Molecular modelling researchers here?

[GHBoners]

Dear Bluelight,

I'm a PhD student, soon doctor if I don't die in the process before the end haha. I would like to know if some people here are working on in silico drug design, and more generally molecular modelling and also machine learning approaches?

I love my job. Molecular modelling is so cool. It's so beautiful to look at the atomic level how your receptor behave. I would like to share this a bit with you bluelighters. Just to start and see how to put images, a little GIF of a zoom into the LSD bound 5HT2B receptor crystallographic structure, at the binding site level. The LSD is in magenta, and amino acids of 5HT2B distant from ~3 A or less are in green. You can see the interaction with dashed lines. You have some hydrophobic ones, some ionic bound (hallmark of aminergic classA GPCRs), some T stacking...
​

 

[Fertile]

Fascinating. I remember the days before Reaxys (Crossfire) when Chemical Abstracts meant that it took 3 days to find a reference. This also explains why the (S) sec-butylamine homologue of LSD has the lower Ki. It also explains why their have been a series of larger mono-substituted analogues turning up that I presume are legal wherever they are being produced. They are branched with an ethyl side-chain and increasingly long straight-chains.

I'm keen to know where the =O of the amide binds. I know it IS important to binding because with 25I-NBOMe and analogues, the O-methoxy on the N-benzyl moiety is a requirement. If you take LSD and 25I-NBOMe, perform energy minimisation and overlay them, the O lone-pair of the LSD amide overlays the O lone-pair of the methoxy precisely.

Oh, it's also worth noting that the AMT skeleton is buried within the LSD scaffold. It's chiral which is why (S) AMT is trippy, (R) AMT is more MDMA like. Of course, the ULTIMATE LSD analogue according to Nichols is 3-methyl LSD. It should bind just as well as LSD, have a slightly improved LogP and significantly increased serotonin release. I mean, TOTALLY impractical to make, but we did both gaze into the distance wishing things were different ;-)

But please, throw up these in-silico models for every class of drug. It will be most instructive. I'm keen to understand why (S) ketamine is an NMDA antagonist.... and (R) ketamine is a potent dopamine reuptake inhibitor (resolved some K and I got to try the 2 isomers - my then wife said 'this is just like coke; after trying the (R) isomer.

 

[paracelsius]

Methamphetamine bound to the Dopamine transporter:

 

[Fertile]

Correct me if I'm wrong, but doesn't methamphetamine and amphetamine bind at quite different sites? It's something I recall from 30+ years ago, so maybe science knows better by now.

It's interesting that methylphenidate has more in common with cocaine than amphetamine.

But of ALL the stimulants, ciclazindol has the most potential for making a MUCH more potent analogue. Of course, I don't know about it's pharmacokinetics. Yeah, it might be strong BUT if it takes 1 hour to begin working and 3 hours to peak, it's not going to be popular. I know some post-grad student wrote his dissertation on stimulants so he explored every class you can think of (apart from aminorex).

 

[ecstacylover]

Correct me if I'm wrong, but doesn't methamphetamine and amphetamine bind at quite different sites? It's something I recall from 30+ years ago, so maybe science knows better by now.

It's my understanding that they're both DAT substrates, in which case it's probably a good assumption that they bind to the same site as dopamine.

Although I wonder if NET activity is being overlooked as a significant contributor to the dopaminergic activity of many stimulants. I suspect this for two reasons..

1. the NET IC50 for many stimulants that are considered primarily dopaminergic is often orders of magnitude higher than the DAT IC50
2. The adrenergic terminals projecting to superficial cortex co-release dopamine and noradrenaline.

In the case of D-amphetamine the hDAT/hNET IC50 ratio is ~14 while for D-methamphetamine it is 16! Let's just imagine, to a first approximation, that these ratios reflect the substrate activity of these compounds at the respective transporters (after all, binding to the transporter is a prerequisite for substrate activity). In this case, it would seem that these drugs should be thought of as primarily adrenergic, at least in low-concentration regimes. Unless of course the NET activity allows them to enter terminals which co-release dopamine and noradrenaline—thereby allowing for dopamine release at drug concentrations below that required for dopamine release from strictly dopaminergic terminals.

