The fact that not everyone responds to Simufilam is irrelevant to its eventual FDA approval

A very intelligent friend does not share my optimism about Simufilam.

“Is the data really that positive? ADAS-Cog mean scores changed minimally over 1 year in patients with mild-to-moderate Alzheimer’s disease.  47% of patients improved ADAS-Cog over 1 year by 4.7 points. But 23% of patients declined by <5 points. Mild patients responded better than patients with moderate Alzheimer’s.”

Why are these thoughts irrelevant to the eventual approval of Simufilam by the FDA?

First: no drug for anything works for everyone with the condition

Second: The assumption that Alzheimer dementia is a single disease is based on just that: an assumption.

An example: When I was running a muscular dystrophy clinic in MonrN (’71 – ’87), we saw something called limb girdle muscular dystrophy , in which the patients were weak primarily in muscles about the shoulders and hips. Now we know that there are at least 13 different genetic causes of the disorder.

If the clinical picture of Alzheimer’s disease is due to multiple causes, it is unsurprising that Simufilam doesn’t help all of them.

Also it is time for some humility about our knowledge about Alzheimer’s disease.  We have misunderstood what the senile plaque of Alzheimer’s disease really is for 111 years — see the following post written 12/22 — https://luysii.wordpress.com/2022/12/13/111-years-of-study-of-the-alzheimer-plaque-still-got-it-wrong-until-now/

Third (and probably the most relevant for FDA approval):  Less that perfect drugs will be approved if every other treatment is worse.

The example of immune checkpoint blockade therapy for cancer is particularly relevant.

Some absolutely spectacular results for the therapy has led to the approval of 6 different drugs in this class (all of them monoclonal antibodies against proteins involving the immune system).

One example [ Cell vol. 162 pp. 1186 – 1190 ’15 ]:  “20% of metastatic melanoma patients are cured with Ipilimumab, a fully humanized anti-CTLA4 monoclonal antibody.”

Would that results like this were the rule not the exception. Unfortunately — [ Nature vol. 565 pp. 43 – 48 ’19 ] “Most patients with cancer either do not respond to immune checkpoint blockade or develop resistance to it.”

So what.

Immune checkpoint blockade, despite being less than perfect,  is  still being offered to cancer patients, just the way Simufilam with its nearly 50% chance of improvement at 1 year should be offered to Alzheimer patients.  

111 years of study of the Alzheimer plaque still got it wrong (until now)

The senile plaque of Alzheimer’s disease has been known for 111 years  which is when Alzheimer’s first patient died and he studied her brain. For the past 60 or so years, we’ve studied it using every technique at our disposal.  We know its chemistry fairly  well, and understand many of the mutations that cause the familial forms of Alzheimer’s disease.

However, we’ve still been interpreting its structure incorrectly until this month.  In addition to the amorphous gunk of the plaque, electron microscopy has described swollen ‘dystrophic neurites’ in and surrounding the plaque.  The semantics of neurites implies a small nerve process which led us all down the garden path to assume that they are dendrites (which are usually smaller than axons).  Wrong, wrong, wrong, they are axons as a recent paper proves conclusively [ Nature vol. 612 pp. 328 – 337 ’22 ].

It took a lot of technology to reach this point.  First was development of the 5XFAD mouse which gets plaques galore, because it contains 5 mutations spread over two proteins, the amyloid precursor (APP) protein from whence the aBeta peptide of the senile plaque and PSEN1 a protein which helps to process APP into aBeta.  Second was the ability to observe dendrites and axons in the living (mouse) brain for long periods using specialized microscopic techniques and a variety of dyes and fluorescent proteins.  They allow us to watch action potentials pass along axons without sticking an electrode into them (by measuring rapid changes in local calcium concentration).

Each senile plaque contained hundreds of axons with focal swellings (the dystrophic neurites).  Most were present for months, but some disappeared without axon loss.  When an action potential got to a focal swelling (also known as a spheroid) it slowed down (the swelling acts as a sink for the current  due to its ability to store ions  (higher capacitance).  Random slowing of nerve conduction is murder for information processing.  It’s old technology but just think of what happens when you play  of  a 33 rpm record at 78 rpms.  It’ s also why the random demyelination (which changes action potential velocity)  of nerve fibers in MS raises hob with information transmission hence neurologic function.

Why did electron microscopy miss this?  Because it is just a two dimensional (very thin) slice of dead brain.

The paper has a lot more about what’s in the swelling — large endolysosomal vesicles, and a possible way to treat Alzheimer’s — genetic ablation of phospholipase D3 (PLD3) was able to reduce the average size of the dystrophic neurities and improve axon conduction.

It’s actually a hopeful paper, because we’ve been assuming that the dystrophic neurites were either dead, severed  or nonfunctional, and here they are intact and conducting nerve impulses.

