Glucose level in CSF? Glucose level in extravascular fluid adjacent to the cells actually taking it up? Glucose levels inside cells? A recent (1999) "review points out that there is no compartment of glucose in the body at which all glucose is at the same concentration, and that models of glucose metabolism, including effects of insulin on glucose metabolism based on assumptions of concentration homogeneity, cannot be entirely correct." I would be more blunt; such models are wrong. (But I am not writing this for a journal requiring approval by an editor or peer reviewers who may believe in homeostasis).

In obesity, what is "homeostasis" trying to keep constant? Is it total organism energy status or energy status of the brain (the most important part of energy status)? If the part of the brain that controls appetite doesn't have enough ATP, presumably it would call for more substrate irrespective of the amount of depot fat the organism has (which the brain cannot use unless the organism is in ketosis). If there are insufficient mitochondria in the liver to support gluconeogenesis and the Cori cycle to recycle lactate into glucose, the only way to maintain glucose supply to the brain may be via consumption of carbohydrates or cachexia which may result in anorexia or obesity. Anorexia and obesity may be more similar than people think because they are both characterized by reduced energy status in the brain. You can actually have both obesity and cachexia simultaneously. During sepsis, muscle is catabolized to make glucose and body fat levels actually increase (I think to dispose of lactate but that is for a future blog).

Body temperature is pretty well controlled in humans and is often used as an example of one of the "homeostatic" parameters. Temperatures too far above or below "normal" can be fatal, but acute illness often produces a fever, and in experimental animals, blocking a temperature increase interferes with the recovery from infection. Under the homeostasis idea, preventing fever would augment homeostatic temperature control and would restore improved function more quickly. It doesn't. Raising body temperature as a response to infection is found in all vertebrates including endothermic animals such as fish. It is also found in insects.

Elevated temperatures affect how immune cells react. Presumably the body is using fever as a signaling mechanism. Driving mitochondrial respiration very high increases metabolic heat and would produce local heating, precisely the proper signal to attract macrophages and to activate them. Blocking fever may make one feel better, but it is likely to prolong the illness.

ATP Regulation

As a charged molecule, cell membranes are impermeable to ATP. Each cell necessarily regulates ATP via mechanisms which are independent of (but obviously related to) surrounding cells. Even now, there is no technique to measure ATP non-destructively in single cells on the time scale it is controlled at (sub second). Typically ATP is measured by freeze clamping a bit of tissue with liquid N2 cooled copper tongs (to stop production and consumption) and then assay it after dehydration or other denaturing of making and using enzymes. Typical minimum sample sizes are a few hundred to a few thousand cells. Measuring the averages of a few hundred cells doesn't tell very much about what ATP is doing in individual cells (or in sub-cellular compartments). ATP can be measured non-destructively by fMRI, but only in large volumes (very large compared to single cells and the length scale where ATP is actually controlled at).

ATP is exquisitely well regulated. Is there any evidence that it is kept "static"? No, there isn't. There are 3 potential ATP parameters, ATP production rate, ATP consumption rate and ATP concentration. Over the course of a day, oxidation of 2000 calories of glucose (0.56 kg) and efficient conversion to ATP produces some 55 kg of ATP. Pretty obviously the ADP and P are recycled many times and over a day's time the production and consumption is essentially exactly balanced. Since ATP isn't stored, production and consumption is balanced in essentially real time.

How many different pathways consume ATP? Conservatively it must be at least 10^5 in each cell, probably 10^6 or more. There is the synthesis of ~10^4 proteins in each cell. If we assume 10 additional pathways including post translational processing, phosphorylation, de-phosphorylation, ubiquitination, proteolysis, folding, transport by motors, and more, we can easily get to 10^5. The precise number is not necessary, simply that there are very many. Each of those pathways consumes ATP and is (to some extent) necessarily regulated separately from all others. How are all of those pathways controlled? The answer is very well.

We know that ATP consumption (and hence production) can vary by an order of magnitude (in heart muscle). We know that sufficiently severe ATP depletion causes necrotic death. We know that transient periods of acute ischemia induce what is called ischemic preconditioning (IP). Ischemic precondition is a lower ATP state, but more importantly is a lower ATP consumption state. This IP state induces a temporary state where the organism, organ, or cell can better tolerate ATP depletion and survive undamaged. The details of IP are mostly unknown, but because the effect is extremely common and robust, it no doubt represents an early evolved and fundamental stress response of cells.

Each of those 10^5 pathways can only be controlled locally. That is, the consumption of ATP is local to the molecules directly interacting with that ATP, the control of that consumption must be local too. There are many mechanisms, including phosphorylation, conformation changes, binding of peptides to block ATP consumption, many more and combinations of all of these. So what signal is used to communicate the cell's ATP status to each of those pathways and so regulate each of those pathways independently? It has to be a small molecule to be able to get every where. There are not 10^5 different small molecules, virtually all of these pathways must use the same signaling molecule for regulation appropriate to ATP status. That molecule must also couple to ATP. My conclusion is that many (if not most) pathways must use ATP itself as the signaling molecule to adjust activity to cellular ATP status. This is an extremely important conclusion. If so, it would mean that ATP cannot be regulated to be constant as the homeostasis idea requires. In addition to being the energy source, ATP must also be the signal that regulates consumption of that source. ATP would be an ideal control signal to use because it exactly tracks ATP status. No intermediate signal inducing delay or error is necessary. Regulation would be completely stable. Presumably ATP was the first signal that cells used to measure ATP status and regulate pathways accordingly. As cells evolved that signal has become more complicated, but it won't be replaced so long as it works. We know there are many proteins that bind to other proteins depending on ATP status (the heat shock proteins for example).

We know that some pathways can consume ATP until the cell dies, for example muscle contraction. A useful survival feature. The heart continues to pump until the heart kills itself from ATP depletion. One can continue to run from a bear until one drops dead from exhaustion. This demonstrates that muscle contraction has a low ATP threshold for being turned off. Other pathways such as axonal transport have higher thresholds for being turned off (see discussion below), that is they are turned off earlier as ATP levels fall.

Ischemia occurs when there is disruption of the supply of substrates (O2 and glucose) to cells. With insufficient substrate, cells cannot make ATP via their normal pathways. As ATP falls (due to insufficient substrate), cells must either make more ATP or use less (or both). Disruptions to ATP supply are leading causes of cell death. In one sense, the ability to generate sufficient ATP to maintain cellular integrity is virtually the definition of what it means to be alive. If the ability to make ATP is blocked by insufficient supply of substrates, the only way to preserve ATP concentration is to use less. With no storage, ATP not consumed is as good as new ATP generated. Following invocation of IP, cells do use less ATP, and this persists for the length of the IP period.

If IP reduces ATP consumption, can cells stay in an IP state indefinitely? It is quite clear they cannot. IP can only be a transient state, because if it was possible for cells to maintain long term a state of reduced ATP consumption, they would have evolved to do so, so that more resources could be devoted to reproduction. The IP state would become the new default metabolic state. This has not happened. We can conclude that a long term IP state is incompatible with life and/or reproduction. Since IP happens in cells that don't divide (nerves in the CNS), long term IP is incompatible with life.

Since IP reduces ATP consumption and is protective over some period of time, there must be ATP consuming pathways that are not essential during that time period and can be shut off to preserve ATP, however those pathways are essential for longer time periods (or they would not be conserved).

In my blog on acute psychosis I use the acute stressor of "running from a bear" as the archetypical stress where to be caught is virtually certain death. To escape from a bear virtually any injury short of death is an excellent trade-off. The trade-offs in IP are similar. Any injury short of death is also an excellent trade-off if it prevents death from ATP depletion. All instances of ATP depletion are either transient or the cell dies. Normally, the ATP crisis will only last until the compensatory pathways have increased ATP production to compensate. The transient nature of all ATP crises can indicate which ATP consuming pathways can be shut down. If a pathway is not crucial for life (over the length of the crisis) and is not involved in increasing ATP production (over the length of the ATP crisis) it can increase ATP availability by being shut down. This is the essence of the difference between the "fight or flight" state and the "rest and relaxation" state, the ATP status, as measured by the ATP level. This is discussed in my blog on placebos.

Let us consider nerves. Nerves only obtain ATP from oxidative phosphorylation and then only by oxidizing carbohydrates, ketone bodies and other small organic acids. They cannot obtain ATP from glycolysis or from the oxidation of lipid. Nerves are also large cells. The cell body contains the nucleus where the DNA is stored, and so is where virtually all protein synthesis occurs. The vast majority of the metabolic activity of nerves occurs in the axons, where most of the cytoplasm is. Axons are small diameter extensions of the cell that may extend for inches, or in the case of motor neurons as much as a meter from the cell body. It is down these axons that the action potential is transmitted. What is also transmitted down these axons are all the proteins and organelles synthesized in the cell body that are used at the tippy end of the axon including mitochondria. These are transported out via ATP powered motors with variable velocities, and then transported back when they get "tired". There are two modes, a slow mode (~1 micron/second), and a fast mode several times that. Usually the fast mode is for transport of endocytosed receptors on the tippy end back to the cell body. In either case, the transport time is very long compared to the duration of an ATP crisis (minutes), and this slow transport can't contribute to ATP production, so it is something that could be shut down. ATP consumption by actin is a major energy consuming pathway in neurons.

