Rapid Discovery of Potent Inhibitors of the Main Protease of SARS-CoV-2

November 02, 2022

Information

Yale Cancer Center Grand Rounds | November 1, 2022
Presentation by: Dr. William Jorgensen

ID8224

To CiteDCA Citation Guide

00:00On behalf of my Co leader,
00:02Barbara Burtness and I,
00:03I'm pleased to introduce Bill Jorgensen,
00:06one of our developmental therapeutic program
00:08members and also my long term collaborator.
00:11Bill is a graduate of Princeton and Harvard,
00:14spent 15 years on the faculty at Purdue,
00:17and in 1990 he moved to Yale,
00:19where he's currently Sterling
00:21Professor in the Chemistry department.
00:24Bill is internationally recognized
00:25as one of the world leaders in
00:28computational chemistry and drug design.
00:30His research has been recognized
00:32by many honors,
00:34and among among those include the American
00:36Chemical Society Cope Scholar Award,
00:39the ACS Award for computers.
00:42Chemical and pharmaceutical research,
00:44the ACS Hildebrand Award,
00:47the ISTQB P award in computational biology,
00:51the Sato International Award from
00:54the Pharmaceutical Society of Japan.
00:56He's been elected to membership
00:58in the International Academy
01:00of Quantum Molecular Science,
01:02American Academy of Arts and Sciences,
01:04and the US National Academy of Sciences.
01:07Another recent honor in
01:092020 includes one of the 16.
01:11Researchers selected for a Nobel Laureate
01:15Citation for individuals considered
01:18doing Nobel Nobel Class Research
01:20that has been cited over 2000 times.
01:23Today he's going to tell you a little bit
01:26about some of his work on SARS COVID 2.
01:29So without further ado,
01:31Bill,
01:31take it away.
01:33Yeah. Well, thank you very much
01:35Karen and the pleasure to be here
01:37the other side of the campus.
01:39So the our work I'll tell you about today
01:42is totally a collaboration between my
01:45research group and chemistry and Karen's
01:47group over here in the the Med school.
01:50So I'll talk a little bit in general about
01:53computer aided drug discovery and then
01:56specifically about our work and finding
01:59very potent protease inhibitors for SARS.
02:03Move two. So a key element of drug design
02:08is the fact of trying to make inhibitors
02:12that bind to an enzyme typically.
02:15So and we'll be talking about again a small
02:19molecule binding to SARS Cove 2 protease.
02:22And so this is governed by an equilibrium
02:24where you have the protein and water,
02:26the inhibitor and water,
02:28there's a free energy of
02:31binding and then the complex.
02:33So the free energy of binding the G because
02:36we're working in the constant pressure,
02:39constant temperature world is for Gibbs.
02:42So it's a Gibbs free energy and I put the
02:46stamp of our former colleague Jay Willard.
02:50Gibbs here is the father of.
02:53Thermodynamics.
02:53So the free energy binding just to introduce
02:58the concept of a nanomolar inhibitor.
03:01So the free energy of binding is given
03:04by minus RTL and the dissociation
03:06constant if you have a dissociation
03:09constant of 10 to the minus nine molar.
03:12That would correspond to A1 nanomolar.
03:16Inhibitor or an inhibitor that has that KD,
03:22one that has a KD of 10 to the
03:24minus six would be a micromolar
03:26inhibitor and our binder.
03:28And the reason I bring this up
03:32is that most drugs turn out to be
03:36typically one to let's say 20 or
03:39so nanomolar in a binding assay.
03:42And this all ultimately has to do with
03:45the farm human pharmacology and just
03:47how big a pill one is willing to take.
03:50So this obsession with nanomolar
03:53inhibitors just to, you know,
03:56reflects this fact,
03:58so.
03:58Ultimately here we're going to have to
04:01do simulations on computer simulations
04:04of proteins binding to ligands in water.
04:07And so how did this arise?
04:09When did it with such things happen?
04:12And the answer is,
04:13there really wasn't any significant
04:16work on doing computer simulations of
04:18molecular fluids before the late 1970s.
04:21And then of course it grew slowly after that.
04:25The problem is you have a lot of particles,
04:27you're using classical force.
04:29Those describe the interactions,
04:31but there's still a lot of particles
04:34and you have to.
04:35Observe the system over a significant
04:38time period.
04:39So if you're doing molecular dynamics,
04:41we might want to run the molecular dynamics
04:44for picosecond hundreds of picoseconds,
04:47nanoseconds and this just we didn't
04:50have the computer resources to do that.
04:53And then making it more complicated
04:56by putting a protein into it and
04:59describing the energetics of the
05:01protein and the water.
05:03That really didn't happen until mid.
05:061980s and my colleague here,
05:08Julian Torrado Rivas and I
05:10published one of the first
05:13calculations for a protein in water
05:16where we did molecular dynamics for
05:19100 picoseconds and that was in 1988.
05:22So doing the type of calculations
05:25we're talking about today is
05:28relatively recent phenomenon.
05:30This is a picture we'll talk
05:32about HIV reverse transcriptase
05:33and just to get the sense,
05:35I usually give this to less.
