Session 8: Human Evolution

Transcript of Part 3: African Genomics: Natural Selection

00:00:07.17	For the last part of my lecture series,
00:00:10.11	I wanna talk about examples of natural selections in humans,
00:00:14.29	and the two particular examples
00:00:17.01	that I'm going to be talking about
00:00:19.00	are the evolution or lactose tolerance in east Africa,
00:00:22.10	and of pygmy short stature.
00:00:25.04	So if we're going to be talking about natural selection,
00:00:27.11	we have to first of course
00:00:28.29	acknowledge Charles Darwin,
00:00:31.12	who came up with the theory of natural selection.
00:00:36.10	In fact, to quote from Darwin, he said,
00:00:39.13	"This preservation of favourable variations
00:00:42.03	and the rejection of injurious variations,
00:00:44.21	I call Natural Selection.
00:00:47.06	Variations neither useful nor injurious
00:00:49.27	would not be affected by natural selection,
00:00:52.18	and would be left a fluctuating element,
00:00:55.01	as perhaps we see in the species called polymorphic."
00:00:58.10	And that was from his classic book
00:01:00.06	On The Origin of Species,
00:01:01.24	published in 1859,
00:01:03.25	and you might recognize from our first lecture,
00:01:07.03	that this is really talking about genetic drift,
00:01:09.27	random fluctuations.
00:01:13.13	However, part of the evolutionary change that we see
00:01:18.12	is not just going to be due to random genetic drift,
00:01:21.03	it's also going to be due to natural selection.
00:01:24.18	And so, according to that theory,
00:01:27.10	natural variation exists and is heritable,
00:01:30.02	more organisms are born than can survive,
00:01:32.11	and therefore organisms best suited to the environment
00:01:35.07	survive more often,
00:01:36.25	and slight differences can accumulate in a species over time.
00:01:40.24	So this is the idea of gradual evolution of a species
00:01:43.27	by natural selection.
00:01:45.18	And this is Huxley,
00:01:47.03	who was also known as Darwin's bulldog
00:01:50.11	because he was the big proponent of his theory,
00:01:52.17	and he said,
00:01:54.05	"How extremely stupid not to have thought of that!"
00:01:57.12	So when Darwin first came up with his theory of natural selection,
00:02:01.08	there was really no concept of genetics
00:02:04.12	as we know it today.
00:02:06.02	In fact, it wasn't until the late 1800s
00:02:08.12	that Mendel proposed his theory of genetics.
00:02:13.02	So in the 1930s and 1940s
00:02:15.22	there was sort of a synthesis of natural selection
00:02:19.09	and genetics and mathematics,
00:02:22.16	population genetics,
00:02:23.25	and at that time it was proposed that genetic variation in populations
00:02:27.05	arises by chance through mutation and recombination,
00:02:31.15	that evolution consists primarily of changes in the
00:02:34.12	frequencies of alleles between one generation and another,
00:02:37.25	largely as a result of genetic drift,
00:02:40.24	gene flow,
00:02:42.05	and natural selection.
00:02:43.27	And that speciation occurs gradually when populations
00:02:46.00	are reproductively isolated, for example,
00:02:48.20	by geographic barriers.
00:02:52.14	And so if we look at this timeline,
00:02:54.12	starting with the Origin of Species,
00:02:56.21	and then Mendelian inheritance
00:02:59.04	is actually rediscovered in 1900,
00:03:01.19	it was first proposed in the late 1880s,
00:03:04.01	but very few people knew about it at that time.
00:03:06.27	And then in the early 1900s
00:03:09.09	we have the theoretical foundations of population genetics
00:03:12.21	and then, as I mentioned,
00:03:14.09	the modern synthesis in the 30s.
00:03:16.19	And then in the 70s we have Kimura's theory of neutral evolution,
00:03:21.19	which was proposing that most changes and speciation events
00:03:25.07	are simply due to random genetic drift
00:03:27.22	and to new mutation events.
00:03:29.20	And I think that today we would say
00:03:31.14	it's a combination of all of the above.
00:03:33.19	There's certainly a lot of genetic drift that occurs,
00:03:36.02	but we know that natural selection
00:03:37.29	is having a very important influence
00:03:40.26	on the variation that we see
00:03:43.05	in terms of phenotypic variation and even disease susceptibility.