Is axonal transport shut down when there is insufficient ATP? One of the most reliable symptoms of all of the neurodegenerative diseases is what is called white matter hyperintensities (WMH). On MRI, the white matter (primarily axons so named because myelin is white, cell bodies are gray) exhibits decreased water diffusion. The precise mechanism of WMH remains unknown. I discuss some of the physiology of this in my blog on fevers and autism. Acute ischemia (from blocking a cerebral artery) causes WMH to occur very rapidly I suspect it is a controlled shut-down of axonal transport to conserve ATP. The reduced water diffusion results from reduced convective transport of water entrained by moving cargo in the axons (my hypothesis). WMH is observed to spread (spreading depression), and that spreading occurs slowly, not via action potentials.

But this blog is about homeostasis specializing in ATP homeostasis, not WMH. All of the neurodegenerative diseases characterized by WMH also are characterized by reduced metabolism, by SPECT, by fMRI, by cerebral blood flow, by MRS, by just about every measure that can be made.

What does the idea of homeostasis say about it? Not much. Homeostatic mechanisms have failed. What does a skeptic of homeostasis say about it? If a substantial fraction of ATP consumption is reduced, how many pathways are involved? 1, 10, 100's or more? For a substantial fraction to be disrupted, presumably a substantial number of the 10^5 or more pathways are disrupted. Either the regulation of hundreds of pathways has "gone bad" simultaneously and in characteristic ways for each of the diverse neurodegenerative diseases, or the regulation remains good, but around a bad setpoint. Since multiple cells are involved, and each of these pathways is regulated local to the individual cell, bad regulation of hundreds of pathways via independent mechanisms simultaneously seems extremely unlikely. That leaves good regulation around a bad setpoint. Since spreading depression can propagate and signal a cell to reduce its metabolic activity, presumably that is a normal regulatory process.

Many neurodegenerative diseases are also characterized by accumulation of damaged proteins, amyloid in Alzheimer's, Lewy bodies, tau inclusions in taupathies, alpha-synuclein in Parkinson's, polyglutamine inclusions in Huntington's, SOD1 inclusions in G93A ALS mice, and lipofuscin in all of them. Amyloid also accumulates in non-neuronal tissue in diabetes, obesity, dilative cardiomyopathy, end stage kidney failure and other degenerative diseases characterized by amyloidosis. The association of these protein inclusions is virtually universal but whether they are a cause, an effect, or merely associated with the degenerative disease is unknown. In most cases these inclusions are poly-ubiquitinated, that is they have been tagged for disposal but that disposal hasn't happened. It has been suggested that sometimes aggregation of some of these proteins may be protective. During hibernation, some animals polyphosphorylate tau, a process that is thought to precede the pathological association into aggregates observed in the taupathies. How does homeostasis explain hibernation? what is being kept static?

Cells have 3 main mechanisms for disposal of proteins that are not longer needed or that have been damaged, the proteasome and autophagy. The proteasome is used to dispose of protein molecules one at a time. They get linked to ubiquitin and then the ubiquitin-protein complex gets transported to the proteasome where first ATP powered unfoldases unfold the protein into a single length, then that piece is fed into the proteasome where ATP powered proteases break it up into little bits. The little bits are then recycled separately. The proteasome is used for control purposes also, some proteins are synthesized and then degraded and when their degradation is impeded, then they reach a level that activates something. For example Hypoxia Inducible Factor 1-alpha (HIF-1α) is rapidly synthesized and rapidly degraded under normoxia, but when the O2 tension drops, the degradation stops and HIF-1α accumulates and activates transcription.

The other two protein degradation systems are related to each other. There is Chaperone Mediated Autophagy (CMA) and autophagy. CMA can also process single proteins, autophagy is more of a bulk process where macroscopic quantities of cytoplasm are engulfed for degradation and recycle. The only way that whole organelles such as mitochondria can be recycled is via autophagy. The material to be degraded is engulfed, protease precursors are ported in, the pH is lowered via V1ATPase which pumps in protons. The low pH activates the proteases and they degrade the contents. Conditions are made reducing via porting in cysteine which reduces disulfide bonds which are exported as cystine to the cytoplasm where cysteine is regenerated. This takes ATP to produce the pH gradient. The V1ATPase is inhibited by oxidative stress, which makes sense to save ATP during IP triggered by ROS.

So what if there isn't enough ATP? Proteins that are damaged and ready to be disposed of are not doing anything constructive, they are just taking up space. All they are good for is as substrates, but they need to be taken apart and the bits remade into proteins or oxidized to make ATP. That takes time and ATP to do so. Their recycle can be put off for a little while if ATP is needed for something more important (such as running from a bear). If you have something better to do, you can always put off getting rid of the garbage. Garbage takes up less space if it is aggregated together (a trick I have found to work in living space too ;).

In conclusion (finally!), the idea of homeostasis posits that important physiological parameters are kept constant. We know that isn't the case, but many persist in clinging to that belief and invoke "homeostasis" as a principle of physiology even though there is no data to support it. It is long past time that the idea was abandoned.

Tuesday, January 1, 2008

Resolution of ASD symptoms with fever. Is it real, and if so, what does it mean?

…when you have eliminated the impossible, whatever remains, however improbable, must be the truth. S. Holmes

There is a report that children with ASDs sometimes acutely exhibit improved behaviors during a fever, and that their behavior then returns to baseline after the fever passes.

What can we infer about the mechanisms behind ASDs and ASD symptoms from this?

I only saw this on Friday (12/07), so have mostly missed the blogging about it at Autism Vox and at Prometheus. I have seen the comments there and will attempt to answer some of the questions raised. The more I get into it, the more complicated it becomes and it is very hard to force myself to be simplistic about anything.

I will talk about some of the background in fever therapy, and what I think the acute effects in treating some mental disorders derive from. There are non-acute effects which likely derive from resolution of the infection (when that is what is causing the symptoms such as in neurosyphilis), but the acute effects on mental activities are probably not from that. My conclusion is that the improvement is real, and that it is due to increased basal NO from iNOS expressed during the immune system activation. I think that inducing a fever can only produce transient improvement while the NO level is acutely higher. I think it is very likely that autistic symptoms will be made chronically worse by transient fevers (even if there is acute improvement during each fever). I discuss this in the low NO ratchet section.

I spend a lot of time considering other signaling molecules in addition to NO. The simplest explanation is that the effects are due to NO. Maybe other things are involved too, perhaps as intermediates, but in my opinion, NO is the most likely candidate. I spend a fair amount on what it can't be. For the acute, short term effects, nothing but NO seems credible to me.

I think that my bacteria can produce some improvement too, but unlike immune system stimulation I think the improvement will last as long as there is a biofilm of the bacteria on the skin increasing basal NO levels. How much improvement? If autism is characterized by low basal NO as many symptoms indicate, then there will be some improvement. All disorders characterized by low basal NO will be improved by an increase in basal NO with no threshold. The reason is, because NO is used as a signaling molecule in many feedback regulatory control loops, the basal NO level is necessarily coupled to the basal NO level. When that basal level is perturbed, so are the operating points of those NO mediated control loops. Because NO is already in the active range, any change in NO changes the output of those control loops. If those control loops are in a dysfunctional range because of low basal NO, increasing basal NO will move them into a more functional range.

A long term improvement in basal NO levels may (very likely) have much greater normalization of behaviors by normalizing physiology and allowing normalization of neuronal remodeling and a restoration of a more normal neurodevelopmental path. This remodeling could not occur during the short term effects observed in this study. How much is far too complicated to predict a priori, and is no doubt idiosyncratic for each individual and will depend on many other factors too, including age and how much of the ASD neurological phenotype the individual has (which depends on prior neurodevelopment including in utero).

This article is not to be taken as any type of medical advice. It is provided for informational purposes only. If anyone is interested in doing an actual clinical trial along the lines I suggest, please contact me so it can be done in the proper way with real IRBs and stuff. If you have any questions, or if you have found any errors, please let me know. I have tried to be pretty conservative in introducing new ideas, everything in here is pretty well supported in the literature, perhaps not cited as well as it should be, but I wanted to get this out fast (which is why parts of it are still kind of rough). I cite a lot of stuff, there is a lot more that I haven't cited. I haven't "cherry-picked". All of the stuff I have cited is what I consider to be pretty main stream and pretty much correct. I don't agree with every conclusion in every paper I cite, I am mostly citing the data which they report and which I then use to draw my own conclusions. My conclusion that NO is involved is "robust", that is it doesn't hinge on one or even a few assumptions that the data in the papers is correct. There are many independent lines of research that point to the same conclusion. Virtually all of them would have to be wrong for NO to not be involved.

There have been some comments that people want to use their own anecdotes of feeling and thinking worse while they have a fever to discount this data that some people with ASDs have improved behaviors. Use of anecdotes to discount other people’s data is no better than using anecdotes as being equivalent to data. Even if only a subset of people with ASDs respond this way to a fever, it may still be real, and may be useful in understanding ASDs. In any case, fevers make people feel "sick". They do not cause NTs to have ASD behaviors or the opposite of ASD behaviors. Nausea is one symptom is be made worse by a placebo. I see that as evidence that nausea is due to elevated NO (see my piece on the placebo effect). Nausea during sickness (and during pregnancy) likely relates to this. If your basal NO level is low, and is acutely raised but not to the level where you feel nauseous, then you could experience an improvement in low NO symptoms without feeling "sick".

