Session 4: An Introduction to Polyketide Assembly Lines

Transcript of Part 2: Dissecting Polyketide Assembly Lines

00:00:08.07	Greetings.
00:00:09.18	My name is Chaitan Khosla
00:00:11.09	and I'm a professor at Stanford University,
00:00:15.19	and this is part two
00:00:18.01	of the trilogy of my lectures
00:00:19.28	on assembly line polyketide biosynthesis.
00:00:24.11	In the previous lecture
00:00:25.27	I introduced you to the evolutionary biology
00:00:30.15	of these remarkable assembly lines,
00:00:33.00	the chemistry that happens
00:00:34.10	on these assembly lines,
00:00:36.07	and I gave you a general idea
00:00:38.05	of what these assembly lines look like.
00:00:42.02	So, we looked at the
00:00:44.12	6-Deoxyerythronolide B,
00:00:47.03	or DEBS,
00:00:48.22	synthase,
00:00:50.11	that is responsible for making
00:00:52.23	this precursor of erythromycin.
00:00:57.21	I ended the previous lecture
00:01:00.15	by giving you a sense of what
00:01:02.18	we think this assembly line looks like
00:01:04.22	and how that insight was derived.
00:01:08.21	What I'm gonna start this module with
00:01:12.02	is an introduction to the kinds of tools
00:01:14.24	we use to interrogate
00:01:17.01	the biochemistry
00:01:19.06	of these remarkable assembly lines.
00:01:22.12	So,
00:01:25.13	these assembly lines exist,
00:01:27.27	as we discussed in the first lecture,
00:01:30.22	in relatively esoteric sources.
00:01:33.17	They usually come from bacteria
00:01:35.22	whose names many of us have a hard time spelling,
00:01:39.11	or sometimes even worms,
00:01:41.25	or often times just sequence information
00:01:45.04	that was derived from DNA
00:01:47.04	that was collected from some place.
00:01:50.16	In order to study these systems,
00:01:53.07	one of the first sets of tools
00:01:55.14	that we developed
00:01:57.24	was to be able to take these
00:01:59.15	very complex metabolic pathways,
00:02:02.12	like the DEBS metabolic pathway,
00:02:04.26	and put them into
00:02:08.00	genetics-friendly microorganisms;
00:02:10.17	hosts like E. coli.
00:02:13.21	So, today you can make
00:02:16.20	6-Deoxyerythronolide B
00:02:19.15	in E. coli
00:02:21.07	by growing it in the presence of
00:02:24.12	glucose as a source of energy
00:02:26.11	and propionic acid
00:02:28.00	as a source of all the carbon
00:02:30.05	that's used to make the product,
00:02:32.04	so long as the recombinant E. coli
00:02:34.11	contains DNA
00:02:36.26	that instructs for the biosynthesis
00:02:39.10	of those three very large proteins
00:02:42.09	that comprise DEBS.
00:02:44.18	This is,
00:02:46.16	for obvious reasons,
00:02:48.18	a very powerful tool
00:02:50.15	to interrogate DEBS
00:02:52.06	because, now, if I have a bacterium
00:02:54.29	that makes my product for me,
00:02:57.09	I can go in there
00:02:59.15	for the price of a $200 kit,
00:03:02.12	manipulate the DNA
00:03:04.12	that encodes this assembly line
00:03:06.19	and ask,
00:03:08.06	what are the consequences of this assembly line?
00:03:11.03	And these tools have been
00:03:13.02	in our armamentarium
00:03:15.09	for the better part of the past two decades.
00:03:17.20	My previous generation of lectures
00:03:20.01	talked quite a bit about these tools,
00:03:22.11	so I won't spend a lot of time
00:03:24.01	doing so again,
00:03:26.05	but these tools
00:03:28.12	have played a critical role
00:03:30.03	in our understanding
00:03:31.21	of the biochemistry
00:03:33.01	of these assembly lines.
00:03:34.17	Now, what is more challenging,
00:03:37.12	but is perhaps arguably more important is,
00:03:41.28	if you want to study this remarkable enzymatic assembly line,
00:03:46.01	you'd like to be able to peel off the wrapper
00:03:49.28	that surrounds these remarkable proteins.