05:36Sophisticated audiences
05:37to point out the yellow,
05:39little yellow pieces inhibitor
05:41and that's enough to shut down
05:44this enzyme and this is an example
05:47of one of the compounds that
05:49developed through Karens and in
05:52our work that is a inhibitor of
05:55HIV RT that little molecule.
05:57So here's the way we do it.
06:00We normally start with an X-ray
06:02structure and the first phase of
06:05this we're looking for micromolar
06:07hit compounds that then we have to
06:09do a lot of hard work on to bring
06:12them to the low nanomolar level.
06:15So we normally start with an X-ray
06:17structure and this can be from
06:19you know somebody else's work and
06:20we remove the ligand that might
06:23be in that X-ray structure and
06:25then we try to design our new
06:27our own inhibitors and that this.
06:30Started out we do a virtual
06:33screening which is docking.
06:36And I'll tell you a little
06:37bit more about that,
06:38where we literally fly molecules
06:40into the protein structure and
06:42see which ones look the best.
06:45Or are we do denovo design where we
06:47use a growing program that I wrote
06:50a while back that starts with the
06:53little seed core of a molecule of,
06:56say benzene you place in the binding site.
06:59And then the program will build libraries
07:02of compounds starting from that core,
07:04growing them out in the binding site.
07:06And then you have to.
07:08A score of them,
07:09evaluate them in the same manner
07:12that this invariably gives us
07:14these micromolar hit compounds.
07:16But we've never been fortunate enough
07:18to do this initial part of the work
07:21and end up with nanomolar inhibitors.
07:23We're close, you know,
07:26single digit micromolar.
07:27So then the hard part is the lead
07:30optimization because we're going to
07:32have to refine the micromolar hits by
07:35making small changes that we decide.
07:38What to do?
07:39By a lot of structure building
07:41and energy minimizations.
07:43So this bond program of mining can
07:45rapidly build protein ligand complexes.
07:48We can energy minimize them.
07:50That's just a fast calculation
07:52compared to adding the water
07:54doing the molecular anamax.
07:56And so we do a lot of the structure building,
07:58energy minimization and then for
08:02select cases we will do, excuse me,
08:07the free energy calculations
08:09that are sort of our hallmark.
08:12We call them FEP,
08:14free energy perturbation calculations.
08:16Virtually all pharmaceutical
08:17companies today are,
08:19you know,
08:20jumped on this.
08:21Everybody's doing FP calculations
08:24for a drug design.
08:26So then you have to make a decision on
08:29what molecules to synthesize their assay.
08:32So you need somebody like Karen to help out
08:35in the assaying and the crystallography,
08:39the crystallography isn't.
08:41Critical,
08:42but it sure is helpful if you know very
08:45much helps reinforce what
08:47the modeling is doing.
08:49And also sometimes you'll see that
08:51the there the there's a change in
08:53the protein structure from what you
08:55originally started with that you
08:57will see in the crystallography.
08:59You don't necessarily see
09:00it in the computation.
09:01So the crystallography is really helpful.
09:06When the HIV area,
09:08Karen and I got along for quite a few
09:11years without a crystal structures,
09:13but then once we certainly
09:15current lab start getting them,
09:16it certainly made life a lot more confident.
09:20So you repeat the cycle until
09:23you get the potency you want.
09:26All the while we are mindful of properties,
09:29so we want the compounds to be drug like.
09:33And that requires a having things
09:36like reasonable solubility,
09:37reasonable cell permeability,
09:39no reactive functional groups.
09:41So we have software that checks that
09:44and then we also do some measurements
09:47of solubility and cell permeability,
09:50OK.
09:50So the FP calculations are done for
09:53where you do molecular dynamics or Monte
09:57Carlo simulations for protein ligand
10:00and a typically a ball of several 1000.
10:04Water molecules.
10:05And you do a calculation where
10:08you're comparing the green ligand,
10:11green inhibitor with the blue.
10:13So you do calculation.
10:15We have protein green legging to give
10:17complex protein blue ligand to give complex.
10:20And what we do on the computer,
10:21it turns out to be easier is to mutate
10:25the green leg into the blue unbound
10:27in water and then bound protein.
10:30And the difference in the two
10:33vertical numbers there then gives us
10:35the difference in predicted free.
10:37Energy binding.
10:39And so this type of calculation wasn't
10:42done at all before 1985 or just the
10:46simple green to blue in water FP calculation.
10:50That was something that I.
10:53It will take credit for doing the
10:56first calculation of that type again.
10:58Then there's no software.
10:59You had to write all the software,
11:01you know the force fields we
11:03had to develop etcetera.
11:04So it was very different world in 1985, OK.
11:08So here are just a little bit on HIV.
11:11HIV is still a big problem.
11:14Some of the statistics
11:16are shown there for 2021.
11:18They're,
11:19you know,
11:19close to 40 million people in the world
11:22that are infected with HIV and about 1
11:26to 2,000,000 each year are becoming infected.
11:30And they're on the order of 650,000 deaths.
11:34So that's down quite a bit from what it was.
11:38But still,
11:39you know,
11:40from a very serious problem and
11:43a Long story short.
11:45We have worked on with Karen on
11:48the reverse transcriptase and the
11:50so this is a an RNA virus and it
11:54has a reverse transcriptase which
11:58converts the RNA to DNA which
12:01is incorporated into the host
12:03cells a genome by HIV integrase.