Physiology is inherently non-linear. Using linear extrapolation from common everyday experience will be completely ineffective in estimating physiological responses in regions of the physiology parameter space which are not linearly related to the normal state. Activation of the immune system is highly non-linear with feedback and hysteresis. Any sort of linear model isn't going to capture what is important.

This result is completely consistent with my low NO hypothesis of ASDs. My explanation for the disparity between how some ASDs and NTs respond to fever is because how a neural network reacts depends on which side of the percolation threshold it is on. NO increases connectivity. If the network is above the percolation threshold, that increased connectivity moves it away from the peak and decreases functionality if it is below it, the change moves it closer and increases functionality. The increase when the network is below is much larger than the decrease when it is above.

First, is it credible?

This study was the result of one author's PhD thesis, at Johns Hopkins University. That is a first rate school, and the other co-authors are first rate scientists from first rate institutions. The journal it is published in is Pediatrics, a first rate peer reviewed journal. There are a number of anecdotal reports in the literature of the effect of fevers on ASDs which they cite. This is a prospective study, not a series of retrospective anecdotes. None of the authors are selling "fever treatments", they are serious researchers, and have written many serious research papers with multiple different collaborators and on multiple different subjects, some relating to autism, some not, published in first rate journals. There is no comparison between this report in a first rate journal with the "mercury causes autism" or the "MMR causes autism" crap. I looked at the publication records of each of the co-authors, and none of them seem like wackos. They haven't published anything wacko before (that I could find), they don't suggest a mechanism, let alone mechanisms that are implausible or that contradict anything that is well known in physiology. They don't cite anything wacko. They seem like credible responsible scientists reporting data as they measured it. The data collectors were the children's parent or guardian, the people who know them best. They would be the most accurate assessors of the child's behavior, provided that confounding factors did not introduce bias. The anecdotal comments of people in blogs expressing skepticism would suggest that any bias would be to expect worse behavior during a fever. It is doubtful that these authors are going to set up clinics to induce fevers to treat autism. They would have disclosed it, it isn't patentable, and no IRB would let them do it. I don’t think that they have an agenda they are trying to push. In any case, a paper like this isn't close to a sufficient basis to justify giving children fevers.

History of Fever Treatments and rationale based on modern understanding

There is a long history of "fever treatments" for mental disorders. Hippocrates mentioned that malaria improved epilepsy. The person who pursued it the most Wagner-Jauregg and is credited with originating it received the Nobel Prize (1927) (though he credits AS Rosenblum who published 'Relation of febrile diseases to the psychoses' in 1876-77 but which wasn't translated into English until 1943). He used malaria, typhoid and recurrent fever. This was before the days of antibiotics and any type of psychopharmacology (except for opiates and cocaine). It was called fever therapy, but it was actually inoculation with a disease that caused fever. They tried a number of different ones, and the one that seemed to work the best was malaria. Malaria due to Plasmodium falciparum is the most severe (and life threatening), and gives repeated bouts of fever due to the life cycle of the parasites as they reproduce. The type of malaria that was most used for fever therapy was the kind that is actually considered the mildest (P malariae) and so is the safest. Safety was probably the major reason it was selected. It worked, and wasn't so life threatening. It does produce fevers every 72 hours. One day of fever, then in 3 days another. It was mostly used against syphilis and the treatments were first done on people with paralysis and nerve damage from syphilis. This was called GPI (generalized paralysis of the insane), and affected mostly men, often of high intelligence, a few percent of people who were infected with syphilis. They had very good success. There were large numbers of well documented improvements. There were also not a few deaths, tens of percent. It was a desperate treatment for a debilitating disease that was certain death. Multiple types of fever producing methods were tried, inducing malaria was pretty clearly demonstrated to work the best. They would infect the patient, let them go through a number of cycles, usually around 10, then cure the malaria with quinine. The induction of malaria was the "standard of care" for neurosyphilis for decades. There was considerable resistance to using only penicillin when it was introduced.

He did try simply inducing fever using killed cultures of bacteria, this did cause remissions in the paralysis, but the syphilis and paralysis did relapse. Experiments with killed bacteria preceded those with active disease agents because while killed bacteria helped some, the results were not satisfactory. Detection of neurosyphilis is quite difficult even today. Recovery from the symptoms of paralysis does not occur instantly upon eradication of infection. Also, infection can remain even after the paralysis symptoms have gone, but the paralysis then returns (usually, but the testing of the time was not sufficient to be completely sure). It isn't exactly clear, but I think the symptoms of paralysis resolved faster with fever therapy than they did with penicillin. Which may have been why treatment with penicillin was followed up with fever therapy for years after penicillin was introduced. I think this relates to the effects of fever on neuronal function and resolution of neurological symptoms of neurosyphilis not being solely due to eradication of the infection (because symptoms would remit and then return if the infection was not cleared).

The diagnosis of the cases of insanity and psychosis were probably not up to modern standards, and some of these cases may have been due to some type of infection (as was neurosyphilis), which resolved after the infection was resolved during the fever. There were some cases of epilepsy that were treated and said to have been cured. The "causes" of many of the disorders that were treated remain mostly unknown. However, this remains true today. There is no known "cause" of psychosis, of schizophrenia, bipolar, of some other mental disorders, and no instrumental or other diagnostic tests other than interview by a clinician. Clinicians 100 years ago didn't have the DSM to consult, but then papers written about fever therapy were not written by work-a-day clinicians but by very senior researchers. There may have been some misdiagnosis, but it is not safe to assume that every successful case was a misdiagnosis. Were the successful cases only from something like neurosyphilis which had an infective cause? Perhaps, but epilepsy and psychosis isn't usually causes by an infection, and virtually all infections either are resolved, or kill the host. Neurosyphilis is a rare chronic bacterial infection. There are a few others, but it is likely unreasonable to assume that all the mental disorders other than neurosyphilis were other type of chronic bacterial diseases (which remain unknown today).

Fever treatment was used for a lot of mental disorders, labeled "insanity" or "psychosis". Precisely what those disorders were is hard to tell from the multi-hand accounts I have access to. What is notable is that some of the recoveries were quite prompt, a day after receiving an injection of tuberculin (a sterile solution containing growth products of tuberculosis) the sister of the patient remarked "What have you done with my sister? she has suddenly become intelligent".

That diverse mental conditions were effectively treated by fever therapy implies that on some level, some of the fundamental pathology in all of these different disorders is "the same", and that fever therapy was working on that common fundamental pathological state, and that restoring that fundamental state to "normal", restored normal function, sometimes extremely rapidly, in less than 1 day. That is too rapid for repair or remodeling of neuroanatomy, but must instead be a modification of a regulatory pathway(s). That a "normalization" occurs implies that the regulatory pathway is intact, it is simply exerting control at a dysfunctional operating point.

White matter hyperintensities and ATP status of the brain

In neurosyphilis MRI imaging, there are enhancements in white matter hyperintensities, (WMH ), called leukoaraiosis, which is also characteristically observed whenever there are abnormalities in brain perfusion and brain metabolism. The precise mechanism of leukoaraiosis is not well understood, however it is a decrease in the local diffusion of hydrogen, usually thought to be water as observed by MRI. My interpretation is that in the white matter (which is mostly axons, called white because of the abundant myelin covering them), a major source of that water movement (and apparent diffusion on MRI) is active axonal transport and not passive diffusion. This would explain the anisotropy of the diffusion also. WMH are observed in just about all of the neurodegenerative diseases including Alzheimer's, Parkinson's, Huntington's, Lewy body neuropathies and some others. These are all also characterized by reduced metabolism, and reduced concentrations of ATP or related energy metabolites.

All components of a neuron are made in the cell body, where the nucleus and all the DNA is. Everything then is carried out to the tippy end of the axons by ATP powered motors moving on the actin skeleton. That includes mitochondria which have only a 30 day half life (in rat CNS), it is probably longer in humans, how much longer is unknown. Under conditions of insufficient ATP (as in ischemia, hypoxia), ATP conservation pathways will kick in, and reduce ATP consumption. I think the ultimate "cause" of WMH is low ATP. It is estimated that ~50% of neuron ATP consumption is via the actin skeleton. The ATP conservation pathways would have to address such a major ATP consuming pathway. With less stuff being moved, or being moved slower, there would be less water entrained, less water movement, and so less apparent diffusion of water due to that shear and mixing. WMH does correlate pretty well with ischemia from stroke, in vascular depression, and other neurodegenerative diseases characterized by reduced brain metabolism. This axonal transport has a maximum velocity of a few microns per second. The transport is usually not continuous, but proceeds and then stops. Different cargo is transported at different velocities, but much of that difference is likely due to different times the cargo is stationary. The fastest transport is about 250 mm/day.

Actin does have regions of enhanced water mobility surrounding it, and these regions are affected by ATP levels. The addition of ATP liberates strongly bound water forming weakly bound water. In MRI, this would show up as increased water mobility at higher ATP levels.