00:03:53.08	That is easier said than done,
00:03:55.20	but today, we can reconstitute
00:03:58.20	the entire 6-Deoxyerythronolide B synthase
00:04:02.15	from purified proteins.
00:04:05.06	What I show you on the lower-left corner
00:04:09.09	is a protein gel, an SDS-PAGE,
00:04:12.27	that shows five proteins.
00:04:17.20	The two proteins to the far right
00:04:21.01	are the second
00:04:23.23	and the third protein
00:04:26.15	of the erythromycin assembly line.
00:04:30.20	And each of them, as you can tell,
00:04:33.12	has a monomeric molecular mass
00:04:36.01	that exceeds 300 kilodaltons.
00:04:40.01	The third protein,
00:04:41.22	which is the first of these proteins,
00:04:45.03	could not be expressed
00:04:46.23	for love or money
00:04:48.14	in E. coli
00:04:50.13	in a form that gave adequate yields
00:04:53.07	of pure protein
00:04:54.28	to study biochemically,
00:04:56.28	and so we had to break it up
00:04:58.26	into three pieces,
00:05:01.04	which are shown in the
00:05:02.24	first three [lanes] of this gel,
00:05:05.14	and purify those pieces independently.
00:05:10.08	And now you can put those three pieces
00:05:13.14	together with the other two proteins
00:05:16.03	to make a cocktail of proteins
00:05:19.25	that, in the presence of appropriate substrates
00:05:23.06	-- propionyl coenzyme A,
00:05:25.11	NADPH,
00:05:27.04	and we don't use methylmalonyl coenzyme A itself,
00:05:31.01	instead we use an in situ enzymatic generation method
00:05:35.22	for methylmalonyl coenzyme A
00:05:38.00	where we use free methylmalonic acid,
00:05:41.01	coenzyme A,
00:05:42.24	and an enzyme called malonyl-CoA synthetase --
00:05:46.07	and so when you put these five proteins,
00:05:49.14	which have been purified,
00:05:51.13	together with all these precursors
00:05:53.26	in a test tube,
00:05:55.21	you see 6-Deoxyerythronolide B.
00:05:58.24	And what I show you on the lower right
00:06:01.07	is a mass spectrum of the product
00:06:04.00	that has been synthesized
00:06:05.29	in a biochemical equivalent
00:06:08.02	of an earth/air/water/fire type
00:06:10.19	of an experiment.
00:06:12.16	What this allows us now to do
00:06:14.24	is to probe this machine
00:06:17.05	with all the power
00:06:19.19	that you're used to using
00:06:21.24	to study your favorite enzyme
00:06:24.27	once you've purified it to homogeneity.
00:06:28.06	So what I show you in this
00:06:30.13	is a very simple graph
00:06:33.05	that gives you a sense
00:06:35.03	that we can turnover
00:06:36.26	this entire assembly line
00:06:38.27	in a test tube,
00:06:40.18	with a rate constant
00:06:42.15	that's approximately about 1/min.
00:06:47.10	So, approximately once every minute
00:06:49.17	this assembly line is releasing
00:06:51.21	6-Deoxyerythronolide B
00:06:53.22	in a test tube that is presented with the appropriate precursors,
00:06:57.27	and that is roughly the rate
00:06:59.24	we might expect this assembly line
00:07:01.18	to be working at
00:07:03.14	inside a cell.
00:07:05.06	I also wanna point out,
00:07:06.24	as the inset shows,
00:07:08.21	this assay is remarkably efficient.
00:07:12.29	Every equivalent of 6-Deoxyerythronolide B
00:07:16.28	has stoichiometric mapping
00:07:19.17	to an equivalent
00:07:21.26	of the propionyl-CoA primer
00:07:24.01	that is used,
00:07:25.28	and uses six equivalents of NADPH,
00:07:29.08	whose consumption is being measured
00:07:31.05	in this simple spectrophotometric assay.
00:07:35.07	Okay, so we have to tools to be able to study
00:07:39.04	the entire assembly line
00:07:40.25	inside a recombinant E. coli-like cell.
00:07:44.06	We have the ability
00:07:46.00	to study the entire assembly line
00:07:48.01	in a purified, reconstituted form.
00:07:51.03	We also have the ability, today,
00:07:53.21	to study the individual steps
00:07:56.12	in the catalytic cycle of these modules
00:08:01.09	in isolation.