12:06So HIV reverse transcriptase has
12:08been the principal target for anti
12:11HIV drugs and there are two classes,
12:13the nucleosides.
12:14And the non nuclear science,
12:17Karen has worked on both.
12:19So in our collaboration with Karen
12:21we've only worked on non nuclear sites,
12:23the NRT I and there are allosteric
12:26inhibitors. They bind in this little
12:29pocket that is about 10 angstroms or
12:32so from the polymerase active site.
12:35It's one of the few examples of of allosteric
12:39inhibitor that's that have become drugs.
12:41It's very, very, very very I'm, I'm, I'm.
12:44Have to think of it to find others.
12:47This is the principal example.
12:49The crystal structure.
12:51Again a Yale connection.
12:53The original crystal structure of HIV
12:55RT was done in the sites lab 1992.
12:58This is a very big. You know,
13:02discovery at the time because the HIV
13:05crisis was so severe and it's a big protein,
13:08thousand of residues.
13:10So Long story short,
13:12we've tried to make better
13:15non nucleoside inhibitors.
13:16The original ones have limitations.
13:19They're susceptible to mutations
13:21that arise quickly.
13:22They also had some undesirable pharmacology.
13:26So the way we proceed on
13:29HIV is the same with the.
13:31COVID and the trick in lead optimization is
13:37making systematic changes small changes in.
13:42Substituents on rings,
13:44the rings themselves and groups
13:46that link rings together,
13:49and if you know the right changes to make,
13:52they can have profound effects.
13:54So this is an early HIV compound
13:58of of ours that we came about from
14:02a de Novo design and Karen's lab.
14:05The assay they're running is an infected T
14:09cell assay and this compound had an EC50.
14:12For inhibiting the reproduction
14:15of of of the HIV in the infected
14:19cells of 10 micromolar,
14:21so 10,000 nanomolar.
14:23So that's a reasonable starting place,
14:25a small molecule, but we've got to
14:28increase the potency by a thousandfold.
14:31So I I point out here that if you
14:34happen to know to put a cyano group
14:36in the four position of this ring,
14:38you get a very big boost,
14:4050 fold boost to 200 nanomolar.
14:44OK.
14:45Then if you happen to change
14:48the thiazole into a pyrimidine,
14:50you get another tenfold boost and
14:53you're at 17 nanul. So quite amazing.
14:56And then if you happen to know to
14:59put a methoxy group and the the
15:01three position of the pyrimidine
15:03ring here 2 nanomolar.
15:05So you have more potency than
15:06you need for a drug.
15:08So this is all fine and this is
15:10what we use the FEP calculations to
15:13help us with because these changes
15:15are in a sea of possible changes.
15:17So we do however scans where we
15:20have we have a compound like this,
15:22we'll scan in chlorine atoms at each open.
15:26To see if we can add a little
15:29beef to it and that might have,
15:32if we did that it would show that this
15:35four position is good for chlorine,
15:37well if it's good for chlorine
15:39and may also be good or even
15:41better for cyano because they're
15:42both somewhat electronic drawing.
15:44So then we would try siana.
15:46But we do these initial scans,
15:48we also do heterocycle scans of
15:51five and six membered rings because
15:53they are obviously affect hydrogen
15:55bonding patterns and hopefully
15:57that would have picked up that
15:59the pyrimidine was the
16:00way to go. And then finally we do
16:03another substituent scan on the
16:05pyrimidine of methyls and chlorines,
16:08we would see that substitution and the
16:10three position is a good thing and
16:13before long we would come to the methoxy.
16:15So that's the way it's done.
16:17And that's attuned animal or very potent
16:19compound we did in collaboration with
16:22Eddie Arnold got a crystal structure
16:24of that there's quite a quite a bit
16:27later and that was the only crystal
16:29structure we had until Karen's group
16:31started getting some around 2012.
16:34OK, so here is just some of the
16:37work with Karen.
16:39These are all publications on different.
16:43And an RTI's and you might say well
16:47Gee and from 2006 you have these two
16:50national or compound aren't you done,
16:51why are you why are you keeping doing
16:54this and the answer is that that
16:56number is against the wild type virus.
16:59But the virus as you know have
17:01mutates just like COVID is mutating
17:04and there's a whole panel of mutants
17:06with the HIV and you need to have
17:09efficacy against all of the common
17:12mutants with one compound.
17:14So it's tough.
17:15So that initial compound like initial.
17:20Compounds in this class,
17:22such as nevirapine,
17:23was the first approved drug in
17:25this class, like nevirapine.
17:27It was good against wild
17:29type but not not much else.
17:32So these other compounds that
17:34I'll just skip to this one,
17:36one of our better compounds,
17:38we've increased the potency,
17:39but we very much increased the
17:42performance again mutant panels,
17:44so very difficult mutant
17:45is a double mutant K10 3N.
17:50Y181C and this compound
17:52here is A10 animal or EC50,
17:56which is you know good and great
17:58against that whereas the original
18:00compounds here would have had
18:02no efficacy against that mutant.
18:05And we've gone on,
18:07we even see something that looks like
18:08a covalent inhibitor which it is.
18:10We with cooperation with Karen,
18:13we have covalent inhibitors for HIV RT Wild.