WMH are more complicated than just ATP and active axonal transport because they are still observed post mortem, when there is complete ATP depletion and a complete absence of ATP powered motion. Are the axons partially clogged due to accumulation of cargo in "traffic jams"? Axons have a very high aspect ratio, microns in diameter and thousands or tens of thousands of times longer. A small change in cargo velocity would result in accumulation if the change were not exactly uniform across the entire length of the axon over the time scale of the change in velocity. Movement control of cargo, that is regulation of cargo velocity, has to be "local", that is it can only depend on the local conditions of the motor, the cargo, the cytoskeleton, cargo density, and the ATP level (and perhaps other local parameters). Those local parameters may have input from distant signals, but the ultimate transduction into cargo velocity must be locally determined. There are too many different individual pieces of cargo being carried for there to be external non-local control on each one. The axon 100 microns away can't "know" the status of cargo transport other than by local signals transmitted in the axon. Presumably the velocity of cargo can be regulated within some range, but as ATP depletion occurs, that active range can only go down, and eventually transport will stop.

Unless that slowing and eventual stoppage is exactly synchronous, then presumably the cargo would accumulate in slow regions, and could become "jammed", that is become so close packed that the force exerted by the ATP powered motors moving cargo into the "jam" would be insufficient to move the "jam". The shear strength or effective velocity of a granular material is a very sensitive function of the particle-particle spacing. This "jam" would show up as retarded water diffusion because the cargo would be forced into a close enough packing that the essentially motionless hydrogen in proteins becomes a larger fraction of the total hydrogen (proteins plus water).

As the transport along an axon degrades from the "normal" well controlled nearly uniform velocity, it would be expected to degrade non-uniformly, that is each axon would devolve into regions where cargo is sparse plus regions where cargo is "jammed". The effective water diffusivity in regions where cargo is sparse may be a little higher, it is not going to be much higher than normal regions. In contrast, where cargo is "jammed", the apparent hydrogen diffusivity would be expected to be lower. At the "normal" cargo density, cargo moves in both directions, and also moves at different velocities simultaneously. Thus there is lots of free volume for cargo to pass each other. The average cargo density can only change slowly, as cargo is either removed at the cell body or at the tippy ends. The effective diffusivity of water in a volume of cargo will go down as that cargo is segregated into regions of higher and lower density. The loss of effective diffusivity in the regions of highest density will not be made up for by the gain in diffusivity in the regions of lowest density. This starts to happen when the particle-particle spacing becomes less than the diffusion distance of hydrogen during the MRI scan.

A "jam" can only be cleared by withdrawing cargo from the jam at the ends where the jam is adjacent to unjammed regions. A "jam" that is present at death will not be cleared. Do axonal contents become "jammed"? Some axonal components do precipitate and become solids, presumably that occurs when the local concentration exceeds the solubility. Accumulation of inclusions is observed (eventually) in essentially all conditions where WMH are observed, Alzheimer's, Parkinson's, Lewy body neuropathies, amyloidosis, and prion disorders. These are also all conditions where brain metabolism is reduced, in some cases profoundly. They are also observed during seizures. On acute occlusion of cerebral arteries, WMH occur very rapidly, in 2.7 minutes in the rat. It has been suggested that edema is the cause of WMH, however edema does not occurs this quickly, but ATP depletion does.

There may be other changes that result in WMH too. WMH are associated with markers for hypoxia. In any case, the relevance of this diversion into WMH is only to connect WMH to the ATP status of the brain. The association of WMH with low brain ATP is pretty well established, even if the mechanism(s) for that association is not.

There is a report that WMH are associated with lower NOx levels in plasma and also with increased markers of oxidative stress. NOx in plasma is not a good measure of NO. NO concentrations are never much greater than 10 nM/L. NOx is the terminal metabolite, and is present at orders of magnitude higher levels (this report says 70 mM/L, I think that is a misprint, it is more likely to be 70 micromoles/L. Nitrate behaves like chloride, plasma levels of chloride are ~100 mM/L. This report on LPS production of sepsis in rats shows increased NOx to 300 micromolar which is more reasonable.) In this report they show the increase in NOx due to hemorrhagic shock and also due to endotoxin shock. This report shows NOx levels ~50 micromolar in humans. During sepsis NO is greatly increased due to expression of iNOS via NFkB, but the regulation is quite complicated.

NO and superoxide from iNOS are protective against excitotoxic injury, and this protection can be induced via LPS treatment.

Subjects with WMH have reduced density of blood vessels in regions which show hyperintensity. These blood vessels are often tortuous and appear as a tortuous arteriole in a cavity. These images are quite striking. The vessel is quite corkscrew-like, very tortuous and is in an empty cavity, devoid of white matter. These are sites of apoptosis. Presumably, that blood vessel must either be a source of a pro-apoptotic signal, or a sink of an anti-apoptotic signal. It turns out that vessels are sinks for NO, which is an anti-apoptotic signal. However it is not precisely the vessel that is the sink, rather it is the hemoglobin in the red blood cells that is the sink. This is how the tortuosity develops (my hypothesis). It is exactly like the meandering of a stream. The high velocity of the stream on the concave side erodes sediments which deposit in the low velocity region on the convex side. The stream migrates in the direction of the highest erosion rate. Similarly, the blood vessel migrates in the direction where NO is the lowest, where the fluid flow pushes red blood cells up against the vessel wall. With the shortest diffusion path and the highest red blood cell density and velocity, NO is lowest there, and lowest on the outside bend of the vessel. This low NO causes the tissues outside the vessel to regress (via apoptosis) and so the vessel migrates in that direction. Tortuous vessels like this are easily seen in retinopathy (where they accompany WMH) where they are caused by the same mechanism. In retinopathy, often when vessels cross there is observed to be "nicking", that is, a reduction in the diameter of the vessels. This is due to the decreased NO at the site of crossing (my hypothesis) where there is more hemoglobin to act as a sink of NO. This migration and remodeling of blood vessels by local NO levels is part of the normal regulation of capillary spacing (my hypothesis).

The reduced blood vessel density is likely due to a remodeling of the vasculature due to chronic low NO. With O2Hb being the sink for NO, if the level of NO is lower, then less hemoglobin is needed to act as a sink. I think this remodeling of the vasculature is the mechanism behind the lower brain blood flow observed in all of the neurodegenerative disorders characterized by WMH. I think it is also the mechanism for capillary rarefaction in non-neuronal tissues observed in hypertension and other disorders.

In looking more into the details of neurosyphilis, some of the neurodegenerative symptoms also bear some (faint) resemblance to Friedriech's ataxia. Friedreich's ataxia is caused by a mutation in the transcription factor that regulates the production of frataxin, an essential mitochondrial protein which locates to mitochondria, and appears to be essential in the formation and regulation of proteins with iron sulfur clusters (both inside and outside of mitochondria). This is a very large class of enzymes, largely involved in regulation of oxygen consumption for ATP production and for manipulation of oxidizing or reducing equivalents for synthesis of compounds. Deficiencies in frataxin result in mitochondrial dysfunction, iron accumulation, oxidative stress, nerve damage and eventual atrophy and nerve death. It seems that the nerves that are the longest get affected first (just like ALS).

This would be consistent with an insufficient mitochondria biogenesis etiology. Mitochondria biogenesis is triggered by nitric oxide. When there are insufficient mitochondria, they get pushed to higher potentials, where they generate ATP less efficiently (leading to a hypermetabolic state as in ALS, cardiomyopathy, obesity) and generate more superoxide which lowers the NO level more and renders the low NO state permanent (as in chronic fatigue syndrome).

Regulation of ATP concentrations in axons must be "local". That is, it must be regulated within each cell (because cell membranes are not permeable to ATP), but also within each axon because the ATP demand is "local".

In autism, there are reductions in brain concentrations of chemicals associated with brain energy status. There are reductions in T2 relaxation times in gray matter reported for children with autism. There are reports of reduced cerebral blood flow in children with autism. There are reports of neuroinflammation caused by neuroglial activation. Low NO causes the activation of microglia.

In people with autism, these signs point to low NO in the brain, which leads to low ATP via regulation of ATP via sGC. The regulation of sGC by NO and ATP and other phosphates is not simple (although this interpretation has been questioned). Acute sepsis leads to high NO from expression of iNOS, and leads to high ATP concentration. This is not generally appreciated. During acute sepsis, ATP levels are actually higher (if the patient survives) than normal controls. It is my interpretation that the authors of this last report don't appreciate what their own data clearly shows. They show higher ATP levels in skeletal muscle during sepsis than in uninfected controls (p =0.05). ATP is higher because NO is higher. High NO blocks cytochrome c oxidase, so mitochondrial ATP generation is shut down (mostly). This is why septic shock causes cachexia. The body is generating ATP via glycolysis. The mitochondria are shut off by the high ATP, so the body needs to make glucose without using ATP, so it does so by turning the muscles into alanine which the liver can turn into glucose without consuming ATP. All of this glycolysis generates a lot of lactate, which can't be turned back into glucose because the mitochondria are shut down. So the body turns it into fat. That is what septic shock does, it turns muscle into fat. Turning protein into fat and carbohydrate liberates a lot of ammonia. If that is sweated out to the skin, a resident biofilm can turn it into NO/NOx while conserve NOS substrates arginine, NADPH and O2.