00:08:03.10	So, recall in the previous module,
00:08:05.29	I introduced you to some of the
00:08:08.12	core reactions
00:08:10.18	that occur at every module.
00:08:13.00	There's a reaction
00:08:14.19	that we call chain translocation,
00:08:16.24	where the chain moves
00:08:18.08	from the acyl carrier protein
00:08:20.04	in the upstream module
00:08:21.26	to the module that's receiving the chain,
00:08:25.09	and if we want to interrogate
00:08:27.07	just that reaction
00:08:29.15	for one module,
00:08:31.09	what we do is pull out that module
00:08:34.07	from the rest of the assembly line,
00:08:36.17	purify that to homogeneity,
00:08:39.23	present it with
00:08:43.01	a chemoenzymatically-derived acyl carrier protein
00:08:47.00	that has the growing polyketide chain substrate
00:08:51.01	bound to it,
00:08:53.03	and we put it into a test tube
00:08:55.19	so that the chain translocation event
00:08:58.13	-- the movement of that growing polyketide chain
00:09:01.23	into the module --
00:09:03.26	is the slow kinetic step,
00:09:06.00	and everything after that
00:09:08.03	that leads to the turnover of this module
00:09:10.21	is fast.
00:09:12.08	And so you can use
00:09:14.07	this kind of an assay
00:09:15.28	to interrogate,
00:09:18.01	using established kinetic paradigms,
00:09:20.23	that chain translocation step
00:09:23.10	of your favorite module
00:09:25.12	that you're interested in.
00:09:27.22	The same approach can also be used
00:09:31.03	to kinetically isolate the chain elongation event
00:09:34.11	that I introduced you to
00:09:36.03	in the earlier lecture.
00:09:38.18	So when we want to study chain elongation,
00:09:42.03	what we do is we take the module
00:09:44.22	whose elongation biochemistry
00:09:46.19	we want to study,
00:09:48.19	and we prepare just the ketosynthase
00:09:51.03	together with the acyltransferase
00:09:54.08	from that module
00:09:56.02	as one protein.
00:09:57.21	We produce its carrier protein,
00:10:00.04	its acyl carrier protein,
00:10:01.29	as another protein.
00:10:03.27	We put these two proteins together,
00:10:06.21	we present the two substrates
00:10:09.24	into this assay,
00:10:12.05	and we look for chain elongation,
00:10:15.00	which gives rise to the product.
00:10:17.19	And we do this under conditions
00:10:20.06	where the step we wanna probe,
00:10:22.07	the elongation step,
00:10:23.27	is the slow step, and everything else is fast.
00:10:27.20	You can do exactly the same thing
00:10:29.21	to probe the acyl transfer,
00:10:31.29	the selection of that building block
00:10:34.16	-- methylmalonyl coenzyme A-derived building block
00:10:38.12	from metabolism --
00:10:40.08	the same approach can also work over there.
00:10:42.20	And all of these assays are well-developed,
00:10:44.29	they're in the literature,
00:10:46.19	and you can use them to study
00:10:48.08	your favorite assembly line.
00:10:51.07	In addition to those core reactions
00:10:53.28	-- chain translocation,
00:10:55.21	chain elongation,
00:10:57.08	and acyl transfer --
00:10:58.23	I mentioned there are auxiliary reactions,
00:11:01.13	which I lumped under chain modification.
00:11:04.27	Those reactions include
00:11:06.22	ketoreductase-types of chemistries.
00:11:10.11	In this particular assay,
00:11:12.11	I'm adding the ketoreductase,
00:11:15.03	or KR,
00:11:16.26	as a stand-alone protein,
00:11:18.16	to the rest of my system
00:11:21.04	so that I can control the rate
00:11:22.27	at which that step occurs,
00:11:24.27	and I can look at the consequences
00:11:27.13	of putting one ketoreductase in my assay
00:11:30.04	as opposed to some other ketoreductase.
00:11:33.04	And that allows me to interrogate
00:11:35.28	the ketoreductase reaction.
00:11:38.00	You can do the same thing
00:11:40.04	at the level of the dehydratase reaction,
00:11:43.15	which follows after the ketoreductase reaction
00:11:47.29	in certain chain modification sequences.
