Session 8: Protein Kinases

Transcript of Part 2: Architecture of a Protein Kinase

00:00:05.18	In the first lecture I tried to articulate for you
00:00:08.02	why protein phosphorylation is so fundamental for biology.
00:00:11.25	What I'd like to do here is to delve more deeply into the protein kinase structure and function.
00:00:19.17	And if we look at PKA, PKA is, again, a prototype kinase we understand best.
00:00:28.18	And it is activated by hormone neurotransmitter binding to the outside of a cell
00:00:37.00	in this case adrenalin. Every time...adrenalin works the same as glucagon.
00:00:40.25	And every time you secrete adrenalin into your bloodstream
00:00:43.21	you are activating PKA and you're binding to a G protein coupled receptor.
00:00:48.11	You're going through a G alpha subunit and you're hooking up to adenylate cyclase,
00:00:53.08	which leads to the generation of cyclic AMP.
00:00:56.12	And in this case, PKA, again, is dormant and it's activated when
00:01:03.22	cyclic AMP binds to the regulatory subunits and then you unleash the catalytic activity.
00:01:09.02	And there are many substrates.
00:01:10.10	So I'm going to focus here on the catalytic subunit. How does that work as a catalyst?
00:01:16.03	And next time we'll talk about how it's kept in this dormant state by the regulatory subunits
00:01:21.25	and how it's targeted to specific sites by scaffold proteins.
00:01:25.25	But, today I want to focus on the kinase.
00:01:28.07	So if we go back to this kinase history, we began to find...
00:01:35.17	there are many many protein kinases and this came from cloning many different proteins.
00:01:41.25	So you could very quickly get the sequence.
00:01:43.24	Instead of doing this sequence the hard way, by chemically doing that sequence.
00:01:49.15	And so the first protein kinase sequence was PKA. It was solved by Tatoni in 1981.
00:02:00.11	And it served then as a template, as a framework.
00:02:05.13	We had...I told you Src was cloned in 1979 but it wasn't until PKA was sequenced
00:02:11.03	that Margaret Dayhoff realized that Src was related to PKA. The two sequences were related.
00:02:18.12	So that's what put them on the same tree here.
00:02:21.02	And then in 1991, we did the first structure of a protein kinase.
00:02:25.10	And that allowed us to see the 3-dimensional features of this enzyme family.
00:02:32.07	So, let me tell you the kinds of information we gleaned from each of those findings.
00:02:40.04	So here's the kinome. The kinome is based on sequence analysis only.
00:02:46.03	And most genomes now have about 2% of their genome is coding for protein kinases.
00:02:54.11	And plants are the winners. They have 4%. They have the most protein kinases.
00:02:59.04	So it's one of the largest gene families in every eukaryotic organisms.
00:03:05.26	If we look at the sequence. So what does the sequence tell us?
00:03:10.07	It tells us that...this is a chemical description of your molecule.
00:03:15.10	You can see how from that sequence,
00:03:18.00	we could then immediately know that the cloned Src is a homolog of PKA.
00:03:23.11	And then we could do chemical things; modification with an ATP analog, here modified lysine 72.
00:03:31.08	That told us this lysine was close to the active site, close to the ATP binding site.
00:03:36.06	And it also...we later showed that you could
00:03:40.25	cross-link it to two different carboxylic acid residues, acidic residues.
00:03:47.07	So that all three of these residues, which are far apart in the linear sequence,
00:03:52.00	must be close together and near the active site of the folded protein.
00:03:57.07	And then the other piece of information here is the phosphorylation site,
00:04:02.01	which is essential for activity and that was associated with this particular threonine.
00:04:10.02	So those are the little bits of information we had.
00:04:12.09	You could take those two kinases,
00:04:16.02	once you knew that Src and PKA were related and other kinases became cloned.
00:04:20.04	The initial finding was from Margaret Dayhoff that Src and PKA were related
00:04:25.10	and had many conserved residues.
00:04:28.02	And then Tony Hunter...Hanks and Hunter did an analysis of the subdomains
00:04:35.13	and elucidated these various subdomains, each having a specific motif.