 

[GHBoners]

It's so cool to have person passionate like me again these stuffs. I think I will enjoy be there

I'm working in a particular class of GPCRs so what's follow is just an educated guess. But I would say that all compounds binding to the transporters binds in the same binding pocket. Why? Because the DAT/SERT/NET are extremely similar in terms of sequence identity and fold. I retrieved the sequences of these proteins from uniprot database, then aligned them with jalview.

SERT Vs DAT: 47 % sequence identity (that's fucking huge).
DAT Vs NET: 66.5 % (even huger, certainly because of the bigger chemical proximity between DAT and NET ligands).
SERT Vs NET: 48.5 %

A tree based on the calculated sequence identity below:​

Generally, with proteins that similar in terms of sequence (generally, we say that if we have >30 % in seq. id., the fold [aka the way the protein is in 3D] should be similar), we should find exactly the same fold. And yeah, I retrive some AlphaFold model of theses 3 proteins, and aligned them structurally this time. This is what it looks like:​

You have the DAT transporter in green, the NET transporter in blue, and the SERT transporter in pink. You can effectively see a near total overlap between the secondary structures (the alpha helices).

But we are interested in drugs here, right? :D So what I can do next, is to see their difference in the binding site. To do this, I simply used an experimental structure of the Drosophila DAT bound to methamphetamine (certainly the structure that you used paracelcius, pdb 4XP6). I aligned it structurally to our human transporters, and used the methamphetamine molecule to delimit the binding site. Without much surprise, the Drosophila DAT structure share the same fold with its human counterpart (evolution is lasy as fuck haha). In the following picture, you can the human DAT in green, and the drosophila DAT in darkblue, with the methamphetamine little molecule in magenta. This is where the binding site is, certainly. Note also that I changed a bit the direction of looking at the system compared to the previous one, so we can see more the amphetamine.​

Then I propose to zoom and look into the cavity indicated by the methamphetamine. So you still have the human DAT in green, the drosophila DAT in dark blue. I hided the secondary structure and selected only amino acids that where at 4 A or less of the methamphetamine. And how suprise, we can see that most of amino acids between the two proteins are the same. Not perfectly, the binding affinity between Drosophila and human DAT will be different, but we find a lot of overlap there.​

Then what about the difference between human DAT and human SERT and NET?​

So here we can see that DAT, NET and SERT share features, quite like aminergic GPCRs. You can find in each of them an charged - Asp, certainly here to interact with the charged + amine of methamphetamine but all other aminergic drugs. You can see that the rest of the binding site is quite hydrophobic with multiple aromatic circles. I've put an # for each amino-acid that were the same for the 3 transporters. But the binding site are not totally identical. Theses subtles changes are what makes SERT bind more serotoninergic drugs and so on.

Hope you will find this interesting!



​

 

[ecstacylover]

Nice post @GHBoners , do you do much work with MD simulations?

Also, just to expand on what you said, RMSD of hSERT from Drosophila DAT is only 1 A which is about the length of a hydrogen atom. Despite this there are slight functional differences. SERT is actually electroneutral due to co-transporting one K+ ion (in DAT and NET the K+ can transiently bind but is not transported), in addition to a Cl-, an Na+, and the cationic monoamine. This also means that the 5-HT transcellular gradient can be shifted by playing with K+ conductances, while dopamine and norepinephrine transcellular gradients will be insensitive to K+ conductances.

It's my belief that monoamine releasers accomplish "reverse transport" primarily by their modulation of ionic gradients, and that any structural modifications of the transporter are only secondary in importance. These ionic gradients normally provide the free energy for intracellular accumulation of monoamines, and altering them can shift the equilibrium thereby increasing extracellular concentration of monoamines. So while dopamine and norepinephrine releasers might accomplish this by modulating Na+ or Cl- conductances (in addition to VMAT), 5-HT releasers have the additional possibility of modulating K+ conductances.

 

[thegreenhand]

You have the DAT transporter in green, the NET transporter in blue, and the SERT transporter in pink. You can effectively see a near total overlap between the secondary structures (the alpha helices).

that is quite interesting.

i'd never realized how structurally (and thus perhaps functionally) similar they are

 

[GHBoners]

Nice post @GHBoners , do you do much work with MD simulations?