00:11:52.00	So, all of these assays are also set up.
00:11:54.22	The point you need to recognize is that
00:11:57.13	you can probe through, again,
00:11:59.04	a divide-and-conquer approach,
00:12:01.00	the chemistry happening at any one of these steps
00:12:05.05	in the overall assembly line.
00:12:08.22	Using this combination of in vivo and in vitro tools,
00:12:14.02	there are a number of important problems
00:12:16.15	you can study.
00:12:17.28	In the remainder of this second module lecture,
00:12:21.16	I will talk to you about some examples
00:12:24.13	of questions, long-standing questions
00:12:26.26	in the field,
00:12:28.04	having to do with the specificity
00:12:29.28	of these assembly lines.
00:12:31.20	I'll give you two examples of those problems
00:12:33.22	because they have engineering implications.
00:12:36.21	And then in the next lecture we'll talk about
00:12:40.08	the assembly line mechanisms.
00:12:42.27	So, stereospecificity
00:12:46.01	is probably one of the most fascinating features
00:12:50.15	of these complex polyketide antibiotics
00:12:53.11	that these assembly lines make.
00:12:55.24	So, to the right,
00:12:57.14	you're seeing the 6-Deoxyerythronolide B product
00:13:01.16	of DEBS,
00:13:03.07	and for those of you who are looking at that now,
00:13:05.22	you're noticing that it has 10 stereocenters.
00:13:12.02	That is 2^10 possible chiral forms
00:13:17.08	of the same chemical formula,
00:13:20.01	or slightly more than 1000
00:13:22.12	of these chiral forms.
00:13:24.20	If you go to a fermentation plant
00:13:27.10	that makes erythromycin,
00:13:30.08	the large vat that produces erythromycin
00:13:33.23	has one out of those 1000+
00:13:38.04	stereochemical forms in it;
00:13:40.17	the one that I'm showing you.
00:13:43.00	I think most of you would recognize
00:13:45.01	that that is a really impressive feat
00:13:48.04	on the part of nature...
00:13:49.27	how it can program this assembly line
00:13:52.06	to give one, and only one
00:13:54.13	stereochemical outcome.
00:13:56.22	That is a problem
00:13:58.29	that we have quite a good understanding of
00:14:02.16	how that happens today.
00:14:05.08	I've cited some references on this slide,
00:14:09.04	and so I will summarize for you
00:14:11.07	what these references teach us
00:14:13.18	about how stereochemistry is controlled
00:14:17.07	by the DEBS assembly line.
00:14:20.00	So, of those 10 stereocenters,
00:14:25.00	one of them,
00:14:27.12	which is this stereocenter,
00:14:31.06	is generated by
00:14:34.28	this ketoreductase
00:14:37.28	in Module 3 of the assembly line,
00:14:40.19	that I introduced to you as that epimerase,
00:14:44.29	that looks like a ketoreductase
00:14:46.29	in my previous lecture.
00:14:49.06	This is the enzyme that is a homologue
00:14:52.02	of other ketoreductases,
00:14:53.18	but does no NADPH-dependent chemistry.
00:14:57.27	Instead, it epimerizes the C2 carbon atom
00:15:01.20	of the growing polyketide chain
00:15:04.28	that is lodged in Module 3
00:15:06.25	of the assembly line.
00:15:09.05	Of the remaining 9 stereocenters,
00:15:12.07	8 of those stereocenters
00:15:14.22	are shown in red,
00:15:17.09	and they are controlled
00:15:19.21	by the 3 red ketoreductases
00:15:23.02	and 1 blue ketoreductase
00:15:25.17	in Modules 1, 2, 5, and 6, respectively.
00:15:32.23	So, each of these 4 ketoreductases
00:15:37.25	controls 2 stereocenters apiece.
00:15:42.20	For the chemically initiated,
00:15:44.29	these enzymes are not just stereoselective,
00:15:48.22	they're also diastereoselective;
00:15:51.17	so they're setting 2 stereocenters at a time.
00:15:56.01	And these enzymes, we know...
00:15:58.23	these 4 enzymes we know, today,
00:16:01.13	are both necessary and sufficient
00:16:04.10	for the unique labeling...