Like all great scientific papers, it raises more questions than it answers.  Is the swelling due to extracellular aBeta?  Is the swelling an attempt to internalize aBeta and destroy it?  Is there a way to inhibit PLD3 ?   Genetic ablation of a gene in a living human is at or beyond our current technology.

Amyloid structure at last !

As a neurologist, I’ve been extremely interested in amyloid  since I started in the late 60s.  The senile plaque of Alzheimers disease is made of amyloid.  The stuff was insoluble gunk. All we had back in the day was Xray diffraction patterns showing two prominent reflections at 4 and 9 Angstroms, so we knew there was some sort of repetitive structure.

My notes on papers on the subject over the past 20 years contain  about 100,000 characters (but relatively little enlightenment until recently).

A while ago I posted some more homework problems — https://luysii.wordpress.com/2021/09/30/another-homework-assignment/

Homework assignment #1:  design a sequence of 10 amino acids which binds to the same sequence in the reverse order forming a plane 4.8 Angstroms thick.

Homework assignment #2 design a sequence of 60 amino acids which forms a similar plane 4.8 Angstroms thick, such that two 60 amino acid monomers bind to each other.

Feel free to use any computational or theoretical devices currently at our disposal, density functional theory, force fields, rosetta etc. etc.

Answers to follow shortly

Hint:  hundreds to thousands of planes can stack on top of each other.

 

If you have a subscription to Cell take a look at a marvelous review full of great pictures and diagrams [ Cell vol. 184 pp. 4857 – 4873 ’21 ].

 

Despite all that reading I never heard anyone predict that a significantly long polypeptide chain could flatten out into a 4.8 Angstrom thick sheet, essentially living in 2 dimensions.  All the structures we had  (alpha helix, beta pleated sheet < they were curved >, beta barrel, solenoid, Greek key) live in 3 dimensions.

 

 

So amyloid is not a particular protein, but a type of conformation a protein can assume (like the structures mentioned above).

 

 

So start with NH – CO – CHR.  NH  CO and C in the structure all lie in the same plane (the H and the side chain of the amino acid < R >  project out of the plane).

 

Here’s a bit of elaboration for those of you whose organic chemistry is a distant memory.  The carbon in the carbonyl bond (CO) has 3 bonding orbitals in one plane 120 degrees apart, with the 4th orbital perpendicular to the plane — this is called sp2 hybridization.  The nitrogen can also be hybridized to sp2.  This lets the pair of electrons above the plane roam around moving toward the carbon.  Why is this good?  Because any time you let electrons roam around you increase their entropy (S) and anything increasing entropy lowers their free energy (F)which is given by the formula F = H – TS where H is enthalpy (a measure of bond strength, and T is the absolute temperature in Kelvin.

 

 

So N and CO are in one plane, and so are the bonds from  N and C to the adacent atoms (C in both cases).

 

You can fit the plane atoms into a  rectangle 4.8 Angstroms high.  Well that’s one 2 dimensional rectangle, but the peptide bond between NH and CO in adjacent rectangles allows you to tack NH – CO – C s together while keeping them in a 3 dimensional parallelopiped 4.8 Angstroms high.

 

 

Notice that in the rectangle the NH and CO bonds are projecting toward the top and bottom of the rectangle, which means that in each plane  NH – CO – CHR s, the NH and CO are pointing out of the 2 dimensional plane (and in opposite directions to boot). This is unlike protein structure in which the backbone NHs and COs hydrogen bond to each other.  There is nothing in this structure for them to bond to.

 

 

What they do is hydrogen bond to another 3 dimensional parallelopiped (call it a sheet, but keep in mind that this is NOT the beta sheet you know about from the 3 dimensional structures of proteins we’ve had for years).

 

 

So thousands of sheets stacked together form the amyloid fibril.

 

Where does the 9 Angstrom reflection of cross beta come from?  Consider the  [ NH – CHR – CO ]  backbone as it lies in the 4.8  thick plane (I never thought such a thing would be even possible ! ).  It curves around like a snake lying flat.  Where are the side chains?  They are in the 4.8 thick plane, separating parts of the meandering backbone from each other — by an average of 9 Angstroms

 

Here is an excellent picture of the Alzheimer culprit — the aBeta42 peptide as it forms the amyloid of the senile plaque

https://pubmed.ncbi.nlm.nih.gov/27355699/#&gid=article-figures&pid=figure-7-uid-6.

 

 

You can see the meandering backbone and the side chains keeping the backbone apart.

 

 

That’s just the beginning of the paper, and I’ll have lots more to say about amyloid as I read further.   Once again, biology instructs chemistry and biochemistry giving it more “things in heaven and earth, Horatio, than are dreamt of in your philosophy.”