18:19Type and also the Y181C mutant.
18:23But I will go on now to what
18:26we did with COVID.
18:29So fortunately, because of our work on HIV.
18:33We're pretty well positioned just to
18:36try to do something when COVID rolled
18:39around at the beginning of 2020.
18:42So this is the IT again an RNA genome.
18:47And it some of the proteins that
18:50it encodes are indicated here,
18:52and not as many as with the HIV,
18:56but you do have.
18:59The The There's a proteases here
19:01that are sort of papain like protease
19:03and then the main protease and what
19:06we've worked on is the main protease.
19:09There's also,
19:09you've probably heard of the
19:11RNA dependent RNA polymerase.
19:12This is just to reproduce the RNA genome.
19:16That's another possible target,
19:18and some of the structural
19:20proteins are over here.
19:21There's the spike and the famous
19:24spike that is mutating and causing
19:26a lot of problems for the vaccines.
19:31So the cycle, the life cycle involves
19:36the COVID virus binding to the ACE 2
19:41receptors on the cells endocytosis.
19:45The RNA genome is unprocessed by a host of
19:49ribosomes to give you these two polyproteins.
19:53You similar situation with HIV,
19:57generating polyproteins that
19:58have to be cleaved by HIV. Areas.
20:02So here's where if we can stop this
20:05proteolysis step, the rest of the
20:08reproduction cycle stops and it's,
20:11I could say there aren't as many
20:14targets here as with the HIV.
20:16There's no integrase,
20:18no reverse transcriptase.
20:19And So what we picked in the beginning
20:22of 2020 that we would work on the
20:26protease almost because there's
20:27hardly anything else to work on
20:29and there was a crystal structure.
20:31Reported so the first thing we did.
20:36So this came about as as you recall,
20:40things got serious in late January
20:422020 and then in March 2020 is
20:45one thing shut down. So we were.
20:50Sent out of the lab.
20:51You know, we could work from home.
20:53If you had special permission,
20:54you could work in the lab.
20:56But we didn't pursue that.
20:58But we decided for working at
21:00home that what we could do is we
21:02would do docking because we have
21:04the crystal structure,
21:06a crystal structure of the protease.
21:08So we would do docking.
21:10And the typical way docking works
21:11is you have the crystal structure
21:13and you have a library of compounds
21:15and these are typically commercially
21:18available compounds.
21:19There's a.
21:20Famous library called Zinc that
21:22has up to 100 million compounds
21:25and then the computers software
21:28combines them and it makes the
21:31complexes and then it has to score
21:33the complexes which is the weak spot.
21:37Often the scoring isn't very
21:39accurate but you can then test
21:41the high scoring molecules.
21:42Well that's a lot of compounds to deal with
21:46so I thought well we would do 1st instead.
21:50Is to dock known drugs,
21:52approved FDA approved drugs.
21:53So I happen to keep a library of
21:56these in the computer and there are
21:59three-dimensional structures of the drugs,
22:00which is this is all three-dimensional.
22:03And so I asked Muhammad and Julian.
22:07To dock the 2000 known drugs to see if
22:12we could see get some reasonable hits
22:16from that and So what happened was.
22:21The docking was done in a consensus fashion,
22:25meaning they used four different
22:27docking protocols,
22:273 different programs and four ways
22:30of doing the docking because any
22:32one program we don't fully trust.
22:35So we're hoping that there will be a
22:38consensus where you score well in all four.
22:42Protocols. And so we got the list.
22:46Excuse me. I don't have code and I've tested.
22:53But.
22:54We we got the list of the top
22:57compounds and then, very importantly,
22:59we visualize the predicted poses,
23:02the complexes.
23:05And based on that visualization,
23:07we picked compounds that we think look
23:10good in the way they're positioned.
23:13And I also was very concerned about
23:16the idea that we would possibly be
23:20making analogs of these compounds
23:22because I didn't expect to have
23:25again come up with a 10 nanomolar
23:28compound we never have in the past.
23:30So we purchased 17 compounds and.
23:34Gave them to Karens lab,
23:37and Karen had meanwhile obtained
23:40the protein, expressed it,
23:42and she also had implemented the A fret
23:45assay that was from the literature.
23:48So she was ready to go.
23:50And the 17 compounds arrived.
23:54And to our surprise, in Karen's lab,
23:5814 of them showed some inhibition of the
24:02protease activity of Massar Scope 2 Proteus.
24:05So that was.
24:06Shocking.
24:07And so we were did very well on the
24:10compound selection and the most
24:12potent compounds are listed here.
24:14They were single digit micromolar.
24:19And but we had a bunch that were
24:22under about 50 micromolar.
24:24So that this we published and this is a
24:28picture of one of the dock structures.
24:31The binding site is you know is
24:34meant to accommodate a peptide
24:36that's going to get cleaved and we
24:39have site sub sites we call S1S1,
24:41Prime S2 and then this channel S3S4S5.
24:45So here's just a picture of a
24:48compound in that binding site.
24:50So we published that but of course
24:53we were looking very much now.
24:55And one of these compounds we're
24:57going to take and try to optimize it.
25:00And the compound we picked,
25:02we were we didn't say what it was
25:04going to be in this paper and it was
25:06not one of the most potent ones.