Note this ATP measurement during sepsis was in muscle, however the ATP levels of all the cells in the body have to go up and down in sync for physiology to be regulated in a stable way. There has to be a "signal" that communicates the ATP level in cells, so that level can be regulated up and down in sync. This "signal" has to be uncharged to penetrate lipid membranes, and rapidly diffusible to communicate the signal quickly. It is pretty clear that the signal has to be NO. The coupling of NO and ATP via sGC makes perfect sense in this light. NO is the diffusible signal that causes all cell to regulate their ATP levels up and down "in sync". This is especially important in the brain, where everything really does need to operate "in sync" for the brain to function properly.

In ischemic preconditioning, a transient ischemia induces a physiological state which is largely protective of future episodes of ischemia within a certain time frame. Ischemic preconditioning is quite complex, and is not fully understood. I discuss a number of aspects of it in my article on the placebo effect and the one on acute psychosis. Ischemic preconditioning can be mediated by oxidative stress in multiple ways. In the first case, HIF-1α causes expression of the enzymes that mediate glycolysis, increasing capacity for glycolytic ATP production. My interpretation is that acute ischemia, hypoxia, or oxidative stress all trigger reduced ATP, and this reduced ATP triggers the ATP conservation pathways that mediate ischemic preconditioning.

Ischemic precondition is a lower ATP state, but more importantly, it is a lower ATP consuming state. ATP consuming pathways that can be shut off temporarily have been shut off, and this preserves ATP for those pathways that are more essential and cannot be shut off. Ischemic preconditioning is induced in whole organs, comprising multiple different cell types. Presumably the regulation of ATP production and consumption in those diverse cell types is regulated "in sync" for ischemic preconditioning to be effective. For cells to be regulated in sync, there must be signal(s) that passes between them to communicate the need to invoke the ischemic preconditioning pathways. We know that NO is a signal that modulates ATP level (it increases it observed in septic shock), we know that ischemic preconditioning is triggered by ROS, and we know that ROS destroys NO with near diffusion limited kinetics. If diverse cell types are all regulated "in sync", presumably they all use "the same" signaling molecules, and in "the same" control scheme. It is much easier for one (perhaps very complex) control scheme to evolve that covers all cells than for multiple control schemes controlling multiple overlapping cell types to evolve.

So, autism like neurosyphilis and like many neurodegenerative diseases is a state of lower metabolism in the brain, lower blood flow, and lower ATP in the brain. In neurosyphilis there is inflammation, in autism there are also reports of inflammation and oxidative stress. Inflammation and oxidative stress would reduce NO levels and reduce ATP levels.

So what is the mechanism(s) for the acute improvement of people with autism during fever?

Presumably nerve damage during a fever could not "normalize" behaviors (that is improve function from abnormal to more normal). Similarly if the fever caused a repair of nerve damage, that repair would not reverse itself when the fever passed. The fevers in the report are mild, likely not enough to produce damage at all.

Changes in neuroanatomy would not be so acute, and would not be so reversible. Presumably an acute and reversible effect on the function of the brain, has to do with change in the normal acute regulation of that function of the brain, that is, change in the physiological regulation of neural activity, and not a change in neuroanatomy.

If a fever can induce more normal behavior in an ASD child, obviously the brain of that ASD child has the neuroanatomy to support that more normal behavior. It simply needs better acute regulation of that neural activity to produce behaviors that are considered more normal.

Something as important as neural activity has to be under feedback control. While we don't know the details of that feedback control, we do know that it has to have a "setpoint", and mechanisms that sense deviations from that setpoint and which then act to restore what ever physiological processes are involved back to that "setpoint" when there is deviation. That "setpoint" is probably something dynamic and reflects the dynamic needs of the brain region(s) being activated (deactivated) at that particular time.

Dysregulation can either be good regulation around a bad setpoint, or bad regulation around a good setpoint (or some of both). Bad regulation can be simple. Good regulation of something as complex as the brain has to be extremely complex. A change from bad regulation to good regulation is necessarily also complex. For something "simple" to change regulation from bad to good and then back to bad, presumably a small subset of the regulatory pathways are involved. For something "simple", (a fever) to change the regulation of multiple individuals from bad to good, implies that the "bad" regulation is not idiosyncratic, but common to all of the normalized individuals. This implies the dysregulation of neural function in ASDs is good regulation around a bad setpoint and that a fever somehow shifts that setpoint(s).

This is an extremely important conclusion. It supports the idea that the ASDs are a unified set of symptoms from a unified set of causes, and not idiosyncratic with each person having a different set of reasons for their ASD symptoms. ASDs have global characteristic physical symptoms, implying that the shifted setpoint(s) is also global and affects physical as well as mental symptoms.

The brain is composed of neurons, some of which are inches long, and which transmit signals from one part of the brain to another. There is considerable thought that in ASDs, there is reduced connectivity, both constitutive (that is via neuroanatomy), and also functional (that is via the normal regulation by what ever it is that regulates neuronal activity). If the change due to fever is transient, then it is not constitutive, rather it must be functional.

As a neural network doing computations, the brain is a "small world" network with each neuron having more than 10^4 or so connections. Most connections are local (80% less than 2 mm), but a small fraction are long range. When a neuron fires, on average one "downstream" neuron fires. If on average more than one downstream neuron fired, the number of neurons firing would increase exponentially, and soon every nerve in the brain would be firing, leading to a seizure. Similarly, if less than one neuron fired the avalanche of neurons firing would extinguish and neural activity would stop. Thus there is a delicate balance between neurons firing and not firing, and exquisite control of that balance. The regulation of which and how many neurons fire is obviously an important part of the normal regulation of neural activity.

It is this regulation of nerves firing that sets the functional connectivity of the brain, and that functional connectivity is regulated to be in the near percolation threshold. All natural neural networks self-regulate in the near percolation threshold because that is where the network is most sensitive to change. The percolation threshold is a critical point; that is the properties of the network change exponentially with respect to the connectivity in the near percolation region. At the percolation threshold, the sensitivity of the network is infinite, as you go away from the critical point that sensitivity decreases exponentially.

There is a large difference between behavior of a network below and above the percolation threshold. Above the threshold, the network is connected and can function with increased reliability, but with reduced sensitivity. Below the critical point, the network becomes increasingly disconnected and the global functionality of the network rapidly collapses. I think a degree of this disconnection is (sometimes) adaptive in stress, and results in an improved capacity for multi-tasking (but only of simple tasks).

I think it is operation in the sub-critical connectivity region that causes the decreased functionality of the brains of people with autism. To increase the functionality, the connectivity of the brain must be increased. Presumably that is what is happening during the fevers that are causing normalization of behaviors. If someone with connectivity above the critical point got a fever and that increased their connectivity still more, they would experience degraded functionality.

The functionality of the network changes exponentially with respect to connectivity in the near percolation threshold. That is why the brain can be regulated. A slight change in connectivity changes its properties a large amount. This is inherently a highly non-linear process. A linear model is completely inappropriate.

To normalize the activity of an extended object as complex as the brain, presumably the "something" has to act on large parts of it, encompassing whole neurons. While the brain is well vascularized, the blood brain barrier blocks the entry of many compounds in the blood from getting into the CSF and affecting the brain. Virtually all of the compounds that do enter the brain, do so via active transport and are quite well regulated.

A fever is an acute activation of the immune system, preparing the organism to deal with what ever infectious agent has elicited the fever. A number of cytokines are produced, some of which are pro and anti-inflammatory. These cytokines don't normally enter the brain and CSF (very much) and so their presence in a fever would not be expected to result in normalized neural activity. Normal individuals don't normally express cytokines or their receptors at the levels that ASDs express cytokines during a fever, so presumably the normalization of brain function is not due to the cytokine response per se, but rather to a "normal" constituent of "normal" physiology that is transiently upregulated (or down regulated) during an immune stimulation.

There is a very recent review of neurological effects of immune system stimulation. It is mostly focused on the "sickness behaviors" that inflammation and other immune system activation cause and suggests that disorders such as depression may have components that are influenced by the immune system. While the focus is on pathology, they don't discuss any mechanisms that would have the effect of normalizing behaviors.

One effect of cytokines is expression of indoleamine 2,3-dioxygenase which lowers levels of the essential amino acid tryptophan. This does deplete levels of tryptophan, but the time course of that depletion is longer than was the course of fever in this study. Tryptophan is the precursor of serotonin. In cancer patients undergoing cytokine therapy, there is decreased mood associated with reduced serum tryptophan levels. Interferon does cause depression, and that is the major side effect of interferon therapy for cancer, hepatitis. Interferon can also cause delirium, psychosis, and even mania. Cytokines do affect brain neurochemistry. However, the levels of cytokines in non-infected individuals are usually small, and cytokines are usually associated with sickness behaviors, usually depressive symptoms. Cytokine effects have not been clearly linked to any major disease. It would seem unlikely that the cytokines which normally cause "sickness behaviors" in NTs, would result in a normalization of behavior in ASD individuals. Particularly at the mild levels likely in the modest fevers these children had. Most cytokine effects are mediated through transcription. Transcription in neurons occurs in the cell body, and the products have to be carried out to the tippy ends of axons for those axons to be affected by those transcription products.