00:04:39.09	And those subdomains have really held up with time as the definition of the protein kinase core.
00:04:47.20	But then when we had the structure, now you could begin to understand
00:04:52.15	how those motifs fold together to form an active kinase.
00:04:57.09	So now you can begin to map that sequence
00:05:03.22	more comprehensively onto a structural framework.
00:05:07.18	So here we show the two major domains that are conserved in the kinase family.
00:05:12.17	One, the small lobe, is mostly associated with ATP binding.
00:05:16.12	The large one has a lot of residues associated with peptide binding and catalysis.
00:05:21.28	Those are the conserved residues and you can see they're spread out over the whole kinase core.
00:05:27.02	It also has this phosphorylation site here. This is essential for its activity.
00:05:33.01	And another phosphorylation site out here at the C-terminus, outside the core.
00:05:37.18	OK, so this is the N-lobe and this is the C-lobe.
00:05:41.26	OK, so now we can have a more in depth understanding of those motifs.
00:05:48.09	And this is what we now know is the conserved core that is shared by all of those proteins on that kinome tree.
00:05:56.25	They all have this common element to them.
00:06:02.09	OK, so if we look at different examples,
00:06:05.27	I've showed you PKA, I earlier talked about Src down here with its SH3 and SH2 domains.
00:06:13.15	These are just some other kinases. And you can see PKA is actually kind of unusual
00:06:19.06	in that it has a relatively small...it has little tails at each end but it's mostly
00:06:26.04	the kinase domain and its regulatory part is a separate subunit--the regulatory subunit.
00:06:33.00	Whereas other kinases have fused domains: C1 and C2 attached to PKC.
00:06:37.21	They bind to calcium and diacyl-glycerol.
00:06:41.01	Cyclic GMP protein kinase, very similar to PKA, only it's cyclic GMP binding domains are fused to the kinase core.
00:06:48.12	So there's all of those have in common the kinase core but each has these variations.
00:06:54.18	So now, again, if you come back to the core,
00:06:58.23	you can actually map those subdomains onto the kinase core.
00:07:03.22	And so I'm going to show you, again, how these now correlate with the N-lobe and C-lobe.
00:07:10.20	And tell you a little more about the two lobes as well.
00:07:13.14	Here's now the sequence of PKA where we can map the secondary structure;
00:07:20.00	a helix versus a beta strand. You can map it onto to those subdomains
00:07:24.22	and have a 3-dimensional context to those subdomains.
00:07:30.12	So here's those subdomains mapped onto the core of PKA and color-coded again.
00:07:38.14	So you can see the subdomains that comprise each of these lobes, the entire conserved core.
00:07:47.06	And, I'm going to walk you through these really quickly
00:07:50.09	just to show you some information about these.
00:07:54.25	So, this is the first on. Subdomain I is what we call the gylcine-rich loop.
00:08:00.03	And here you can see the gylcine-rich loop here.
00:08:03.21	You can see the 3 gylcines that are conserved in most kinases up at the top.
00:08:08.21	And the features you can begin to sort out what are the features that this motif does.
00:08:15.01	And then that's followed by subdomain II.
00:08:18.10	Here's that lysine, that K. That's the lysine that gets modified by ATP analog.
00:08:24.02	This is the C-helix. This is only helical element that is in the small lobe.
00:08:32.08	And then you have another beta strand.
00:08:37.07	And then you have this subdomain V that actually links the two domains.
00:08:43.12	This beta strand is in the N-lobe and this helix is in the C-lobe.
00:08:49.25	And this little piece joining them is the linker that joins the two lobes.
00:08:55.09	And then the E-helix. This is a very hydrophobic helix.
00:09:00.01	And the catalytic loop.
00:09:02.11	This is going to be essential for catalysis. It's a little beta sheet that's in the C-lobe.
00:09:09.04	And this is another part of that beta sheet. This is the activation loop.
00:09:15.14	And so I'll talk about these in more detail.
00:09:18.03	This is the F-helix--very, very important. This entire molecule is organized around this F-helix.
00:09:25.05	It's unusual in that it goes right through the middle of the C-lobe.