Yeah this is my favorite technique. For those who don't know, MD mean molecular dynamic simulations (I'm doing the "classical", Newtonian based calculations, not the quantum based when you take into accounts the electron clouds that are faaaaaaar more computationaly expensive). When we have a model of our protein like depicted before, we can add whatever ligand we are interested in the system, add a phospholipid membrane if the protein is a membrane receptor, to mimic the milieu. Then we had water molecules to form a box around the system. And at this point, we can calculate the dynamical evolution of the system, integrating the second law of Newton. The beauty of it is that we are making such a huge approximation! It's like we are considering atom like planets, but we still succeed to makes meaningful predictions with this. That's really cool, I will show you some videos whenever I find time for it.

And then from these simulations, you can retrieve multiple information at the atomistic level, like binding free energy, the important movements of an ion channel or an activated receptor, identify a binding site and the path that takes a ligand into its binding site, try to understand the impact of a polymorphism on the properties of the protein, etc etc... Relative to your system, this is quite calculation expensive. We are using server of GPU cards. At this time in the field, and in our lab, we can simulate from nanoseconds to a few hundred of microseconds, not much more (and it takes months), on system of 100000 atoms so around a GPCR + its membrane + the water box around.

Also, just to expand on what you said, RMSD of hSERT from Drosophila DAT is only 1 A which is about the length of a hydrogen atom. Despite this there are slight functional differences. SERT is actually electroneutral due to co-transporting one K+ ion (in DAT and NET the K+ can transiently bind but is not transported), in addition to a Cl-, an Na+, and the cationic monoamine. This also means that the 5-HT transcellular gradient can be shifted by playing with K+ conductances, while dopamine and norepinephrine transcellular gradients will be insensitive to K+ conductances.

It's my belief that monoamine releasers accomplish "reverse transport" primarily by their modulation of ionic gradients, and that any structural modifications of the transporter are only secondary in importance. These ionic gradients normally provide the free energy for intracellular accumulation of monoamines, and altering them can shift the equilibrium thereby increasing extracellular concentration of monoamines. So while dopamine and norepinephrine releasers might accomplish this by modulating Na+ or Cl- conductances (in addition to VMAT), 5-HT releasers have the additional possibility of modulating K+ conductances.

Quite interesting indeed, thanks for these precisions! What I would love to see in MD is the actual transport of a neurotransmitter accross the transporter with its ion. It may have already be done I dunno. I would also love to see where the antidepressive drugs bind to hinder the SERT channel, for example. Is it an allosteric site, or do they fix on the site that we see in the few previous pics?

 

[paracelsius]

Correct me if I'm wrong, but doesn't methamphetamine and amphetamine bind at quite different sites? It's something I recall from 30+ years ago, so maybe science knows better by now.....
But of ALL the stimulants, ciclazindol has the most potential for making a MUCH more potent analogue...

They both bind at the same site on DAT/NET/SERT. Only difference is their selectivity: (+)Amph is more potent at DAT~NET than (+)Meth but way less active at SERT. Actually even cocaine-like pure reuptake inhibitors (MPHs, pyros, desoxy,..etc) bind same site. Highly potent DNRIs like mazindol, ciclazindol (and pipradol, d2pm, dpmp..etc) all bind same site.
But those are too lipophilic to be recreational. High logP gives them long time to reach maximum plasma concentration Tmax (at least orally). Which actually is dangerous with highly potent compounds: Isn’t it what sent ppl to psych ward back with full blown psychosis back in desoxy (aka Ivory wave) 10years ago?? because it takes hour(s) to kick in, ppl thinking they haven’t taken enough redose and redose only to regret later! I mean desoxy is active starting at 0.5mg! Plus their super long half-life (36h for ciclazindol, up to 3 days for dpmp) doesn’t help either.

....Although I wonder if NET activity is being overlooked as a significant contributor to the dopaminergic activity of many stimulants. I suspect this for two reasons..

1. the NET IC50 for many stimulants that are considered primarily dopaminergic is often orders of magnitude higher than the DAT IC50
2. The adrenergic terminals projecting to superficial cortex co-release dopamine and noradrenaline.....