00:16:08.05	for the unique identification
00:16:11.01	of those stereocenters.
00:16:13.21	The last stereocenter,
00:16:16.01	which is this stereocenter,
00:16:19.13	is at the 6 position,
00:16:22.01	is a more complex output,
00:16:25.08	and it is generated by 3 enzymes
00:16:27.28	in Module 4 of DEBS.
00:16:30.28	There is a ketoreductase,
00:16:34.03	a dehydratase,
00:16:36.04	and an enoylreductase,
00:16:38.08	that all collaborate with each other
00:16:41.06	to set this one stereocenter.
00:16:47.06	In addition to stereochemistry,
00:16:49.20	there is another very important
00:16:53.15	specificity that is encoded
00:16:56.11	in this assembly line
00:16:58.22	at each module,
00:17:00.24	and that is the specificity
00:17:05.17	that corresponds to the choice
00:17:08.19	of the extender unit.
00:17:10.19	In my introduction,
00:17:12.07	I pointed out that all of the modules
00:17:14.26	of 6-Deoxyerythronolide B synthase
00:17:18.04	use a methylmalonyl coenzyme A
00:17:21.14	extender unit.
00:17:23.22	In the case of DEBS,
00:17:27.06	the R group that is shown
00:17:29.25	in this enzymatic scheme
00:17:32.07	would be a methyl group.
00:17:34.17	Other polyketide synthases
00:17:37.06	can use coenzyme A thioesters
00:17:39.29	that contain other functional groups
00:17:42.16	in place of a methyl group, over here.
00:17:46.11	And all of these choices
00:17:48.18	are made by the acyltransferase.
00:17:53.00	These acyltransferases
00:17:55.10	are relatively specific.
00:17:58.01	Not only do they have high specificity,
00:18:01.04	as shown in this graph
00:18:03.15	for the coenzyme A precursor
00:18:05.28	they're picking from the metabolic soup...
00:18:09.09	so what you see in this graph over here
00:18:13.06	is the rate...
00:18:15.24	the velocity versus substrate concentration
00:18:18.20	of the preferred substrate,
00:18:20.20	which is methylmalonyl coenzyme A,
00:18:23.22	and down here are the rates
00:18:25.25	if R is one methyl short,
00:18:28.13	so in other words it's a hydrogen instead of a methyl,
00:18:31.14	or one methyl longer,
00:18:33.23	which is an ethyl group.
00:18:35.29	And as you can tell,
00:18:37.25	this acyltransferase
00:18:39.18	that we're showing you data for in this slide
00:18:42.08	is highly selective
00:18:44.24	for a methyl group
00:18:46.25	instead of one smaller or one larger.
00:18:50.16	Now, in addition to being specific
00:18:52.25	for its cognate substrate,
00:18:56.09	this enzyme is also specific
00:18:59.11	for its protein partner,
00:19:01.09	which is the acyl carrier protein,
00:19:03.22	that is being used.
00:19:05.25	And here I'm introducing you
00:19:07.25	to a concept that is gonna come back
00:19:10.11	in a more significant way
00:19:12.13	in the last of my three lectures,
00:19:14.16	which is the importance
00:19:16.10	of protein-protein interactions
00:19:19.04	in the assembly line biochemistry
00:19:22.00	of these systems.
00:19:23.22	In this case, what you're seeing is
00:19:27.03	that the acyl carrier protein
00:19:29.21	is being strongly recognized
00:19:32.00	by the acyltransferase,
00:19:34.05	because if you give this same acyltransferase
00:19:37.18	other acyl carrier proteins
00:19:39.25	from other modules of DEBS
00:19:41.23	or elsewhere,
00:19:43.19	they work much more poorly
00:19:46.09	than the natural acyl carrier protein.
00:19:49.19	So in addition to recognizing
00:19:51.15	the coenzyme A precursor,
00:19:53.27	you also have recognition
00:19:56.00	of the acyl carrier protein.
00:19:58.18	Now, for those of you who are familiar
00:20:00.16	with enzymes kinetics
00:20:02.16	know that from data like this,
00:20:04.18	you can derive mechanisms
00:20:06.10	of how these enzymes work.
00:20:08.15	So, in this case,
00:20:10.19	this acyltransferase
00:20:13.02	has a ping-pong
00:20:15.03	bi-bi-type of a mechanism.