25:08In fact, it was this one param panel.
25:11Which is only 100 to 250 micromolar,
25:15so a relatively weak hit.
25:18But the fact was I liked the way it looked.
25:24And this was the dock structure.
25:28I'm orienting them all in the same
25:31way as 1S Primus 2 and I felt that
25:34the dock structure looked reasonable.
25:36Sometimes they they have features,
25:39they just say this doesn't feel right.
25:41But this looked reasonable.
25:43But I could also see that it had
25:45features that were not optimal.
25:47So looking at it over here,
25:49so the yellows are carbons,
25:52Reds are oxygens, Blues or nitrogens.
25:55I could see features that were not optimal.
25:57There's a histidine here and it could,
26:00it would be nice if it could form
26:03a hydrogen bond with this ring.
26:04So you probably want to put a nitrogen
26:06in here, this nitrogen of the purity,
26:08and that's not doing any good.
26:10So we can get rid of that.
26:13It's just spacing out into solvent.
26:15There's an NH over here that's.
26:18I would like to be in a hydrogen bond,
26:20but it isn't.
26:21Meanwhile,
26:21this carbonyl is just interacting
26:24with solvent,
26:25so maybe I could flip that from
26:27moving over left to there.
26:29Plus it looked like there was a
26:32little space in the meta position
26:35of that right ring.
26:37So.
26:39What happened next was we did some
26:42FEP calculations to test those
26:44ideas and this is what those are
26:46raw data looks like in an Excel
26:48sheet so that the the things I'm
26:51trying here are for the left ring.
26:55I'm going to try different rings.
26:57So ring scan where I did 234
27:01pyridinyl 4 pyrimidine 2 triazine,
27:04so a bunch of different rings there.
27:07They also did a calc and that those
27:10calculations said that the three pyridine
27:13the the negative number here is good.
27:16This is the change in free energy
27:19of binding relative to benzene.
27:21So this was saying go for the three pyrenee.
27:25Also I checked that ring flip of the
27:28carbonyl and that was very good,
27:30minus 4.7 and then over on the right
27:33side checking to see if we could
27:36put something in that meta position,
27:39indeed the meta position when we did
27:42a chlorine scan at each position,
27:45the meta here shed very good.
27:48Looks like we should put a chlorine there.
27:50So combining those three ideas
27:53led to then the.
27:55Three initial compounds
27:57that were synthesized.
27:58So here.
27:59Now I'm aligning everything so you
28:01can see the changes from parent panel,
28:04the three pyridyl.
28:06The carbonyl's been flipped and we've
28:09added the chlorine and we've left the
28:12the cyano phenyl from parent panel.
28:14I also from modeling with my bond program.
28:19Again, the,
28:19the slow part in all of this is synthesis.
28:22So we have plenty of time to do computer
28:25work while people are doing synthesis.
28:28So it's a natural thing to, you know,
28:32look very hard at these structures.
28:33And I had looked hard at this and I
28:36recognized maybe I could do something
28:38over with this ring because there's an
28:40edge that will show more clearly here
28:43of a loop that could use some hydrogen bonds.
28:47And I thought a uracil might work,
28:49so I'd modeled.
28:50Got with the program complex
28:51has looked very good.
28:53So we synthesized a uracil and also
28:56just this 35 dot clock compound.
28:59So this is a very happy day now.
29:02Because the potency of those original 3
29:06compounds was 10-6 and four micromolar.
29:09So here we've gotten a huge boost as
29:12expected from the FEP calculations.
29:14And this was the wonderful and I'll
29:17tell you the timing more in a bit,
29:19but this is now June of 2020.
29:23So we didn't get back into our lab until May.
29:27And now in June we have these,
29:31this 4 micromolar. Compound.
29:34We've only,
29:35and then it came a little later was
29:38actually October and Karen's group
29:40got a crystal structure for that
29:43dichloro compound and it's basically
29:46identical to what we've predicted.
29:48There's the carbonyl and hydrogen
29:51bond we wanted.
29:52There's a hydrogen bond between
29:54the pyridine and the histidine.
29:56We still have the nitrile hydrogen
29:58bonded in what he called the oxyanion
30:01sort of hole and the dichloro compound.
30:05Is again looking very good.
30:07Furthermore,
30:08we have this channel running N from
30:11the upper chlorine there and so we're
30:14ready to think about putting some
30:16of the something in that Channel.
30:19So the next thing was to try to grow
30:23substituents into that Channel and
30:25just for grins and I mean really not
30:28interested in methyl particularly,
30:30but just for grins, we did FP
30:32calculations for methyl ethyl propyl,
30:34O methyl ethyl propyl albuterol and
30:37then some ones with a hydroxyl that
30:40I figured probably wouldn't be very
30:42good problem with hydroxyl is it's
30:44very happy unbound said waters around
30:47and if you go bound it may be happy.
30:49Again, but you're not going to gain much.
30:52The way you gain is by having more
30:55hydrophobic pieces that are binding into
30:58hydrophobic part of the binding site.
31:00So this told us.
31:04Tried the O propyl compound
31:06so we synthesize on.
31:08There are two synthetic chemists
31:10are working on this so Lizzie and
31:14Chun way and so we they made.