You right: pure dopaminergics are actually not really recreational. Needs norepinephrine. Actually I’ve seen a paper where they claim Norepinephrine release (not DA) is responsible for "subjectives effects of stimulants" (academic term for stim rush-euphoria”). But it is little complicated. Pure NRI/A are not recreational either. Example: (-)Meth is potent selective norepinephrine releaser. But it is not really recreational. It is the shitty isomer of racemic meth; only increase peripheral effects (tachychardia, Bp, vasoconstriction, stimdick..etc) without psychoactivity. So it is not NE or DA alone responsible for stim rush. Need both. Ppl makes mistake when they talking about “Dopamine rush”. It is more like “Dopamine-Norepinephrine rush”. Both are required. The key is balance ratio of DAT-NET-SERT.

....Drosophila DAT bound to methamphetamine (certainly the structure that you used paracelcius, pdb 4XP6)..... I would also love to see where the antidepressive drugs bind to hinder the SERT channel, for example. Is it an allosteric site, or do they fix on the site that we see in the few previous pics?....

Yes that is PDB 4px6 crystal. As you said, yeah those 3 transporters are structurally very much alike. They pretty well conserved especially at the ligand binding site. Makes sense since the natural ligands DA, NE and SER molecules are very similar. Guess they came from same ancestors: I think even insects have similar transporter proteins tho they use octopamine instead of DA/NE (octopamine is Norepinephrine without the meta-hydroxy on the phenyl).

The only difference between DAT, NET and SERT is the hydrophobic pocket you mentioned (where the aryl fits). It is pretty similar for DAT and NET and larger for SERT. As a matter of fact, in some part of the brain Dopamine and Norepinephrine are used interchangeably. In the prefrontal cortex, Dopamine uses Norepinephrine transporter NET while in the limbic area (the part of the brain that gives feel good rush-euphoria) Norepinephrine uses DAT.

Here is the SSRI antidepressant Sertraline in complex with Serotonin Transporter (PDB 6awo). In this specific case, it is not allosteric but ligand binding site. Notice how similar to DAT

Good job you did with that blast alignment. Very instructive.

 

[ecstacylover]

You right: pure dopaminergics are actually not really recreational. Needs norepinephrine. Actually I’ve seen a paper where they claim Norepinephrine release (not DA) is responsible for "subjectives effects of stimulants" (academic term for stim rush-euphoria”). But it is little complicated. Pure NRI/A are not recreational either. Example: (-)Meth is potent selective norepinephrine releaser. But it is not really recreational. It is the shitty isomer of racemic meth; only increase peripheral effects (tachychardia, Bp, vasoconstriction, stimdick..etc) without psychoactivity. So it is not NE or DA alone responsible for stim rush. Need both. Ppl makes mistake when they talking about “Dopamine rush”. It is more like “Dopamine-Norepinephrine rush”. Both are required. The key is balance ratio of DAT-NET-SERT.

I agree with you there, although my point was more so that the high-affinity NET activity might allows them to produce DA release at lower concentrations than would be expected from just looking at their DAT affinities.

 

[Fertile]

I've been fascinated by 2 compounds in particular:

(±)-McN 5652 & JNJ-7925476.

US Patent 6,162,417 is the index patent (all relevant patents and patents reference back to this).

People will notice that they are refinements to the nomifensine series (for want of a better term) which are themselves derived from the ampetamines by way of the diphenylethylamines (such as pipradrol).

I warn you in advance that synthesis is torturous. I know Dr. David Nichols is aware of them and I wouldn't be surprised if he has suggested research on them as a subject for his post-graduate students, but AFAIK none of them have been confident enough (or possibly stupid enough) to consider making more analogues.

Ring-substitution does seem to follow that of the amphetamines. Of course, the LogP means that a sublingual formulation (for example) would be needed to ensure a reasonably fast onset time. I would also consider a sacrificial moiety that is added to allow the body a reasonably efficient metabolic pathway.

Things like mephedrone are clever in that the p-Me balances it's NED/DAT/SERT activity AND provides a really good target for non-specific blood enzymes.

BTW it's excellent to find a post with really educational content. I'm 30 years out of the loop and so I'm having to read up on stuff that I forgot, that wasn't known back then or I missed at the time. Thank you to everyone.​

 

[polymath]

I would like to see a simulation result of how well the N-demethylated versions of cyclizine and diphenhydramine bind to the dopamine transporter and other monoamine transporters. Those would be expected to inhibit dopamine reuptake, but without the antihistamine and anticholinergic effects of the original compounds. But this doesn't tell much about the subjective effects because many substances inhibit dopamine reuptake without having much effect on mood states.