00:09:29.20	It's very hydrophobic. Just analyzing it you would think it was a membrane-spanning helix.
00:09:35.04	Extremely hydrophobic.
00:09:37.14	And then this G-helix serves as a docking site for proteins and H-helix.
00:09:42.10	And here just to show you an example comparing the subdomains of PKA and Src,
00:09:48.10	you can see there are little differences but overall, those subdomains are conserved.
00:09:53.10	The fold is conserved in all of those protein kinases, all of those 500.
00:09:58.28	So, I'm going to tell you about these lobes first
00:10:02.06	because I think it will give you an understanding of how those subdomains work together
00:10:07.02	synergistically to create an active kinase.
00:10:10.00	And so this is the N-lobe and it contains this glycine rich loop
00:10:15.02	which I told you about. Those are the three glycines.
00:10:17.16	And then the C-helix is the other really conserved feature here.
00:10:22.09	And you can see the lysine 72 and Glu91.
00:10:26.03	Those are the residues that were linked together by our chemical studies.
00:10:30.14	So when we first saw this structure, my first thing was to say,
00:10:34.07	"Where's that lysine? And where are those residues that cross-link to it?"
00:10:37.19	And it was...they were all right next to each other just as we thought from the chemical studies.
00:10:43.17	But we couldn't understand how they related to each other until we saw that structure.
00:10:49.04	OK, so the glycine loop is the most mobile part of the kinase.
00:10:58.18	And it has to open and close.
00:11:01.08	And when it closes it fits down on top of the gamma phosphate of ATP
00:11:05.28	and positions it to be transferred to a protein substrate.
00:11:11.08	So this opening and closing of the glycine loop is essential for catalysis.
00:11:16.01	And that hydrogen bond there between the glycine loop and the gamma phosphate of ATP is critical.
00:11:26.22	Now we go to the large lobe.
00:11:28.17	So it's mostly helical but it has this beta sheet that's right at the active site cleft.
00:11:33.20	And some of those conserved residues lie here in these two loops
00:11:39.18	and those are very essential for transferring that phosphate.
00:11:42.29	And that lies on top of this very stable helical subdomain.
00:11:48.03	The N-lobe is very malleable. This lobe is very, very stable.
00:11:53.00	And some key features that the phosphate that's important for activating the kinase
00:11:59.16	is here, and when you phosphorylate that site, the active site is maximally active.
00:12:10.10	Without that phosphate it's not active.
00:12:12.09	So even though it's 20 angstroms away from the active site, it is essential to create the active kinase.
00:12:18.15	And without that phosphorylation, the enzyme really unfolds very easily
00:12:23.16	and its chemical properties are quiet different.
00:12:27.02	One phosphate, just one phosphate.
00:12:30.02	This is the catalytic loop. Many of the catalytic residues that are important.
00:12:34.07	This P+1 loop is important from docking peptide substrates.
00:12:40.11	So, if we look at a mimic of what we think might be a transition state for transferring the phosphate,
00:12:47.06	this is the...aluminum fluoride serves as a mimic for the gamma phosphate of ATP
00:12:54.19	and you have the residues from the catalytic loop in the C-lobe,
00:12:58.17	the residues from the glycine rich loop in the N-lobe.
00:13:01.15	They converge on that gamma phosphate and transfer it to the peptide substrate.
00:13:06.17	So, very beautiful chemistry of how these come together at the active site.
00:13:12.05	So let's just look, for a moment, at catalysis.
00:13:15.03	So, the kinase must open and close as part of its catalytic cycle.
00:13:20.19	So, when it's open, it's actually quiet malleable.
00:13:24.28	Then, it binds ATP and its peptide substrates and here, for that transient moment of catalysis,
00:13:32.05	it's in a closed conformation.
00:13:34.08	So, if we look at how these look in the molecule, this is the open conformation on the left
00:13:44.20	where you have a hole there, that's where the ATP is going to fit into that big hole.
00:13:48.19	And then the closed conformation, you can see how that active site cleft really comes down and closes.