31:19The proxy compound in both the
31:22cyano phenyl and the urea series,
31:25and this turned out great,
31:29140 nanomolar and 120 animal later on.
31:33This wasn't in sequence.
31:35We had made the trifluoromethyl analogs
31:38to that. They're more hydrophobic.
31:40They're probably going to be better
31:42binders as they were showing this
31:44one even down at 25 an animal,
31:47but generally I don't like CF.
31:49Big groups and drug like molecules
31:51because they really hurt the
31:53solubility of the compounds.
31:55So, but we're doing very well here 120.
31:59An animal or and I'll show
32:01you the timing on this,
32:02but this,
32:03this I think is in August now and
32:07Karen's group again got a crystal
32:10structure in October and it was exactly
32:14as expected including this bent.
32:17Hard at the at the end of the Propoxur group.
32:19And so it's a Ghosh. We call it a gauche.
32:22You've all taken organic chemistry, I'm sure.
32:27So that's the course you hated the most,
32:30but maybe, maybe not.
32:32But there it is.
32:33There's this gosche OCC and we
32:36had figured that was the case.
32:38The modeling told us that because at
32:41that terminal methyl fits right in
32:43the S4 site of the of that Channel.
32:48And so there's a lucine or problem,
32:50so hydrophobic site and so put
32:54it right in there also.
32:56Again, like I said,
32:57there's lots of time to do computing
33:00and so we considered benzel oxy groups.
33:04So you can imagine a benzene ring
33:08sitting here and potentially projecting.
33:12A substituent into that pocket.
33:15So sure enough we did modeling on
33:18these benzyl oxy analogs and did a
33:21chlorine scan on the fennel which said
33:24in a methyl scan and both methyl and
33:27chlorine were predicted to be very good.
33:30And so those compounds were made and
33:34the parent compounds 120 micromolar,
33:37but the ortho chloro compound
33:4118 an animal compound and this
33:44we had in October of of 2020.
33:48And Karen's group again got a
33:51crystal structure for the bend the
33:55parent Benz loxy components and
33:59is positioned as one expected.
34:02So this is just now a little video.
34:06To. Have those. Break this.
34:10This is a dimer, so they're two.
34:12This is Karen's crystal structure
34:16of the propoxur compound and
34:19just zeroing in on it.
34:22There.
34:29OK, so you can.
34:34Run it again. So that little
34:39molecule is enough to shut down the
34:43enzymatic activity of that protein.
34:50OK, so this we published and.
34:54I was also saying a a second here,
34:57we're going to of course the
34:58well I've shown you so far
35:00is just protease inhibition.
35:01We've got to go into cells,
35:02infected cells and so that
35:05we published 28 compounds.
35:07Of course by the results
35:09I've talked about so far,
35:10we have lots of compounds here
35:13under the 50 nanomolar and
35:14you can see there are authors,
35:16lots of people involved and from
35:18the medical school, you know,
35:20fair and Isaacs and Brett Lindenbach,
35:22grouper and very important.
35:24Along with Karen in doing the
35:27cell assays that will describe in
35:29a Miller in a minute and Scott
35:31Miller in chemistry had donated
35:33his graduate student Lizzie Stone
35:36to help us with the synthesis,
35:38along with my postdoc Chunwei Zang.
35:42So that was good.
35:43We published that in ACS Central
35:46science in February 2022.
35:47A little later we also replaced
35:50the benzyl Oxy with heterocycles.
35:52This is a standard, I'd say,
35:54medicinal chemistry.
35:55This isn't, you know, genius stuff.
35:59Heterocycles often have some desirable
36:02properties over a substituted benzene.
36:05So we published some more compounds
36:07in the summer than of a 2021.
36:10We also.
36:11Tested cell permeability with
36:14a pampa assay in our lab and
36:18measured aqueous solubility.
36:20So now we have uracil's with
36:22the hydrogen or with a methyl.
36:24So the ones with the methyl are going
36:27to have better cell permeability.
36:29And so that is an issue because we want
36:33to show that we have efficacy and sell assay.
36:37So this is where the,
36:39again the folks here in the Med
36:40school are so important to us.
36:42The BSL three facility was used.
36:45There's krassimir getting suited up
36:49because COVID, of course, is airborne.
36:51He has to have a full breathing
36:54apparatus and the assays that were done.
36:57Karen certainly can describe these far
36:59better than I can, but there's one.
37:01It's a a plaque assay using infectious virus.
37:05And so you have the live these are Vero
37:09cells infected with large live SARS Cove two.
37:14And there's also then the replicon.
37:16Assay and the Republican
37:18isn't using infectious virus,
37:20but it's giving us a very
37:22virtually identical readout.
37:24So we're testing our compounds and we have
37:28as a as a reference compound remdesivir,
37:32which is A1 micromolar EC50
37:37and the assays that were done.
37:40And long short, we have many
37:42compounds that are one micromolar.
37:45We also have some compounds.
37:46This one's 38 nanomolar EC 50,
37:49that's inhibition of the
37:52of the protease activity,
37:54but it's not active in the replicant housing.
38:00And this simply because it
38:01doesn't get into the virus.
38:03Cell permeability is too low,
38:05so the cell permeability is critical.