The chemical 6a in this publication, tested for DAT binding, is like a diphenhydramine with both n-methyl groups removed and with the ether oxygen replaced by a sulfur atom.

 

[paracelsius]

I've been fascinated by 2 compounds in particular:

(±)-McN 5652 & JNJ-7925476....​

I warn you in advance that synthesis is torturous.​

Very fascinating indeed. Yeah I think those McNeil pyrroloisoquinoline compounds have been overlooked. As a class, quite easily beat any Stims RCs like PVs..etc. It is not really that hard synthesis especially leading to cis-trans mix. Sure only the cis isomers are active. but who cares! some of them are so potent like McN-5908 why bother making chiral cpds. I mean this last cpd McN-5908 is about 100-200x cocaine as DNRI (IC50 0.88nM DAT, 0.20nM NET, 15nM SERT). Racemates will be as good.

One of the most potent and interesting DNRIs I ever came across!! Near perfect drug-like properties (logP, PSA, Lipinski..etc) + they're certainly unlike any banned chemical anywhere (except the UK! which bans anything that affect anything in the brain..kind of ridiculous tho! but the law is the law).

wonder why Maryanov team at McNeil didnt develop them further. I guess they were looking for SSRIs, the craze at that time and got beaten by Lilly Prozac to the market. but who knows. Could be possibility of abuse liability issues too.

 

[Fertile]

If you look at the synthesis, I think you will see that they must have had about a dozen post-grad students doing the chemistry! I still think you would need a sacrificial moiety. Maybe borrowed from nomifensine? I'm pretty sure that the aromatic nitro was added to reduce duration. If I remember the report, it's duration was 6-8 hours. I think that of dichlorofensine is a lot longer.

 

[paracelsius]

Certainly to develop the chemistry initially especially the stereospecific synthesis. + they made literally hundreds of analogs. But for the racemates it is actually only 2-steps from commercially available 2-phenylpyrrolidine and styrene oxide.. 4 if you add (straightforward) nitration-reduction to get the Nitro and then the 4-amino analog 5908. 10.1021/jm00391a028
The nitro was relatively inactive tho isnt it?. maybe used as prodrug. I suspect the amino would have reasonable T1/2 certainly less than unsubstituted or chloro, dichloro..etc. since it has a polar amino metabolic handle.
edit: synthesis talk is now allowed so @MOD remove post if it cross line. Oh btw even the racemates is enriched with the cis (active) enantiomer:
"....afford a mixture of amine products, in which the cis isomer generally predominates by a 3:l ratio" same ref

 

[Fertile]

Nice paper there.; very nice. Finding an appropriate optically active salt to resolve the isomers is a fascinating area of chemistry.. My first idea would be to check the activity of the trans pair. If it's not going to interfere (too much) with the activity of the active, leave it. Especially if you have a 2 step synthesis!
Of course, sertraline and related are always resolved. I've often wondered what the problem was with the mixture.

 

[paracelsius]

PyroTHIQ

Finally have some time to go through that paper. Pretty extensive SAR I’d say. Old fashion medchem (one compound at a time). Now I realize some of these compounds are even more potent than I thought, pretty insanely potent stimulants! I mean the parent compound (7b) is ~40xcocaine at DAT, 1300x at NET and 14x at SERT. The one I mentioned 52b is even more potent (500, 4000 and 7x). Pure stimulants NDRI selective. I’d stay away from those tho for sake of harm reduction. Potency is insane! Too much risk of psychosis+too adrenergic.

But I think the best with more potential in the series is 8b (3’,4’-dimethoxy-pyro-THIQ). I mean in terms of recreational (functional) potential. This one has more reasonable and optimal PK-ADMET profile. I’d expect it to be very much like MDPV and similar pyros. Not that MDPV is not potent (It is still what?like 20xcocaine? as DNRI).

MDPV: DAT~30nM, NET~40nM and SERT~9300nM
8b: DAT 15nM, NET 50nM and SERT~1500nM
Properties (logP, logS, PSA, BBB....etc) are pretty similar.