00:13:55.09	And now, in this...this is just going to give you an image of how this kinase opens and closes.
00:14:01.19	And it's morphing the open and closed states.
00:14:03.24	And I'm just going to point out, before we see the movie, the lysine 72 and the Glu91.
00:14:09.22	Those were two of the residues that cross-linked.
00:14:11.24	So now you can get a sense of how opening and closing of the active site cleft is taking place.
00:14:21.11	And so this is just a morphing--the open conformation and the closed conformation,
00:14:25.25	essentially the breathing of a protein kinase molecule.
00:14:30.21	OK, so if we look up close now at that closed conformation where ATP is bound,
00:14:37.06	you can see most of the ATP is buried in that deep hole there.
00:14:40.22	And just the little gamma phosphate is sticking out at the edge of the interface
00:14:46.12	between the glycine rich loop from the N-lobe and the catalytic loop from the large lobe.
00:14:53.00	And then you can see here how a peptide fits into that site.
00:14:57.01	So, it docks mainly onto the large lobe.
00:15:00.27	Here's the P+1 residue. It's a nice, in this case, hydrophobic pocket
00:15:04.20	where that hydrophobic residue docks.
00:15:07.29	This is a pseudo-substrate. It's an alanine instead of a serine.
00:15:11.06	If that was a serine it would accept the phosphate and transfer the phosphate.
00:15:16.08	And then, PKA likes basic residues and so you have acidic residues on the protein in red
00:15:23.06	that recognize those basic amino acids in the peptide.
00:15:26.09	And then for the inhibitor peptide which is bound in...this is a very high affinity binding peptide.
00:15:35.17	It has a helix that docks to the active site cleft and this serves as a tethering site.
00:15:45.08	So I've told you how PKA works. It's the kinase that we understand best in terms of structure and function.
00:15:53.10	But now we have this whole kinome tree.
00:15:55.22	And, what can we learn now from a large family?
00:16:02.15	We have...because these are so important for diseases, we have many kinase structures.
00:16:07.17	Kinases have become a major target for drug discovery. So we have many kinases.
00:16:12.22	So I've shown you what we can learn from in depth analysis of one kinase.
00:16:18.13	What can we learn from this structural kinome now where you have
00:16:22.26	not only many sequences but you have many structures?
00:16:26.16	And, we can learn different kinds of things.
00:16:28.28	I'm going to tell you some global lessons we can learn about the entire family.
00:16:32.10	One can delve deeply into a single family
00:16:35.05	and find out what are the unique features of one subfamily versus another.
00:16:39.27	So this is just six of the different kinase structures, kinase cores.
00:16:47.25	And you can see that the subdomains are mostly conserved in all of these kinases.
00:16:53.22	What can we learn from that structural kinome?
00:16:59.04	One of the things that we did was to develop a method which we call Local Spatial Pattern Alignment.
00:17:07.01	And this is just rapidly comparing any two structures and looking at the spatially similar residues.
00:17:15.09	And this provides you with a pattern.
00:17:18.05	You can do this independent of any sequence alignment.
00:17:20.25	Just two structures, compare them.
00:17:23.07	And you get figures like this that shows the spatial relatedness of residues in two protein kinases.
00:17:34.00	And so if we look here you can see a couple of key residues
00:17:38.09	that these edges indicate a spatial link between two amino acids.
00:17:46.23	And the more edges that go to a particular amino acid, the higher its involvement score.
00:17:52.15	We call that an involvement score.
00:17:54.24	So we make a network of spatial related residues in two protein kinases.
00:18:01.24	And this is called the involvement score. The higher the involvement score,
00:18:05.15	the more spatially-related amino acids interact with that particular amino acid.
00:18:12.24	So we use this first to compare active and inactive kinases.
00:18:20.04	What is different?
00:18:21.06	Can you find spatially-related differences that are unique to
00:18:28.01	the active kinase and that are not there in the inactive kinase?
00:18:31.06	And from this analysis we defined what we call a spine.
00:18:38.06	We've subsequently called it a regulatory spine
00:18:42.00	It's a hydrophobic spine that is spatially conserved in every active kinase
00:18:47.12	and is broken when the kinase is inactive.