The other uses of amyloid (not all bad)

Neurologists and drug chemists pretty much view amyloid as a bad thing.  It is the major component of the senile plaque of Alzheimer’s disease, and when deposited in nerve causes amyloidotic polyneuropathy.  A recent paper and editorial casts amyloid in a different light [ Cell vol. 173 pp. 1068 – 1070, 1244 – 2253 ’18 ].  However if amyloid is so bad why do cytomegalovirus, herpes simplex viruses and E. Coli make proteins to prevent a type of amyloid from forming.

Cell death isn’t what it used to be.  Back in the day, they just died when things didn’t go well.  Now we know there are a variety of ways that cells die, and all of them have rather specific mechanisms.  Apoptosis (aka programmed cell death) is a mechanism of cell death used widely during embryonic development.  It allows the cell to die very quietly without causing inflammation.  Necroptosis is entirely different, it is another type of programmed cell death, designed to cause inflammation — bringing the immune system in to attack invading pathogens.

Two proteins (Receptor Interacting Protein Kinase 1 — RIPK1, and RIPK3) bind to each other forming amyloid, that looks for all the world like typical amyloid –it binds Congo Red, shows crossBeta diffraction and has a filamentous appearance.  Fascinating chemistry aside, the amyloid formed is crucial for necroptosis to occur, which is why various bugs try to prevent it.

The paper above describes the structure of the amyloid formed — unusual in itself, because until now amyloid was thought to involve the aggregation of a single protein.

The proteins are large: RIPK1 contains 671 amino acids, and RIPK3 contains 518.  They  both contain RHIMs (Receptor interacting protein Homotypic Interaction Motifs) which are fairly large themselves (amino acids 496 – 583 of RIPK1 and 388 – 518 of RIPK3).  Yet the amyloid the two proteins form use a very small stretches (amino acids 532 – 543 from RIPK1 and 451 – 462 from RIPK3).  How the rest of these large proteins pack around the beta strands of the 11 amino acid stretches isn’t discussed in the paper.  Even within these stretches, it is two consensus tetrapeptides (IQIG from RIPK1, and VQVG from RIPK3) that do most of the binding.

Even if you assume that I (Isoleucine) Q (glutamine) G (glycine) V (valine) occur at a frequency of 5%, in our proteome of 20,000 proteins assuming a length of amino acids IQIG and VQVG should occur 10 times each.  This may explain why 300/20,000 of our proteins contain a 100 amino acid  segment called BRICHOS which acts as a chaperone preventing amyloid formation. For details see — https://luysii.wordpress.com/2018/04/01/a-research-idea-yours-for-the-taking/.

Just another reason to take up the research idea in the link and find out just what other things amyloid is doing within our cells in the course of their normal functioning.

 

Will flickering light treat Alzheimer’s disease ?

Big pharma has spent zillions trying to rid the brain of senile plaques, to no avail. A recent paper shows that light flickering at 40 cycles/second (40 Hertz) can do it — this is not a misprint [ Nature vol. 540 pp. 207 – 208, 230 – 235 ’16 ]. As most know the main component of the senile plaque of Alzheimer’s disease is a fragment (called the aBeta peptide) of the amyloid precursor protein (APP).

The most interesting part of the paper showed that just an hour or so of light flickering at 40 Hertz temporarily reduced the amount of Abeta peptide in visual cortex of aged mice. Nothing invasive about that.

Should we try this in people? How harmful could it be? Unfortunately the visual cortex is relatively unaffected in Alzheimer’s disease — the disease starts deep inside the head in the medial temporal lobe, particularly the hippocampus — the link shows just how deep it is -https://en.wikipedia.org/wiki/Hippocampus#/media/File:MRI_Location_Hippocampus_up..png

You might be able to do this through the squamous portion of the temporal bone which is just in front of and above the ear. It’s very thin, and ultrasound probes placed here can ‘see’ blood flowing in arteries in this region. Another way to do it might be a light source placed in the mouth.

The technical aspects of the paper are fascinating and will be described later.

First, what could go wrong?

The work shows that the flickering light activates the scavenger cells of the brain (microglia) and then eat the extracellular plaques. However that may not be a good thing as microglia could attack normal cells. In particular they are important in the remodeling of the dendritic tree (notably dendritic spines) that occurs during experience and learning.

Second, why wouldn’t it work? So much has been spent on trying to remove abeta, that serious doubt exists as to whether excessive extracellular Abeta causes Alzheimer’s and even if it does, would removing it be helpful.