00:20:18.11	The coenzyme A precursor first comes in,
00:20:21.21	it is bound by the acyltransferase,
00:20:25.05	the acyltransferase picks the methylmalonyl extender unit,
00:20:29.02	coenzyme A leaves,
00:20:30.23	the carrier protein comes in,
00:20:32.29	is recognized by the acyltransferase,
00:20:35.25	and takes away the product
00:20:38.06	- the methylmalonyl extender unit.
00:20:40.23	So that ping-pong element comes into
00:20:43.12	this kind of a mechanism.
00:20:45.12	Now, you can also ask,
00:20:46.29	in addition to these gatekeeper acyltransferases
00:20:50.25	that control the choice of the building block,
00:20:55.18	there are many other enzymes
00:20:57.18	in this assembly line
00:20:59.17	that lie downstream of each choice
00:21:02.15	that a module makes
00:21:04.16	of its building block.
00:21:06.09	To what extent do they influence
00:21:09.05	the overall substrate specificity?
00:21:11.29	Do they care about what the upstream module chose
00:21:16.17	as its precursor for elongating
00:21:20.01	the growing polyketide chain?
00:21:22.13	We can ask questions like that
00:21:24.16	using the assays... biochemical assays
00:21:27.11	I showed you earlier on,
00:21:29.16	and from those experiments you learn something
00:21:31.22	quite interesting.
00:21:33.16	So, you learn that the downstream steps,
00:21:36.14	beyond the acyltransfer step,
00:21:38.28	in many modules
00:21:41.12	are not that discriminatory,
00:21:43.09	analogous to what Henry Ford
00:21:45.11	had contemplated for his assembly line.
00:21:48.26	The downstream steps
00:21:51.01	have low, but not a significant amount,
00:21:54.16	of specificity.
00:21:56.21	So if, by whatever mechanism,
00:21:59.14	you can fool this acyltransferase
00:22:02.25	to put a hydrogen instead of a methyl
00:22:06.16	at this position on the carrier protein,
00:22:09.26	the elongation enzyme
00:22:12.06	that elongates the chain
00:22:14.10	and puts this R group in the growing polyketide chain,
00:22:19.14	primarily loses
00:22:22.12	about 2- to 4-fold specificity
00:22:25.10	as a result of this mistake
00:22:28.13	that the upstream step made.
00:22:31.00	That's not much in the grand scheme of things,
00:22:33.25	but what you have to remember is
00:22:37.02	these assembly lines have
00:22:39.18	many, many downstream steps.
00:22:41.27	So, one of these assembly lines,
00:22:43.23	the first module over here, or the second module,
00:22:46.28	has four or five modules downstream
00:22:50.17	that are looking at the consequences
00:22:53.01	of what that module did.
00:22:55.00	And so these small effects
00:22:57.09	at the level of substrate specificity
00:22:59.23	then have quite significant impact
00:23:02.21	on the final product,
00:23:04.18	and you can see this
00:23:06.18	in the context of assays like this.
00:23:09.06	So here,
00:23:10.24	in the spirit of engineering,
00:23:12.17	what we're doing is we're taking that natural...
00:23:15.17	the natural assembly line that nature uses
00:23:18.04	to make 6-Deoxyerythronolide B,
00:23:20.27	we're taking that full assembly line,
00:23:23.00	and instead of just presenting it
00:23:25.27	methylmalonyl coenzyme A,
00:23:28.14	we're now presenting it a 1:1 mixture
00:23:32.06	of methylmalonyl coenzyme A
00:23:34.11	and ethylmalonyl coenzyme A,
00:23:36.28	and we're asking what's gonna happen.
00:23:40.11	Are you gonna get just 6-Deoxyerythronolide B?
00:23:44.19	Are you gonna get something else,
00:23:47.10	one or two other products?
00:23:49.06	Or are you gonna get a zoo of products?
00:23:51.26	And the answer to that is,
00:23:53.28	you get some analogues
00:23:56.23	that are produced competitively
00:23:58.25	with the natural product 6-DEB,
00:24:02.23	but these compounds
00:24:05.24	aren't immediately obvious
00:24:08.00	why these should be formed
00:24:10.02	and other ones shouldn't be formed.