There are other cells in the brain that are affected by cytokines that are not so extended as neurons. Cytokines might induce more rapid changes in them, however for neurological behaviors to be affected, neurons must be affected. Non-neuronal cells may be affected by cytokines and then communicate that change to neurons via some other signaling molecule. If that signaling molecule is NO, then it is the NO that is normalizing the neuron behavior. Axons are mostly covered in myelin, so any signaling molecule would need to be able to get into the neurons so as to affect them. Myelin is permeable to NO, most everything else would require specific receptors.

So what aspect of a fever is normalizing the regulation of neuronal activity?

Presumably it is "something" that is present at an appropriate level in non-ASD individuals at all times, but is transiently increased (or decreased) in these ASD individuals during their fever. Presumably it is a signaling molecule, because any feedback mechanism is essentially a signaling process.

The paper suggests a number of possible mechanisms which may be involved, although in quite generic terms.

1. neurobiological effects of selected proinflammatory and/or anti-inflammatory cytokines.
2. modification of neuronal and synaptic function secondary to variations in body temperature that influence neural conduction velocities or synaptic transmission,
3. modification of dynamic neural networks as a result of changes in cellular signal
transduction and gene transcription that regulate synapse formation and function,
4. increased production of other stress-related proteins, such as heat-shock proteins,
during fever that might modify energy consumption and mitochondrial activity,
5. stimulation of the hypothalamic-pituitary-adrenal axis leading to modifications
of neurotransmitter production and interaction.

While these mechanisms could well modify the function of the brain, it is difficult to see how the specific effects of a fever (or the immune system activation accompanying a fever) would acutely modify the function of the brain of someone with an ASD through any of these mechanisms to produce an acute normalization of function. For these mechanisms to produce acute normalization, implies that the dysfunction of ASDs is due to a dysfunctional level of such things in the non-fever state in ASD individuals. There is no evidence that any of the levels of any of these parameters are causal in ASDs. There is evidence that all of these systems are perturbed in ASDs (except for temperature perhaps), but that likely results from the coupling of all of physiology together. One aspect of physiology cannot be modified independent of other aspects. Dysregulation of diverse physiological systems in ASDs implies a shifted setpoint(s) in all of those physiological systems.

Could the regulation be improved simply by temperature? Doubtful. Perturbed body temperature isn't reported or observed in ASD individuals, and a change in regulation of neuronal function by a change in body temperature away from normal would not be expected to normalize function. Perturbing the body temperature of normal individuals to lower doesn't result in ASD behaviors. Perturbing temperature away from normal would be expected to decrease fidelity of control, not increase it. Temperature gradients in the brain during a fever are likely greater than during a non-fever simply because control is not expected to be as good. If there were temperature perturbation of normal neuronal function in ASDs, it would likely be due to specific structural or anatomical changes, changes in structures of proteins, in lipid concentrations, in lipid-protein interactions, in lipid raft properties. Different structures for common neuron components, such as ion channels sufficient to perturb temperature dependence of function in such characteristic ways would be pretty obvious in genetic studies. Genetic studies haven't found that, so that likely isn't what is happening.

Fever may modify synapse formation; however the promptness of the improvement and the promptness of the return to baseline function implies that the changes are not due to structural changes in neuroanatomy implied by synapse changes. If new synapses did form, they wouldn't disappear when the fever passed.

Stress related proteins? Stress related proteins are extremely well regulated (as are all stress responses). Psychological stress exacerbates many symptoms of ASDs. Stress related proteins are regulated locally to each cell. In a neuron, they would necessarily be synthesized in the cell body where the DNA is. They would then need to be transported out the axon to the tippy end. For the proteins to be cleared after the fever passes, they would need to be transported back to the cell body for disposal and recycling. Any kind of transcription regulation has this same problem. It takes time for the proteins to be expressed and transported to the ends of the neurons where (presumably) they have their effects. The promptness of the normalization, and the promptness of the reversion after the fever passes argues against transcription mediated mechanisms (at least transcription inside neurons).

It is likely that it is the longest length connections that are normalized. Why? There are many more short range connections, so there is more redundancy in them. Simple neurological effects such as sensory processing have not been reported to be disrupted in ASDs (they are actually improved). The effects that are disrupted are very complex neurological properties, such as behavior, social interactions and communication. The importance of a few long range connection in the brain (as a "small world" network) are most important in determining global properties, less important in determining local properties such as sensory processing.

Pro-and anti-inflammatory cytokines can have large effects on most any tissue, but there isn't much evidence that they are responsible for the acute neuronal regulation of the types of behaviors that are characteristic of ASDs. Immune suppression drugs have not been reported to produce a normalization of ASD behaviors.

No doubt there are many compounds with different concentrations in ASD and non-ASD individuals. However to normalize behaviors, a compound with a changed concentration during a fever must have specific properties and effects. To summarize, the agent responsible for a normalizing of neuronal behavior during a fever in ASD individuals has to have a number of properties:

1. normal CNS signaling molecule
2. active range in both fever and non-fever
3. present in normal CNS, body
4. perturbed levels in ASD CNS, body
5. change in fever results in a more normal level in ASDs
6. passes through blood brain barrier
7. change up and/or down is rapid
8. change not easily induced by other means

CO2 is reportedly decreased during fever (observed as an increase in blood pH due to hyperventilation). Supplemental CO2 does extinguish those febrile seizures. In rats, maximum susceptibility to febrile seizures occurs at an age when the ventilatory response to CO2 is at a minimum. However neither hypercapnea nor hypocapnea produce symptoms characteristic of ASDs, so a resolution of hypercapnea during a fever would not be expected to resolve them. Rett Syndrome is characterized by breathing regulation irregularities, which may well indicate abnormal regulation of breathing by CO2. However other ASDs are not characterized by breathing abnormalities. People with known genetic causes of autism (such as Rett Syndrome) were excluded from this study.

Nitric oxide does have all of these properties. It is lower in autism and ASDs, it is acutely increased in fever. During immune system activation, iNOS is expressed, and NO levels are increased. The production rate of NO can get very high. However, the basal level of NO still does not get above ~10 nM/L (0.3 ppb) which is the EC50 for sGC. If the basal level did get to 10 nM/L, there would be 50% activation of sGC and generalized vasodilation leading to severe hypotension. There can be severe hypotension during septic shock, and that can be sufficient to be fatal, but it isn't that high in a mild fever, and didn't get that high in these ASD individuals who normalized their behavior during their fever (if it did, they would have collapsed from hypotension).

The low NO ratchet

The mechanism for the NO increase is that activation of NFkB causes the expression of iNOS, which generates NO via open loop control, that is, the NO generated is not regulated other than by the amount of iNOS produced and the availability of substrates and the presence of inhibitors. This NO inhibits NFkB and prevents the expression of more iNOS. Thus the level of NO before NFkB activation determines in part the amount of NO after NFkB activation. However it is an inverse regulation. The lower the initial NO level, the higher the iNOS expression and the higher the ultimate NO level. This high NO level then reduces the expression of eNOS and nNOS, lowering the basal NO level when the iNOS is degraded in a day or so.

This behavior of NO physiology is extremely important. When basal NO is low, any immune system activation raises NO levels higher (due to less inhibition of NFkB) than if NO was higher before immune system activation. I hypothesize that this can lead to what I call the low NO ratchet, where activation of the immune system under conditions of low basal NO causes basal NO levels to ratchet lower each time the immune system is activated. I think this is one of the fundamental causes of regressive autism, and also of low basal NO in adults as characterized in chronic fatigue syndrome (CFS). Many people with CFS can identify when they acquired it, and it corresponded with an acute serious infection. Similarly, many parents anecdotally identify the immune reaction of a vaccination as a precipitating event leading to regression. However the large scale epidemiology shows no change in incidence of autism with changes in vaccination. My hypothesis is that in susceptible individuals, any immune system activation is sufficient to activate the "low NO ratchet", a vaccination, or one of the zillions of infections of childhood. It is the low NO ratchet that (I hypothesize) causes Gulf War Syndrome. Receiving multiple immune system activations (vaccinations) during a high stress period (being deployed to a war zone) causes basal NO to ratchet lower with each immune system activation until it saturates and produces chronic fatigue. This takes a few weeks, while the mitochondria turn-over and are not replaced (due to the low NO level from the chronic stress). Once mitochondria numbers are low, the low NO state is perpetuated due to superoxide from too few mitochondria being pushed to higher potentials to supply the same ATP. With continually low NO, mitochondria biogenesis can't occur enough to get back to the level that is "normal".

There are some reports that there are WMH observed in a subset of CFS patients, and some that there are not. The primary symptom of CFS is muscle fatigue, implying insufficient mitochondria in muscle to supply sufficient ATP for continued exertion. Mitochondria biogenesis in both tissues is initiated by NO, but the details of that regulation, including time constants may be quite different. Muscle has storage of ATP equivalents as phosphocreatine in much larger quantities than does the brain, and muscle can derive ATP from glycolysis which neurons cannot. While there should be some similarities, we should not expect identical behaviors.