Now for some fascinating detail on the paper (for the cognoscenti)

They used a mouse model of Alzheimer’s disease (the 5XFAD mouse). This poor creature has 3 different mutations associated with Alzheimer’s disease in the amyloid precursor protein (APP) — these are the Swedish (K670B), Florida (I716V) and London (V717I). If that wasn’t enough there are two Alzheimer associated mutations in one of the enzymes that processes the APP into Abeta (M146L, L286V) — using the single letter amino acid code –http://www.biochem.ucl.ac.uk/bsm/dbbrowser/c32/aacode.html.hy1. Then the whole mess is put under control of a promoter particularly active in mice (the Thy1 promoter). This results in high expression of the two mutant proteins.

So the poor mice get lots of senile plaques (particularly in the hippocampus) at an early age.

The first experiment was even more complicated, as a way was found to put channelrhodopsin into a set of hippocampal interneurons (this is optogenetics and hardly simple). Exposing the channel to light causes it to open the membrane to depolarize and the neuron to fire. Then fiberoptics were used to stimulate these neurons at 40 Hertz and the effects on the plaques were noted. Clearly a lot of work and the authors (and grad students) deserve our thanks.

Light at 8 Hertz did nothing to the plaques. I couldn’t find what other stimulation frequencies were used (assuming they were tried).

It would be wonderful if something so simple could help these people.

For other ideas about Alzheimer’s using physics rather than chemistry please see — https://luysii.wordpress.com/2014/11/30/could-alzheimers-disease-be-a-problem-in-physics-rather-than-chemistry/

Baudelaire comes to Chemistry

Could an evil molecule be beautiful? In Les Fleurs du Mal, a collection of poems, Baudelaire argued that there was a certain beauty in evil. Well, if there ever was an evil molecule, it’s the Abeta42 peptide, the main component of the senile plaque of Alzheimer’s disease, a molecule whose effects I spent my entire professional career as a neurologist ineffectually fighting. And yet, in a recent paper on the way it forms the fibrils constituting the plaque I found the structure compellingly beautiful.

The papers are Proc. Natl. Acad. Sci. vol. 113 pp. 9398 – 9400, E4976 – E4984 ’16. People have been working on the structure of the amyloid fibril of Alzheimer’s for decades, consistently stymied by its insolubility. The authors solved it not by Xray crystallography, not by cryoEM, but by solid state NMR. They basically looked at the distance constraints between pairs of isotopically labeled atoms, and built their model that way. Actually they built a bouquet of models using computer aided energy minimization of the peptide backbone. Another independent study produced nearly the same set.

The root mean square deviation of backbone atoms of the 10 lowest energy models of the bouquets in the two studies was small (.89 and .71 Angstroms). Even better the model bouquets of the two papers resemble each other.

There are two chains of Abeta42, EACH shaped like a double horseshoe (similar to the letter S). The two S’s meet around a twofold axis. The interface between the two S’s is form by two noncontiguous areas on each monomer (#15 – #17) and (#34 – #37).

The hydrophilic amino terminal residues (#1 – #14) are poorly ordered, but amino acids #15 – #42 are arranged into 4 short beta strands (I only see 3 obvious ones) that stack up and down the fibril into parallel in register beta-sheets. Each stack of double horseshoes forms a thread and the two threads twist around each other to form a two stranded protofilament.

Glycines allow sharp turns at the corners of the horseshoes. Hydrogen bonds between amides link the two layers of the fibrils. Asparagine side chains form ladders of hydrogen bonds up and down the fibrils. Water isn’t present between the layers because the beta sheets are so close together (counterintuitively this decreases the entropy, because water molecules don’t have to align themselves just so to solvate the side chains).

Each of the horseshoes is stabilized by hydrophobic interactions among the hydrophobic side chains buried in the core. Charged residues are solvent exposed. The interface between the two horsehoes is a hydrophobic interface.

Many of the famlial mutations are on the outer edges of double S structure — they are K16N, A21G, D23N, E22A, E22K, E22G, E22Q.

The surface hydrophobic patch formed by V40 and A42 may explain the greater rate of secondary nucleation by Abeta42 vs. Abeta40.

The cryoEM structures we have of Abeta42 are different showing the phenomenon of amyloid polymorphism.

The PNAS paper used reombinant Abeta and prepared homogenous fibrils by repeated seeding of dissolved Abeta42 with preformed fibrils. The other study used chemically synthesized Abeta and got fibrils without seeding. Details of pH, peptide concentration, salt concentration differed, and yet the results are the same, making both structures more secure.

The new structure doesn’t immediately suggest the toxic mechanism of Abeta.

To indulge in a bit of teleology — the structure is so beautiful and so intricately designed, that the aBeta42 peptide has probably been evolutionarily optimized to perform an (as yet unknown) function in our bodies. Animals lacking Abeta42’s parent (the amyloid precursor protein) don’t form neuromuscular synapses correctly, but they are viable.