00:03:48.04	So let's look what happens
00:03:49.08	when a neutral mutation occurs in a population,
00:03:52.07	as indicated by this individual in green.
00:03:55.25	Let's look what happens as we proceed forward in generations,
00:03:59.08	and you can see there's not too many changes
00:04:01.21	in allele frequency.
00:04:03.29	But what happens when we have a beneficial mutation,
00:04:09.05	which means that it increases the fitness of the individual,
00:04:12.20	meaning that they're more likely to produce children,
00:04:16.12	and their children are more likely to produce more children,
00:04:19.00	and so on and so forth.
00:04:21.15	And so we can see that each generation,
00:04:24.04	this beneficial mutation is going to spread,
00:04:27.22	until eventually it may be nearly fixed
00:04:32.06	in the population.
00:04:34.17	So I want to tell you about some of our studies
00:04:37.20	focused in African populations
00:04:39.14	in which we're trying to identify
00:04:41.02	genetic signatures of natural selection,
00:04:43.19	and regions of the genome that are targets of natural selection.
00:04:47.29	And this is important
00:04:49.22	because it's thought that mutations associated with diseases
00:04:52.16	in modern populations,
00:04:54.13	like hypertension
00:04:56.03	, diabetes,
00:04:57.08	obesity,
00:04:58.10	and asthma,
00:04:59.08	may have been selectively advantageous or adaptive
00:05:01.23	in past hunter-gatherer environments.
00:05:04.04	So if we can identify these regions
00:05:06.25	that are targets of selection, or actual variable sites
00:05:09.18	that are targets of selection,
00:05:11.16	those may be functionally important
00:05:13.14	and may give us a clue about disease risk.
00:05:16.11	So here I'm showing you a few of the populations
00:05:18.12	that we've studied in Africa,
00:05:20.23	and we have people who are living at very different climates,
00:05:23.15	high altitude, low altitude,
00:05:26.03	savannah, and tropical environments, for example.
00:05:30.10	We have people who have very different diets,
00:05:32.22	so agriculturalists,
00:05:34.08	hunter-gatherers,
00:05:35.23	or pastoralists.
00:05:37.14	And they have very different infectious disease exposures,
00:05:40.05	so they've likely undergone local adaptation
00:05:42.13	to different environments.
00:05:45.25	And I'm going to, as I mentioned,
00:05:47.16	tell you about two examples today.
00:05:49.15	The first one is the evolution of lactose tolerance
00:05:51.29	in east African pastoralist populations.
00:05:57.07	So, the ability to digest the sugar lactose,
00:06:00.21	which is quite common in milk,
00:06:03.07	is due to an enzyme called lactase-phlorizine hydrolase,
00:06:07.15	or known as lactase for short.
00:06:09.28	And lactase is expressed specifically
00:06:13.01	in the brush border cells of the small intestine,
00:06:16.25	and in individuals who maintain high levels of this enzyme
00:06:20.29	as adults,
00:06:22.23	they're able to break down the complex sugar lactose
00:06:26.16	into glucose and galactose,
00:06:29.14	which is rapidly taken up into the bloodstream.
00:06:32.23	However,
00:06:35.19	most mammals, and most humans,
00:06:38.19	shut down lactase activity
00:06:40.23	shortly after weaning.
00:06:42.28	So, as adults, they do not have an active form of this enzyme.
00:06:46.24	And what's going to happen is
00:06:48.19	they're not going to be able to break down that complex sugar.
00:06:51.26	It's going to go down into the lower gut,
00:06:54.13	it's going to be attacked by bacteria,
00:06:56.25	and you're going to have severe intestinal distress.
00:07:01.00	Now, it has been noted for many years by anthropologists
00:07:04.27	that there is a very strong correlation
00:07:06.27	between the lactose tolerance trait,
00:07:09.19	or you could think of it also as the lactase persistence trait,
00:07:13.26	because there's persistence of the enzyme activity as adults.
00:07:18.15	And they've seen a strong correlation
00:07:20.20	between the prevalence of that trait
00:07:23.14	with populations who traditionally practice cattle domestication
00:07:28.04	and dairying.
00:07:30.05	So for example, this trait is most common in northern Europe,
00:07:33.19	it decreases in frequency as one moves
00:07:36.23	into southern Europe
00:07:38.29	and into the Middle East.