00:18:50.22	And it's made up of hydrophobic residues.
00:18:53.02	They come from non-contiguous residues. So one is coming from beta-4, one is from alpha-C,
00:18:59.16	one is from the DFG motif, one is from the HRD motif, two are in the N-lobe, two are in the C-lobe.
00:19:08.14	And they align to make this hydrophobic spine.
00:19:12.00	You would not ever find this by sequence comparisons
00:19:15.11	because, by sequence analysis, they're non-contiguous
00:19:17.26	but they are spatially well-defined as a conserved motif.
00:19:25.20	So we then went back and looked at all of the kinases
00:19:29.00	and found that there were actually two spines.
00:19:34.03	And so we found a second spine, which we call the catalytic spine, in addition to the regulatory spine.
00:19:40.22	And the importance of...again, it's hydrophobic residues that are from both the N-lobe and the C-lobe.
00:19:48.04	The unique thing about the catalytic spine is that it's the adenine ring of ATP that
00:19:52.19	completes the spine and links the two.
00:19:54.26	So that means once you add ATP, you now have a new coordination between the N-lobe and the C-lobe.
00:20:04.02	So the other feature that came from this analysis is this F-helix.
00:20:08.09	This a very hydrophobic helix that spans the C lobe.
00:20:12.06	This is very unusual for a globular protein like a kinase.
00:20:17.02	But, it's very highly conserved and it serves as the framework...
00:20:22.22	it's linked to both the regulatory spine and to the catalytic spine
00:20:29.05	And it nucleates...it really provides the architecture on which you assemble an active kinase.
00:20:36.27	And we call this the S2H motif or two spines and a helix.
00:20:41.19	And that is really the fundamental architecture that every protein kinase.
00:20:46.21	So, here I show you just that F-helix.
00:20:49.12	You can see how it goes right through the middle of this C lobe.
00:20:55.02	You would never predict that this was part of a globular protein by sequence analysis.
00:21:01.10	And you can see, as this rotates around, the relationship of that F helix to the rest of the protein.
00:21:13.06	It really...everything is nucleated around that F helix.
00:21:16.25	You can see some of the key residues that emerge
00:21:20.29	that are going to be important for linking to the two spines that are coming out of that F helix.
00:21:26.26	There you can see the two spines,
00:21:28.04	how those two spines are really anchored in a very fundamental way to the F helix
00:21:33.10	and the catalytic loop. All the functional elements are really linked to this F helix.
00:21:44.12	So, again, you can see here, when you open and close the kinase,
00:21:48.22	those conformations...you can see that the two spines stay intact as part of that breathing motion.
00:21:55.10	It does not interfere with the breathing motion of the kinase.
00:22:02.03	So, I'll give you an example here of the...this is the insulin receptor,
00:22:06.15	another member of that tyrosine kinase branch.
00:22:11.03	And this is the active conformation of the insulin receptor and you can see that
00:22:15.00	the two spines are intact. This is a fully active conformation of the insulin receptor kinase.
00:22:20.27	When it's not active (and it's activated by phosphorylation of its activation loop here),
00:22:28.06	when it's not phosphorylated on its activation loop,
00:22:31.13	then you can see the spine in broken.
00:22:33.25	And this is one way in which it can be broken.
00:22:37.28	Actually, a residue from the regulatory spine goes over and fills the adenine pocket
00:22:43.07	for the catalytic spine.
00:22:44.10	But there are many, many variation of how you can break that spine.
00:22:49.20	So, why is that phosphate so important?
00:22:53.00	I talked about this phosphate being really important.
00:22:58.08	It's important for the insulin receptor. It's important for PKA.
00:23:02.04	That phosphate on the activation loop...
00:23:04.22	And I'm just going to show you: Here's a phosphate in PKA
00:23:08.22	and it's making seven different either electrostatic or hydrogen bond interactions
00:23:17.10	with different side chain residues.
00:23:20.09	And it's such an integrating phosphorylation site.
00:23:24.19	It goes to this histidine which is in the C helix, in subdomain 3.