00:24:11.29	So at least one of these, for example,
00:24:13.27	this peak that I show you out here,
00:24:16.04	whose mass spectrum is shown over here,
00:24:18.22	we know with reasonable confidence,
00:24:21.06	has an ethyl group
00:24:23.07	that is incorporated at the C8 position
00:24:26.12	instead of a methyl group.
00:24:28.06	So, somehow,
00:24:30.10	that gatekeeper transferase
00:24:32.29	has enough tolerance
00:24:35.04	for an ethylmalonyl extender unit,
00:24:37.24	and all of the downstream enzymes
00:24:40.13	on the assembly line
00:24:42.11	are sufficiently tolerant
00:24:44.13	that they will let that mistake slide by,
00:24:46.28	so you get this desired product.
00:24:50.05	And if we could predict
00:24:52.10	what's gonna be made and what's not gonna be made
00:24:55.01	through an experiment like this,
00:24:57.05	why, then, we would have a way to precisely engineer
00:24:59.21	an antibiotic like erythromycin
00:25:01.21	to make a molecule like this.
00:25:03.27	But right now, we're just beginning to scratch
00:25:06.10	the tip of the iceberg
00:25:07.22	in terms of what's possible,
00:25:09.13	and what's not,
00:25:11.04	in these kinds of systems,
00:25:12.21	and an experiment like this illustrates
00:25:14.22	what's possible.
00:25:16.28	These kinds of insights can also be used
00:25:19.12	for most sophisticated engineering experiments,
00:25:22.07	analogous to the kinds of experiments
00:25:24.17	you may be familiar with
00:25:26.25	when you think about incorporating unnatural amino acids
00:25:31.18	in proteins that are derived
00:25:33.24	by ribosomal mechanisms.
00:25:36.04	And so, in this,
00:25:38.08	there are some examples of acyltransferases
00:25:42.10	that are what we call stand-alone acyltransferases.
00:25:46.12	They operate outside the assembly line,
00:25:49.21	and because they work
00:25:51.29	very, very fast compared to typical assembly line acyltransferases
00:25:58.03	that exist in assembly lines,
00:26:00.16	you can do some interesting experiments.
00:26:02.25	So in this case,
00:26:04.24	we have knocked out
00:26:07.10	one acyltransferase
00:26:09.17	out of all the six acyltransferases
00:26:12.09	on the 6-Deoxyerythronolide B synthase
00:26:16.23	-- so this is like a site-directed mutant
00:26:19.14	that has inactivated that acyltransferase --
00:26:23.02	and we complement it,
00:26:25.15	in trans,
00:26:27.26	with that very ultra-fast acyltransferase
00:26:30.27	that we get from a different assembly line
00:26:34.02	in nature.
00:26:36.07	And what happens is,
00:26:38.00	because this acyltransferase
00:26:40.09	will pick a malonyl unit instead of a methylmalonyl unit,
00:26:44.13	it can transfer that malonyl unit
00:26:47.07	onto this module,
00:26:49.02	because this module is simply waiting
00:26:50.26	for something to come its way,
00:26:52.29	and this enzyme can do it pretty quickly,
00:26:56.18	and now that resulting intermediate
00:26:58.27	moves all the way down in the assembly line
00:27:01.23	to give you a product
00:27:04.09	that has one and only one change
00:27:07.11	in the entire macrocycle
00:27:10.09	that is made.
00:27:12.08	And what you see over here is,
00:27:13.29	if I have my assembly line that is present
00:27:18.15	at say micromolar concentrations in my assay,
00:27:22.00	at nanomolar concentrations
00:27:24.26	I can get quite respectable incorporation
00:27:29.22	of malonyl coenzyme A
00:27:31.18	at that single position,
00:27:33.16	to give me 12-desmethyl-6-Deoxyerythronolide B.
00:27:39.25	So this is a way you can cheat the system,
00:27:43.26	so long as you have
00:27:46.19	a robust enzyme
00:27:49.00	that can trans-complement the module
00:27:51.15	that you wanna cheat.
00:27:53.22	So hopefully that gives you
00:27:55.24	a flavor of the kinds of tools we have
00:27:58.28	and the way we use these tools
00:28:01.01	to study specificity.
00:28:03.12	Thank you.