38:08The quite remarkable compound is number 19.
38:12So this.
38:14Benzyl oxy compound that has a
38:17methylated uracil and in the assay it
38:21was 80 nanomolar in the infectious
38:24virus assay and 175 and the replicon assay.
38:29So this became our our lead
38:32compound for preclinical work.
38:35Now unfortunately in our world we can't,
38:37you know we're not Pfizer,
38:39so we can't take 10 compounds and put
38:41them all into preclinical studies but.
38:44We did work on 19 and a pharmaceutical
38:48company was very interested in 19.
38:52They took 19 and did their own sell
38:55assay and they came back and their
38:57cell was 15 animals they can confirmed
39:00everything that we we had reported.
39:03So that compound 19 is a very potent compound
39:08in infected cells and Karen's group has
39:12been working on the PK, it has very good.
39:16Basic PK bioavailability.
39:18And they have done with Pretty Kumar
39:21some initial mouse studies and this
39:25is with these humanized mouse mice,
39:27KTH 2 mice.
39:29And again Karen could describe
39:31the current status of this.
39:34But basically we were delighted a very
39:37low dose of the compounds that were using
39:41and if you don't untreated mouse after
39:45six days as this is now fluorescent.
39:48Imaging of where the virus is.
39:50So initially the virus goes into the lungs,
39:53but it makes its way into the brain.
39:57And at day six,
39:59the mouse is again horribly
40:01infected and dies.
40:03So we have tested we meaning Karen
40:07and pretty by both Ivy and oral.
40:12And the results have been very good.
40:14There's only one dose and you see
40:17protection for four days, you know,
40:19completely clean a mouse and even at 6 days.
40:22So with the oral, it's,
40:25you know, really very clean.
40:26So if this was being dosed every day,
40:29the feeling is infection that wouldn't go on.
40:33So we have very, you know,
40:35concur raging data with this compound.
40:38There has been some.
40:40You know again external interest
40:42in this compound, yeah,
40:44we think we if we had the resources we
40:46can come up with lots of other compounds,
40:48but we need support for this and are
40:51you know high level because these
40:54preclinical studies are are expensive.
40:57So just to compare what we've done.
41:00Versus others.
41:01So first of all our compound is a non
41:04covalent inhibitor by most of the other
41:07work in this area been covalent inhibitors.
41:10Up until recently covalent
41:12inhibitors were considered to be.
41:15Not desirable because you're always
41:18worried about off target activity.
41:21But here is how other people progress.
41:23So a lot of these things are peptidic.
41:25Generally we don't like peptidic inhibitors
41:29because they can be proteolysis by many.
41:33Proteolytic enzymes that exist
41:36in humans so but this is some of
41:40the compounds and EC 50 of 720.
41:44Remember we're 50 or 80 nanomolar.
41:48This is the COVID moon shot that
41:49got quite a bit of publicity.
41:51This is just the icy 50.
41:53They obtained an assay 2400
41:57compounds and the best IC50 they
42:01obtained is basically 100 nanomolar.
42:04At 30,
42:05we had made no more than 30 compounds
42:08and we were at 18 nanomolar.
42:10Another peptide peptide peptide,
42:13but this is a PAX lovin.
42:17So Pax Lovid is this neurometrix alvir,
42:21but you have to include a SIP
42:24inhibitor ritonavir.
42:25So ritonavir is an HIV protease inhibitor.
42:29Not something you probably want
42:30to take for a long time and have
42:32their side effects of that.
42:33Of course you're not going to take
42:35packs a little bit for a long time.
42:36So I guess it's OK,
42:38but on the other hand having to
42:41have the SIP inhibitor to keep the.
42:45Protease inhibitor from being chewed up.
42:48Metabolically.
42:49Is clearly not desirable because
42:51you don't want to be, you know,
42:54can have drug, drug interactions.
42:56This is our compound.
42:58Again, by comparison,
42:59other things that you know.
43:02I'm obviously a little bit prejudiced here,
43:04but this to me is a tough molecule.
43:08All the stereo chemistry going
43:10to be tough to synthesize.
43:11You have high cost of goods.
43:13It's peptic. You worry about that.
43:15It is a covalent inhibitor,
43:17covalently modifies the cyano,
43:19but it's probably reversible.
43:21Are covalent.
43:22There have been a synthesis
43:24issues with the compound.
43:27It's also intrinsically not
43:28as potent as our compound.
43:30It's a EC 50 or 740 whereas we're at you
43:35know 10 times more potent with there's no,
43:38we don't we know from our preclinical
43:41work on off target and SIP
43:43activity that we don't have any sip
43:46problems with the compound either.
43:47So the rest of the story.
43:51So why isn't our compound in clinical
43:53trials and that's a probably takes me
43:57more than the last time I I have here.
44:00But the packs of lovin and thermal,
44:03Trevor got into clinical trials very
44:05quickly because it was sitting on the
44:08shelf from the SARS Cove One project.
44:10They made a minor modification to
44:12make it have better solubility.
44:14So it was ready to go and
44:17so it's off and running,
44:19I doubt seriously it's the best.
44:21Drug possible,
44:23and there's no way, and well,
44:26time will tell the problem for the
44:28pharmaceutical companies that they're
44:29all in the business of making money.