00:07:40.24	It's very uncommon in eastern Asia
00:07:43.26	and in the Americas,
00:07:46.13	and it's uncommon in western Africa,
00:07:48.26	which is one of the reasons that we see high levels
00:07:51.17	of lactose intolerance in African Americans, for example.
00:07:55.24	But in regions of Africa where there's a high prevalence
00:07:59.04	of cattle domestication, pastoralism, and dairying,
00:08:03.14	we see a high prevalence of this trait.
00:08:07.15	So, in 2002,
00:08:10.18	there was an elegant study done
00:08:12.20	by Leena Peltonen's group in Finland,
00:08:14.28	in which they identified a genetic mutation
00:08:17.08	that regulates lactose tolerance in Europeans.
00:08:20.20	And it was located near the...
00:08:23.25	upstream of the lactase gene.
00:08:26.01	When we sequenced that region in east African pastoralists,
00:08:29.04	they didn't have it,
00:08:31.10	so we knew they must have something else.
00:08:33.13	So in order to identify those mutations,
00:08:35.21	we did something that's called a lactose tolerance test.
00:08:38.29	So, basically what we do is
00:08:42.11	we give people the sugar lactose in a powdered form,
00:08:46.20	we add water, and it basically tastes like orange Kool-Aid,
00:08:51.09	and then we have to line people up
00:08:54.17	and have them drink the lactose at the same time.
00:08:57.27	This is a group of Maasai women from Tanzania.
00:09:03.28	This is a group of pastoralists from southern Ethiopia.
00:09:11.23	And then we can use a standard diabetes monitoring kit,
00:09:16.03	and what we can do is to measure the blood glucose,
00:09:19.29	starting at baseline before they drink the lactose,
00:09:23.29	and then every 20 minutes we're gonna measure this,
00:09:27.06	over a period of about an hour.
00:09:30.02	And then we're gonna look at the maximum rise
00:09:32.25	in blood glucose.
00:09:35.15	If individuals have a rise
00:09:37.09	that is greater than 1.7 millimolar (mM)
00:09:39.20	we consider them to be lactose tolerant,
00:09:42.20	or to have the lactase persistent trait,
00:09:45.03	shown in light blue.
00:09:47.07	And if they have a rise that is less than 1.1 mM,
00:09:51.12	they're considered to be intolerant,
00:09:53.25	shown in dark blue.
00:09:55.18	So, we measured this trait
00:09:57.12	in nearly 500 individuals
00:09:59.17	from Tanzania, Kenya, and the Sudan,
00:10:02.00	and then we looked for association
00:10:04.10	with genetic variation that we identified
00:10:06.21	by resequencing the region
00:10:08.28	where the European variant had been identified.
00:10:13.13	And in doing so we identified
00:10:15.12	three novel genetic polymorphisms
00:10:18.21	that are associated with the lactose tolerance trait in east Africa,
00:10:22.17	and those are shown here by the boxes.
00:10:26.29	The most common was this one at position 14010,
00:10:31.06	but we also saw those others
00:10:32.24	at positions 13915 and 13907,
00:10:36.03	located roughly 14,000 basepairs
00:10:38.25	upstream of the lactase gene
00:10:41.18	which is located on chromosome 2.
00:10:44.07	Now, one of the really interesting things about this is that,
00:10:48.11	one, these regulatory mutations were pretty far away,
00:10:51.28	about 14,000 basepairs from the gene,
00:10:54.26	and they were located in an intron
00:10:57.25	in a non-coding region of a neighboring gene called MCM6.
00:11:03.00	So this is demonstrating that
00:11:04.25	functionally important variation
00:11:07.13	can actually be located in non-coding regions,
00:11:10.21	and we were able to show,
00:11:13.13	using in vitro cell line studies,
00:11:16.20	that these variants that are derived,
00:11:20.19	shown in the different colors here,
00:11:23.22	that they regulate expression
00:11:26.15	of the lactase gene using the lactase promoter.
00:11:31.10	Now, they're located very close to the mutation
00:11:35.01	associated with lactose tolerance in Europeans,
00:11:38.20	located at position 13910,
00:11:41.14	but they arose independently
00:11:43.17	due to a process called convergent evolution,
00:11:46.13	and probably due to a very strong
00:11:48.27	selective force to be able to drink milk that contains lactose,
00:11:56.15	in these different regions of the world.