A little more dopaminergic selectivity which is good as It would probably be safer than MDPV (less peripheral adrenergic side-effects (cardiovascular, bp..etc).

Oh yeah the trans epimer is the active form. Shouldn’t be problem separating from the cis tho. Just extra work and lower yields but who cares synth is only 2-steps. Actually one pot when you think about it. Not quite “shake’nbake” kitchen chemistry but straightforward really.

Really fascinating class of stims/antidepressants: Thought about those (diclofensine-type) years ago and dismissed at first glance thinking too lipophilic to be usefull. But apparently I was mistaken.

 

[Fertile]

I'm glad your outcomes (of your research) were much the same as mine. It wasn't finding the MOST potent but rather the LEAST potent.

Now I recall an overheard discussion at the 2004 GRC conference. I was amazed how (from only one conference before)m) the optical isomrs suffenly became the KEY. I've mentioned that their are sertraline homologues which I guess would be termed norsertraline (i.e. methylamine ---> amine) as it was well established that the primary amines are the active drugs (but of course, in the 10 years, much research may say othersies( but some Ukrainian friends had made the primary amines (also modified to reduce duration. It was most clearly 'off the record' but they were clearly very ling-acting and very potent,

I very much doubt it would have been an RC but I keep reading about the Chinese Army using medicines that resulted in 48-60 hour shifts in which they were not relieved. So on one-hand they expect the most modern weapons.... and then need drugs to allow such flights could go on week in, week out; month in, month out.

Well - it would be done, if he is OK.

 

[Fertile]

classes hosted at ImgBB

Image classes hosted in ImgBB

ibb.co

There has been much written on the aromatic amine, a moiety often resulting in toxicity in the compound. Nitrobenzodiazepines are a good example. In Sweden, 70% of 'non-violent' suicides are caused by nitrazepam, clonazepam, flunitrazepam and nimetazepam. People have survived multiple grams of diazepam (certainly 7 oz) and survived.

I was unable to divine the toxic mechanism but it IS an accepted fact.

So the aromatic -NH2 here needs close examination. I believe that it increases serotonin activity and acts as a sacrificial moety - an easy way for the body to excrete the drug.

One of the reasons synthetic alcohol was abandonwd was because pynazolam (one of the 2 actives) was sufficiently toxic for us not to proceed with pyezolam (the 1,5 benodiazepine) because not only is it needed to emulate low-dose alcohol, but it's still needed to emulate high-dose ethanol and too many drinkers just keep going until they die.

We did not feel that an RC that killed (albeit in overdose situations) was not an acceptable option. For those of you interested, A 3:1 ratio of pyzolam : pyeyzolam emulates ethanol. Or, rather, it's positive effects. They are both slightly soluble in water and so 3g of pyezolam & 1g of pyeyzolam really emulated wine very well. The flavour wasn't awful and easily hidden using fruit flavours. We produced 8 x 70mL bottles of it for people to try....

It was all going well until conflict in Ukraine meant we couldn't work with our research lab.

I can tell you all this because testing 2 novel compounds is VERY costly and seemingly nobody was prepared to invest. But thanks to some work carried out by Professor James Cook, we were able to produce both effects into a single compound. This cut development costs by 2/3 because instead of trials needing to be carried out on each compound in turn and then the mixture, A single compound could be tested.

People who tried pyeyzolam will testify that it IS like being 'drunk' but what about people who just want a couple of beers? Well, we sorted that out as well.

I'm certainly not suggesting that their are only 2 (combined into 1) compound that can simulate alohol's effects, in fact their are 6 compounds known and so their are actually 36 compounds. If I am honest, we ignored one series because we weren't happy about toxicity. Another we were concerned about toxicity. I mean, just because its LESS toxic than ethanol doesn't mean it isn't toxic. We also felt the need to provide a compound that would reverse the effects.

But what I learnt is that patenting a new (series of) medicines takes a decade and is frought with worried. We realised that working WITH Dr. James Cook was the way forward.

But forget the money- although we would ensure it ONLY ended up in drinks, it would be possible to 'spike' people with it - a concern.

But believe me when I say that their is one GABA subtype without a known ligand and we would like to research it. During the search for an alcohol mimic, we discovered MANY compounds that produced unusual ASCs,