44:32And so the before the end of last fall.
44:37People are getting kind of cocky about,
44:39you know, covid's under control,
44:42the vaccines are working.
44:44And then Omicron came along around December
44:48of last year and that's changed things a bit.
44:53But we'll see who has the, you know,
44:57stamina to advance additional protease
45:00inhibitors into the clinic because
45:02of the cost of the clinical trials.
45:05This is a timeline just showing
45:06the power I think of our approach.
45:09So June 15th all we had was parent panel.
45:12By August 3rd we had these six
45:15and four micromolar compounds.
45:17By.
45:18September 2nd we had the proxy
45:20140 nanomolar compound.
45:22September 10th we had the corresponding of.
45:26Benzyl oxy,
45:27uracil and then we started getting
45:30some crystal structures October 3rd.
45:33We had the first crystal structure October
45:374th and also the Propoxur compound and
45:41but the speed here which we got to the.
45:46These sort of loading animal compounds
45:49again to get to 18 animal we had
45:52synthesized about 30 compounds and a
45:54few of them were things we probably 8
45:57or 10 of them were real or wild shots.
46:02And this synthesis was done by a gun.
46:04Postdoc chunwei and graduate student Lizzie.
46:07So that's the story and I I hope
46:11I've told you a little bit about
46:14what Karen and I do and the.
46:16Hour of combining the computation with the,
46:21you know,
46:22reliable assaying and crystallography
46:24is such a different world than
46:26what we lived in 20 years ago.
46:28So just thanking people in my lab notably.
46:33And Julian is a long-term associate other
46:35so he's a senior research scientist.
46:38Anna and Joe were both associate research
46:42scientist and other people listed here.
46:45Karen of course my.
46:47Wonderful collaborator and
46:49other Pi collaborators,
46:51pretty yosi on our Jack projects and Brett,
46:57Brett and Faron on the COVID project.
47:01So pleasure to be here with
47:02you and thank you very much.
47:12What a whirlwind journey.
47:13Yeah, amazing.
47:15Are there any questions here,
47:18Emily? Are you monitoring
47:19questions in the chat, Tommy?
47:35Yeah, these the COVID compounds
47:37are all binding to the active
47:41site of the Proteus.
47:43So the cysteine, that's a,
47:44there's a cysteine,
47:45it's a cysteine protease,
47:46there's a cysteine it.
47:48We're sort of in the middle of all
47:50the structures I showed you and
47:52that's the active site cysteine.
48:02Even fine.
48:08Yes, the the Pfizer compound
48:12binds in that same site, and it
48:15covalently modifies that cysteine.
48:18And you? Does not covalent.
48:27Yes.
48:34Dog.
48:40Quite understand, he's asking if
48:43in the crystal structure does
48:45it bind to the cleavage. Yes,
48:48and this is the.
48:51The cysteine there cysts 145
48:54and this histadine over here.
48:58Are the catalytic residues,
48:59so our compound sitting right on top of them.
49:03And the Pfizer compound covalently modifies
49:07that cysteine as do most of the other.
49:11There's a very few coat non covalent
49:13inhibitors have been reported for this.
49:16But we from the getco we wanted
49:18to pursue non covalent inhibitors
49:20just to avoid the potential
49:23issues of covalent inhibitors.
49:29So you know, we're we're
49:31extremely familiar with the hyper
49:33immutability of this virus in the
49:36spike protein to evade immunity.
49:38I wonder if you've done sort of low dose
49:41exposure and if there's a mutational.
49:44Response to to a protease
49:46inhibitor like this?
49:49Yeah, I haven't maybe.
49:50I mean, we haven't to my knowledge,
49:53unless Karen's been up to
49:55something I don't know about,
49:57the SARS Cove 1 protease and SARS Cove
50:012 protease are extremely identical.
50:05They're the only differences are quite
50:07far from the the protease active site.
50:10So it's it's hoped that there won't be.
50:15A lot of mutations possible
50:18for the Proteus, however,
50:19it hasn't been under pressure.
50:21So I think with the Pax lovid treatments
50:24we will probably begin to see some
50:28mutations closer to the binding site.
50:31And there there was a recent
50:33paper in science,
50:35I believe is it science indicating some
50:40mutations that might arise in this Proteus,
50:43so it was under some pressure that they.
50:46Put it,
50:47but we haven't looked into that yet.
51:01So this is a this is a related question,
51:03but how different is the COVID 2
51:06protease active site from that
51:08of other common human proteases?
51:11Umm. Well, I would say it's it's the
51:17COVID active site is quite unique,
51:20but it's virtually identical to
51:23the SARS Cove one active site,
51:25but I don't think there's been a.
51:30I don't think that these inhibitors are
51:33generally inhibiting other proteases.
51:39So I I haven't heard that,
51:42so I don't expect it, but if they were,
51:45it would certainly be, I'd imagine
51:47it'd be assisting a protease would
51:49be the ones you'd be looking at.
51:53Well, we're we're at the hour.
51:55I can't thank you enough for this
51:58lucid explanation to a bunch of
52:00non chemists was really beautiful.
52:02Also on behalf of healthcare
52:04workers who, you know,
52:05see people with COVID all the time.
52:07It's wonderful work.
52:07Thank you very much. Thank you.