00:12:00.27	What's also interesting
00:12:02.19	is that the variants that we identified
00:12:04.16	have a very distinct geographic distribution.
00:12:07.09	So the one that we found that was most common in our study
00:12:10.06	was at position 14010,
00:12:12.08	and we can see that it is pretty localized
00:12:15.01	to east Africa, to Tanzania and Kenya,
00:12:17.26	and that's the most likely site of origin of that mutation.
00:12:21.11	Interestingly, we also see it a bit in south Africa,
00:12:26.01	probably reflecting migration of pastoralists
00:12:28.29	from east Africa into that region.
00:12:32.16	The variant position at 13915
00:12:35.08	appears to have originated in the Middle East,
00:12:37.21	and we could see that it was introduced into northeast Africa,
00:12:40.17	probably by migration.
00:12:43.10	And then the variant at position 13907
00:12:46.26	likely arose in northeast Africa.
00:12:49.21	But again, one of the important take-home points is that
00:12:53.02	we have a functionally important variant
00:12:55.07	that's occurring at high frequency, sometimes as high as 40%,
00:12:59.12	and it's very geographically restricted,
00:13:02.22	and there are likely to be other mutations like that,
00:13:05.06	some of which may have implications for disease susceptibility,
00:13:09.11	again emphasizing the importance
00:13:11.23	to look amongst ethnically diverse Africans.
00:13:16.29	So the next thing we wanted to do
00:13:19.21	was to look for a signature of positive selection,
00:13:23.17	and this is the method in which we can do that.
00:13:27.21	So imagine, here in red,
00:13:30.25	imagine that this is a new mutation that has occurred, say,
00:13:34.15	one of the mutations associated with lactose tolerance.
00:13:38.00	And it's adaptive,
00:13:39.10	meaning that it increases the fitness of individuals who have it,
00:13:43.25	meaning that they're more likely to have children,
00:13:45.27	and their children are more likely to have children,
00:13:47.22	and so on.
00:13:49.28	And so it's going to increase in frequency
00:13:52.24	in the population,
00:13:54.29	and it's going to drag with it
00:13:57.15	the neighboring variants nearby.
00:14:00.03	So, you can see that when it originated, it had...
00:14:02.29	it was on a chromosome with a green variant
00:14:05.18	and a black variant.
00:14:08.03	And now these got dragged along to high frequency,
00:14:11.04	through a process known as hitchhiking.
00:14:14.07	Now, if this had gone to fixation,
00:14:17.12	meaning that everybody has it,
00:14:19.00	we would have called it a full selective sweep.
00:14:21.20	In this case, it hasn't quite reached a full selective sweep,
00:14:26.15	so we call it a partial sweep.
00:14:29.24	Now, that could just mean that
00:14:31.17	there hasn't been time for it to go to a full sweep,
00:14:33.06	or it could be that for some reason
00:14:35.17	there may be some negative aspects of having it,
00:14:38.02	and there's a reason that both variants are maintained in the population.
00:14:43.17	Now, after the sweep occurs,
00:14:45.19	you're going to have new mutation events
00:14:47.26	and new recombination events
00:14:49.21	shuffling up the variants
00:14:52.04	that are linked to the mutation that's adaptive.
00:14:56.12	And so that will decrease the association
00:15:01.00	observed between the mutation and the flanking variation.
00:15:05.00	And in fact,
00:15:06.15	if we have an estimate of the recombination rate,
00:15:08.27	we can use computational methods
00:15:10.24	to estimate how old this mutation is.
00:15:14.18	And that's exactly what we did here.
00:15:17.12	So shown on top
00:15:19.28	is an example from the most common mutation
00:15:22.27	that we found associated with lactose tolerance,
00:15:25.03	at position 14010.
00:15:27.18	Individuals who have the C variant
00:15:29.26	are able to digest milk,
00:15:31.22	and individuals who are homozygous are shown as red.
00:15:35.28	And what we did is we genotyped markers
00:15:38.28	going a distance of about 3 million nucleotides,
00:15:42.26	and what we would do is that if someone is homozygous,
00:15:46.20	starting at the lactose tolerance mutation,
00:15:49.02	and then we go to the next mutation.