If an acute fever reduces the symptoms of ASDs, is malaria "protective" against acquisition of an ASD? It would appear not. The epidemiology of ASDs in regions where malaria is endemic is not good, however there was a case series from Tanzania where medical histories of 14 children with autism were examined, and 3 of 14 were reported to definitely occur precipitously with malaria infection, 4 of 14 possibly coincident, including cerebral malaria and one case of Salmonella meningitis. 7 of 14 were reported not associated with a severe infection. The onset of autism in many of these children was many years earlier. Presumably they continued to be exposed to malaria while living where malaria was endemic and this ongoing malaria exposure did not "cure" them of their autism. This is malaria from P. falciparum which is the most severe type, and not the milder form from P malariae. 4 of 14 cases had dysmorphic facial features, and none of them had genetic work-ups, so how comparable these cases are to standard cases where malaria is not endemic is unknown. The autism in some of these cases may be quite different than in the cases in the current fever study (which did exclude known genetic causes). There may have been brain damage due to these very serious infections. The degree of immune system stimulation from cerebral malaria is completely different than the mild immune system stimulation from childhood vaccines. Childhood malaria is a leading cause of death where it is endemic.

The blood brain barrier is completely transparent to NO which has physical diffusion properties close to that of O2. Normally, blood is not a source of NO, but rather O2Hb is the sink for NO. NO is produced by the endothelium by eNOS, and also in nerves by nNOS.

Precisely where immune system activation increases NO production is not well characterized. Mostly it is thought to be generated in immune cells, which are spread throughout the body. Some in the blood, but most in the extravascular space and especially in the lymph nodes and at specific sites of infection where those cells are attracted. The NO they produce locally is important in inducing the hyperemia of inflammation. The supply of O2Hb by that hyperemia keeps the NO level from getting much above ~10 nM/L. Sites of infection do attract immune cells, and the NO they produce is enhanced under conditions of fever. Local concentrations might get very high, but those very high concentrations can't be systemic or there would be systemic vasodilation and systemic hypotension.

Normal brain activation can be monitored using BOLD fMRI. What is measured is the magnetic susceptibility of deoxyhemoglobin and how the concentrations of that change before, during and after neuronal activation. The precise physiology that is going on is mostly unknown. The suggestion is that the increased blood flow is primarily nutritive, however the flow exceeds the nutritive demands to support the increased metabolic load. There has been a suggestion that NO is related to this. I think the idea that NO is related is precisely correct, and that in fact the dominant physiological effect is the release of NO. This NO blocks O2 uptake by cytochrome c oxidase, increasing O2 levels. The NO also activates sGC producing cGMP which relaxes vascular smooth muscle and causes vasodilation and increased flow of O2Hb to the NO affected site. This O2Hb then takes up the NO and converts it into nitrite and nitrate at near diffusion related kinetics. The flow takes the nitrite and nitrate away and the vasodilation subsides. The vasodilation directly relates to the magnitude of the NO signal, and so the blood flow directly relates to the NO signal. Under conditions of low basal NO, the NO signal reaching sGC is less, less cGMP is produced, there is less vasodilation, and less blood flow. I think that low basal NO is the precise reason for the reduced cerebral blood flow observed in autism, and in other neurological disorders characterized by reduced blood flow.

Because the "normal" increased blood flow exceeds the nutritive requirements for the metabolic burst in activity, when that blood flow is reduced, so will be the metabolic burst in activity. Over time, the metabolic rate of that region of the brain will be reduced. This is not "dysfunction", the metabolic rate is being controlled by the "normal" control parameters, the "problem" is that the "setpoint" is bad. This is how low basal NO leads to eventual neurodegeneration.

This NO has a lot of simultaneous effects in the local site. NO regulates long term potentiation. NO binds to heme and inhibits heme enzymes in mitochondria, but also in microsomes where the cytochrome P450 enzymes make a variety of neuroactive compounds such as steroids and prostaglandins. NO destroys superoxide, and so regulates signaling via superoxide. When superoxide is destroyed by NO, peroxynitrite is produced, which can nitrate proteins and affect enzyme activity.

NO diffuses through myelin, and so NO can affect mitochondria (and other things) inside axons that merely pass through the NO affected region. That is, there need not be any synapses or other neural terminations for NO to have effects on axons in the vicinity of the NO release. Nerves that are not connected by a synapse must communicate in some way before a synapse can be formed.

NO regulates synaptic remodeling, including synaptogenesis following injury, no doubt it regulates it before injury too. Many growth factors have effects mediated by NO, they are endocytosed in receptors where nNOS is activated and the combined receptor/growth factor/nNOS complex is transported to endosomes for recycling. NO may modify those endocytosed complexes via nitration.

There are many ways by which NO affects neural activity. I am cutting short the discussion here because I could go on for many pages, and that isn't necessary. There are many mechanisms by which NO regulates neural activity. Which ones are important in which region of the brain is unknown but it is likely that many are important with different levels of importance in many (probably all) regions of the brain, and that those differences may well be idiosyncratic and depend on genetic as well as epigenetic factors.

Many neural mechanisms known to be affected in autism have effects mediated by NO, and are affected in the direction that would be expected if the basal NO level were lower. For example, maternal bonding is mediated through oxytocin and NO. While NOS is inhibited postpartum, ewes do not develop smell memories of their lambs. If NOS is inhibited and NO is supplied they do. Maternal bonding is the primary social behavior in mammals. If that depends on NO, likely many other social behaviors (if not virtually all) do too.

Many organs are known to be programmed in utero in humans, including heart, liver, kidney, vasculature, pancreas, immune system, bone, endocrine system, reproductive system, muscle. Some of those programming mechanisms involve NO. It would be surprising (actually beyond surprising, it would be inconceivable) if the most important organ, the brain were not programmed in utero too. There are some indications of long term programming of the brain, for example the cycle of violence.

Therapeutic considerations.

The objective of any treatment is to restore normal function in the long term without producing adverse effects. That fever produces a transient improvement implies that function is only abnormal due to good regulation around a bad setpoint. To successfully treat this requires shifting the setpoint long term. Simply moving the setpoint temporarily (as in a fever) won't produce long term normalization. Attempting to do this long term (as in a permanent fever) is extremely likely to induce side effects, because physiology will most likely adapt to the perturbation.

In fever therapy, malaria was used to induces multiple bouts of fever in rapid succession (with a 4 day period). The time constant for iNOS expression and degradation is about a day or so. One of the things that maintains low basal NO is reduced mitochondria number, which pushes mitochondria potential to higher levels (to generate the same ATP), which generates more superoxide which perpetuates low NO and retards mitochondria biogenesis. NO from iNOS during a fever would override all of that and produce high NO during mitochondria biogenesis and so increase mitochondria levels. Mitochondria are generated at night during sleep, so multiple cycles of fever as used in fever therapy could result in increased mitochondria numbers and may resolve a condition of insufficient mitochondria. This might be a mechanism by which multiple bouts of fever may have very different effects than single bouts. Single episodes of fever may activate the "low NO ratchet" and worsen chronic low NO. Multiple bouts of fever may activate processes with different time constants (for example mitochondria biogenesis) which may over multiple episodes counter the adverse effects which may occur in single episodes. NO physiology is not well enough understood to predict this a priori, and some individuals may have idiosyncratic responses.

NO physiology is under intense regulation. There are no generally accepted methods for raising NO levels long term. Many pharmacological methods have been tried. There has been no report of a long term successful trial at raising NO systemic NO levels or in ameliorating chronic disorders characterized by low NO through any pharmacological treatment. There are short term methods that have been used, but they only work short term. So far there has always been compensation if the study was extended long enough. Most studies have been too short to show this.

NO can be given in inhalation air, however there are essentially no systemic effects. The NO is converted to nitrite and nitrate by oxyhemoglobin in the lung even before the blood leaves the lung. In air, NO is rapidly oxidized to NO2, which is toxic at ppm levels (NO is non-toxic at hundreds of ppm). Inhaled NO must be continuously monitored for NO2, and can only be done in a hospital setting.

NO can be given intravenously, as actual NO, or as sodium nitroprusside. NO and SNP have very short lifetimes (few minutes at most), and can only be administered in hospital settings.

Organic nitrates are sometimes (erroneously) referred to as "NO donors". They are not NO donors. Organic nitrates such as nitroglycerine are sometimes given, however they produce what is called "nitrate tolerance", where the therapeutic effect are diminished and oxidative stress effects occur. The precise physiological details behind these effects are unknown.

L-arginine is the substrate for nitric oxide synthase, and when given acutely, does transiently raise NO levels. However over time, the increased NO is diminished and long term use of L-arginine doesn't improve low NO mediated diseases such as peripheral artery disease and perhaps may lead to "arginine tolerance". This is thought to be due to upregulation of NOS inhibitors such as asymmetrical dimethyl arginine and arginase which turns arginine into ornithine plus urea. Production of urea via arginase may be a mechanism to increase urea excretion in sweat to increase NO/NOx via biofilm (my hypothesis)

Oxidative stress does lower NO levels by consumption of NO via superoxide. There is some data in the literature that short term administration of antioxidants does transiently resolve some symptoms of oxidative stress and presumably those of low NO. However there is no data to suggest that long term administration of antioxidants has any therapeutic benefit and much to demonstrate that it doesn't. Every long term placebo controlled trial of antioxidants has shown zero or slightly negative effects. A large meta-analysis has shown this quite clearly. It has been suggested that oxidative stress is a regulated parameter of physiology, and that a state of oxidative stress is due to good regulation around a bad setpoint rather than bad regulation around a good setpoint. This is discussed in Chapter 8 of Oxidative Stress: Clinical and Biomedical Implications. This explanation fits the data observed in all the large placebo controlled trials, and also explains the very robust observation that a self-selected diet rich in antioxidants is inversely correlated with the diseases of oxidative stress. The conclusion is that a self-selected diet is part of the control system that physiology uses to regulate the state of oxidative stress that the setpoint calls for. Excess antioxidants must be destroyed at some metabolic cost; in a high oxidative stress metabolic state it is better to not consume them in the first place. In other words, individuals with a high oxidative stress setpoint self-select a diet low in antioxidants. This is why in the JAMA review article consumption of excess antioxidants correlates (slightly) with adverse effects. There is a metabolic cost to destroying those excess antioxidants.