00:23:30.10	It goes to lysine 189 which is actually in beta-strand 9.
00:23:37.26	It goes to arginine 165 which right before the catalytic loop.
00:23:44.26	And it integrates this whole activation loop.
00:23:48.24	So, it's playing a critical role in integrating all of the subdomains of the protein.
00:23:56.18	So in this last part I wanted to go back to the kinases and disease
00:24:00.07	and again, relate to what I told you about these subdomains and the spines.
00:24:08.25	There are probably more now, but at least 30% of kinases are implicated in various diseases
00:24:13.28	and that is just growing all the time as we delve more deeply into the different kinases.
00:24:20.08	Many more are likely to follow as we get more and more genomic information
00:24:24.21	and disease correlations.
00:24:28.10	Kinases are tractable as drug targets because you can inhibit them and
00:24:34.00	I'm going to show you can example of that.
00:24:36.01	So this is the human kinome and I'm showing you one kinase in the tyrosine branch there.
00:24:42.28	This is Abl, a relative of Src.
00:24:46.07	And in this fatal disease chronic mylogenous leukemia
00:24:51.23	it's this Abl which is modified. It's fused to another protein and it makes it an oncogene.
00:24:56.27	So, it's constitutively active. It's turned on all the time.
00:24:59.23	That's what makes an oncogene--you can't turn it off.
00:25:03.03	So, this Gleevec was discovered as a very specific inhibitor of BCR-Abl.
00:25:12.07	And this is the real proof of principle that says that kinases are very tractable drug targets.
00:25:23.07	If we look at Abl with ATP bound to it you can see here the two spines;
00:25:28.09	the regulatory spine, the catalytic spine. This is active Abl.
00:25:32.16	And then you compare that now to the structure where Gleevec is bound
00:25:37.17	and this was done by John Kuriyan's laboratory.
00:25:42.04	It is bound to the active site cleft but you can see when Gleevec binds,
00:25:46.22	the regulatory spine is broken and so it's actually weaving through
00:25:52.07	both the catalytic spine and the regulatory spine.
00:25:57.10	So it binds to an inactive conformation
00:25:59.12	and you can see precisely the features that are being taken advantage of by Gleevec.
00:26:05.03	So, one of the sites that is frequently associated with resistance to Gleevec
00:26:11.20	is a mutation of this threonine residue in the ATP binding pocket here.
00:26:17.10	So this threonine is sometimes referred to as the gatekeeper.
00:26:20.04	And I think in the previous slide...here's the gatekeeper residue.
00:26:25.09	I show you where the gatekeeper residue is.
00:26:27.09	It's a threonine. And so the mutation that takes place often is that it's converted to methionine.
00:26:33.10	So threonine is a small residue. Methionine is a much bulkier residue.
00:26:37.24	It's so bulky that it will prevent Gleevec from binding.
00:26:41.15	It's a steric hindrance so it blocks that.
00:26:43.19	But, it does more than that and this recent study by Daley and Kuriyan shows that
00:26:51.07	when you have the methionine there instead of the threonine,
00:26:55.08	methionine is a nice hydrophobic residue and it fills a gap between
00:27:01.22	the catalytic and the regulatory spine and what it does is it now creates a very stable, active spine.
00:27:08.06	So by changing that one residue, threonine, to an isoleucine (methionine),
00:27:13.17	you not only abolish the binding of Gleevec, you also create an oncogene.
00:27:19.03	You now have stabilized this regulatory spine in a way that no longer requires phosphorylation.
00:27:26.01	So this is not only resistant to Gleevec, this protein is an oncogene.
00:27:35.06	So, I've talked about the subdomains and how they're conserved in the family
00:27:39.26	and I have given you an appreciation of how kinases work
00:27:42.24	and why they are a tractable target for drug discovery
00:27:46.18	and then what I want to talk about next time is how the kinases regulate
00:27:51.23	and to really do that we're not going to be so interested in the catalytic machinery,
00:27:56.01	we're going to be really interested in the surfaces of the kinase
00:27:59.18	and how does PKA in particular bind to other proteins.
00:28:04.04