00:15:51.02	If they're homozygous,
00:15:53.00	then we continue going.
00:15:55.13	If they underwent a recombination,
00:15:57.05	we stop the line.
00:15:59.13	And what we can basically see is that homozygosity
00:16:02.28	extends about 2 million basepairs
00:16:06.01	on chromosomes that have the lactose tolerance mutation.
00:16:09.15	But if we look at chromosomes that have the ancestral mutation,
00:16:14.00	they have almost no extended haplotype homozygosity.
00:16:19.07	And so this is a classic signature of a selective sweep.
00:16:22.25	It means that this variant
00:16:24.16	was under very strong positive selection
00:16:28.14	and it rapidly increased in frequency in the population,
00:16:32.04	dragging with it the neighboring variation.
00:16:38.16	Now, here I'm showing the European variant,
00:16:41.17	in this case the T variant
00:16:43.20	is associated with lactose tolerance,
00:16:45.27	and it shows a very similar pattern.
00:16:50.22	So using computational approaches,
00:16:52.27	we were able to estimate the age of the African mutation
00:16:57.22	to be somewhere between about 3,000-7,000 years of age.
00:17:01.28	These are the populations
00:17:03.25	that had the oldest age estimates,
00:17:06.08	and they include individuals
00:17:08.07	who speak Cushitic languages.
00:17:10.04	They came from Ethiopia,
00:17:12.10	and they practiced agro-pastoralism.
00:17:15.01	They came into Kenya and Tanzania
00:17:17.16	within the past 5,000 years.
00:17:20.23	And then we saw it at very high prevalence
00:17:23.26	and an old age estimate in Nilo-Saharan-speaking groups,
00:17:27.11	and these would include, for example, the Maasai.
00:17:30.09	Now, they came into the region more recently,
00:17:32.17	from southern Sudan,
00:17:34.08	within the past 3,000 years, so if I were to guess,
00:17:37.05	I would think perhaps this mutation
00:17:39.04	arose in the Cushitic speaking populations.
00:17:42.03	But irregardless, it quickly, rapidly spread
00:17:45.07	to all of the populations in the area
00:17:47.21	because it was so selectively advantageous
00:17:51.22	and adaptive to have this mutation.
00:17:55.03	Now, because we see the correlation
00:17:59.18	between the practice of cattle domestication and pastoralism
00:18:04.11	and the rise in this mutations,
00:18:06.16	this is a really excellent example
00:18:08.22	of gene-culture co-evolution.
00:18:12.03	And in fact, what's really interesting is
00:18:15.01	that the date estimates that we came up with correlate really well
00:18:18.25	with the archaeological data,
00:18:20.17	which shows that cattle domestication
00:18:22.14	arose in the Middle East or north Africa
00:18:27.04	somewhere between 8,000-10,000 years ago,
00:18:29.26	and that corresponds with the age estimate for the European mutation,
00:18:33.18	which we inferred to be about 9,000 years old.
00:18:37.25	But cattle domestication was not introduced
00:18:40.25	south of the Saharan desert
00:18:44.20	until roughly 5,000 or 5,500 years ago,
00:18:48.21	correlating very well with the age estimate
00:18:52.03	for the mutation we found in eastern Africa.
00:18:54.24	And then it was introduced
00:18:56.13	much more recently into southern Africa.
00:19:00.12	But one could argue that perhaps Mendelian traits like lactose tolerance,
00:19:05.04	which are regulated by a single locus or gene of major effect,
00:19:10.23	are in a sense the low hanging fruit;
00:19:12.20	they're the easiest to identify.
00:19:15.04	So one thing that my lab is interesting in doing
00:19:17.10	is looking at more complex traits,
00:19:19.23	and perhaps one of the most classic complex traits is height.
00:19:23.28	So, height is highly heritable,
00:19:26.19	genome wide association studies in tens of thousands of Europeans
00:19:30.12	have identified hundreds of loci,
00:19:33.06	each of very small effect,
00:19:35.06	and explaining only a very small proportion of the variation in height.
00:19:39.20	Now, interestingly, most of these are not part of
00:19:42.22	the growth hormone/IGF1 pathway,
00:19:45.07	which we know plays a very important role in idiopathic short stature,
00:19:49.16	for example.