Whole body vibration does produce bone strengthening. The primary mechanism for regulation of bone stiffness is via NO released during bone strain. This NO increases deposition of bone mineral where NO levels are highest, which corresponds to where bone has the lowest stiffness. Whole body vibration in sheep does release NO and reduced an immunogenic hyper-response and airway inflammation in sheep on antigen challenge. Whole body vibration is reported to be beneficial in muscle function and balance in the elderly. That is an effect that could be mediated by NO. However effects on bone density and muscle strength may be mediated through NO generated via vibration in those tissue compartments and by exerting only local effects. There may be no systemic effects, or no effects on the brain. Local vibration can cause neuropathy and also Raynaud's phenomena. Vibration is probably not a good method to try and raise NO levels in the brain. In vibration produced neuropathy, the conduction velocity of peripheral nerves is reduced. A reduction in conduction velocity in the brain would likely cause greater dysfunction due to asynchronous operation.

Meditation does raise NO levels, and systemic increases in NO have been measured. This is thought to be the mechanism for many of the beneficial effects of meditation. While meditation might be effective, teaching autistic children to meditate is likely difficult. I suspect that many "stimming" behaviors are actually things that increase NO levels which is why they are done, particularly during times of stress when more NO is needed. Humming certainly does.

Parasitic intestinal worms are an effective treatment of Crohn's disease and infection with helminthes and also malaria increases production of NOx metabolites in humans, demonstrating that NO production is upregulated via intestinal parasite infection and also during malaria infection. The precise interaction of intestinal parasites and malaria and NO is somewhat controversial. Some suggest increased NO is pathological and exacerbates malaria however for our purposes all we need to know is that malaria increases NO. NO is a very good anti-inflammatory agent (due to its inhibition of NFkB), and the positive effects of intestinal worms on Crohn's disease are likely mediated through NO as well. It has been suggested that the NO produced during infestation with intestinal worms is protective against malaria.

People with malaria induced anemia and people who have been infected with malaria and have recovered do have a higher basal level of NO/NOx and their immune cells have greater NO/NOx production capacity. NO/NOx levels correlate with severity of malaria infection.

However, while NO/NOx in malaria exposed individuals is higher, this high NO/NOx production is not always well correlated with higher NO/NOx production by blood mononuclear cells, suggesting that there is another source of NO/NOx. The NO/NOx production in individuals was stable over time and didn't correlate (in individuals) with their malaria status, suggesting that something else was controlling the NO/NOx level. They found NO/NOx production was much higher in the study subjects from rural PNG than in urban adult controls from northern Australia. This study was very interesting in that they tested the same individuals after their malaria had been cleared with antimalarial drugs (2-4 weeks later), and again (7 weeks after initial screening). They spend some time discussing the implications of a non-correlation of NO/NOx status with parasitemia, and how to rationalize this with earlier studies showing a correlation. I suspect that they haven't accounted for the source that I am studying. This source was not observed in controls from urban areas, and (my hypothesis) it may be a biofilm of ammonia oxidizing bacteria.

Repeated bouts of fever and infection is probably a more "normal" state than the disease free state that people like to maintain and are able to maintain in modern society. In other words, during evolutionary time, humans were not disease free, they evolved with a significant load of parasites and diseases. The NO regulatory systems probably work better under normal conditions of illness than in the abnormal disease free state. Restoring the NO status of the body to a more "natural" state will probably improve NO regulation.

For fever therapy to achieve long term recoveries, it had to restore regulation of neuronal physiology long term, that is even after the malaria was cured with quinine. Modern measurements to show that NO/NOx levels are raised following malaria infection even after that infection has been cleared. I think this strengthens the case for NO/NOx elevations during a fever as being at least part of the mechanism for the short term resolution of behavioral symptoms and the long term resolution following fever therapy with malaria.

Mitochondrial disorders are associated with white matter hyperintensities. I think that long term oxidative stress is a generic factor that can lead to (or exacerbate) essentially any type of neurodegeneration via inhibition of mitochondria biogenesis and reduced ATP levels and reduced transport of essential cell components in axons. I suspect that the paralysis of neurosyphilis is actually due to the low NO from neuroinflammation, rather than from destruction of nerves. If it were from nerve destruction, the paralysis wouldn't be reversible in a day. If the state of oxidative stress is due to insufficient mitochondria, then the high NO state from fever therapy would have to persist long enough for the mitochondria number to be raised sufficiently to drop the mitochondrial potential, reduce the superoxide production, and normalize the basal NO level. This has to occur over the entire length of the axons in the brain. This cannot happen in a short period of time. Mitochondria must be synthesized in the cell body and transported out the axons. Presumably it would take on the order of the normal mitochondria lifetime to attain a new stable steady state. 10 cycles of fever from P malariae would take 40 days. This is probably long enough for a new steady state to be produced. It would be unreasonable to expect recovery in less time. Even if behaviors are normalized, the feedback regulatory system may not be. It is the feedback system that must be normalized for any recovery to persist long term.

Exposure to pyrogens such as lipopolysaccharide (LPS) causes acute expression of iNOS and acutely higher NO levels. LPS also causes LPS tolerance, where there is a refractory period where additional LPS does not elicit as much of a response. That tolerance is due in part to NO generated by iNOS in activated macrophages. I suspect this feedback inhibition is what prevented non-infective fevers from being effective treatments of neurosyphilis. The high NO state was not maintained long enough for the regulatory pathways to compensate and form a new steady state.

Fever therapy was first tested and was in use before the modern era of daily bathing and daily hair washing with detergents and antimicrobial agents. I think it is likely that institutionalized patients were not bathed every day, and so had an abundant biofilm of the bacteria that I am working with. Malaria is characterized by fever, sweating and chills. The periods of sweating would have nourished any biofilm of ammonia oxidizing bacteria and would likely have contributed to any therapeutic effect of increased NO. Similarly the above results on NOx from malaria and intestinal parasites were done on native people living in "the wild", so presumably they had a more natural biofilm of commensal bacteria, but the actual status of their biofilm is unknown. Increased sweating providing more substrate to their biofilm may have been one of the mechanisms for the increased NOx that was observed.

Presumably the ASD children in the fever study did bathe with normal frequency and so did not have a biofilm of ammonia oxidizing bacteria, and any NO they produced during their fever was generated by iNOS.

A more constant (but still regulated) source of NO, from a biofilm of autotrophic ammonia oxidizing bacteria will improve natural regulation of NO physiology. How much more, and will that improvement be enough to achieve the effects of this study of fever in autism is unknown and unpredictable a priori. However, if any symptoms of autism are caused by low basal NO, then raising basal NO levels will improve those symptoms with no threshold.

NO/NOx from a biofilm of the bacteria I am studying is the method that I prefer and would suggest. It is the only method that is under direct physiological control via sweating. Sweating releases ammonia to the biofilm, and the biofilm delivers NO and nitrite in less than a minute. I think this is the reason for non-thermal sweating, as adrenergic sweating during stress, shock, and when a fever breaks. These autotrophic ammonia oxidizing bacteria are common in the environment, are incapable of growth on any media used for isolating pathogens, and have never been implicated in any infection. An "infection" by them may not be possible (even in immunocompromised individuals) because they are obligate aerobes and lack all virulence factors and cannot metabolize or derive energy from animal substrates other than ammonia.

If my hypothesis is correct, that humans evolved to have these bacteria growing on our skin and providing us with regulated levels of NO/NOx, then normal physiology cannot be attained without them. There is no conceivable artificial mechanism that could provide the same regulatory fidelity in terms of time, dose, and delivery mechanism. NO physiology remains mostly unknown. We know that there are thousands of pathways that involve NO. We don't know the complete details of even a single one of them and we know they are all coupled. There is simply no conceivable method for artificially regulating NO to achieve the proper levels for optimum function of a single one of those thousands of pathways let alone for all of them simultaneously.

In summary

I have shown a plausible case for the improved behaviors observed in this study being due to an acute increase in NO due to expression of iNOS. There may be other effects from other immune system compounds, but using actual disease organisms treat autism the way that malaria was used to treat neurosyphilis is a very high risk approach. Unless the immune stimulation is long term (as in parasitic worm infestation), rebound and the "low NO ratchet" effect is to be expected. This could well exacerbate ASD symptoms in the long term, even if there is improvement at each cycle.

Any questions?