00:19:53.05	Now, in Africa, we see some of the broadest distributions,
00:19:56.21	or ranges in height,
00:19:59.02	ranging from the very short statured Pygmies in central Africa,
00:20:03.28	and then we see some of the tallest individuals
00:20:07.13	in the Sudan and in eastern Africa.
00:20:10.23	And it's thought that these differences
00:20:12.14	may be partly due to adaptation
00:20:14.29	to different environments.
00:20:16.27	So what I want to tell you today is about
00:20:18.19	our genetic studies of short stature
00:20:22.01	in Pygmy populations from central Africa.
00:20:25.14	And, for you to fully understand and appreciate the work we've done,
00:20:29.17	I think I should first tell you a little bit about
00:20:32.07	how we went about collecting these samples
00:20:34.07	and how challenging it could be.
00:20:35.25	So, this is...
00:20:37.18	to get to one of the groups that we studied in Cameroon,
00:20:39.27	you have to cross this river,
00:20:41.28	and you have a person who has a ferry,
00:20:44.05	he's actually using a hand crank here
00:20:47.13	to get us across.
00:20:50.12	And I guess I'm very fortunate
00:20:52.19	because as a woman, I was able to get shade,
00:20:54.15	but not everybody was that lucky.
00:20:56.28	And here are some other hazards that we run into,
00:20:59.15	but I'm smiling because the head is cut off of this snake.
00:21:03.01	But I actually have to give credit to Dr. Alain Froment,
00:21:06.24	who has been studying the Pygmy populations in Cameroon
00:21:09.16	for greater than 30 years,
00:21:11.20	and he did the majority of the sample collection
00:21:14.01	in this case.
00:21:16.21	So, the genetic basis of short stature in Pygmies
00:21:19.26	is a question that's been of tremendous interest
00:21:22.08	to endocrinologists and human geneticists alike
00:21:25.07	for most than 50 years.
00:21:27.08	The particular populations that we studied
00:21:29.25	are located in Cameroon, three different groups from Cameroon,
00:21:34.27	who mean male height is 152 cm.
00:21:40.11	And they live in very close connection and interaction
00:21:44.29	with neighboring populations who speak Bantu languages
00:21:47.29	and practice agriculture,
00:21:50.06	and their mean male height is 170 cm,
00:21:54.04	so that's quite a difference between the two.
00:21:58.16	So, the Pygmy short statured phenotype in humans
00:22:01.26	has arisen independently in different global populations.
00:22:05.12	Typically, these are populations
00:22:07.03	that live in tropical environments,
00:22:09.14	so there have been a number of hypotheses
00:22:11.11	about why this trait might be adaptive.
00:22:14.28	And these include thermoregulation,
00:22:19.04	limited food resources,
00:22:21.19	locomotion - that it may be easier to move
00:22:23.28	in a dense tropical environment if you're short,
00:22:26.18	and more recently there's a theory
00:22:30.10	that this is due to a life-history tradeoff,
00:22:32.10	and I'm going to focus on that theory.
00:22:35.08	And that has to do with the fact that
00:22:37.14	Pygmies have a remarkably short lifespan.
00:22:40.11	Their chance of living to age 15
00:22:42.06	is only about 40%,
00:22:44.18	and if they make it to age 15,
00:22:46.23	the expected lifespan is only around 25 years of age.
00:22:50.01	Now, that is due largely to very high infectious disease burden
00:22:54.05	and a very challenging life in dense tropical forests.
00:22:59.24	Now, what the study showed is that
00:23:02.18	Pygmies appear to be reaching reproduction...
00:23:05.25	they appear to be reproducing and reaching puberty
00:23:08.09	at a significantly earlier age
00:23:11.01	than other Africans.
00:23:13.13	And the growth trajectory in Pygmies
00:23:14.28	appears to be similar to other populations until the point of puberty,
00:23:19.21	and then they lack the adolescent growth spurt.
00:23:22.15	So this may be some sort of a tradeoff:
00:23:24.18	there's selection to reproduce earlier
00:23:26.22	because they're dying very young,
00:23:28.22	but that may be a tradeoff,
00:23:30.24	in that they're not undergoing the adolescent growth spurt.
00:23:35.20	Now, there have been only a handful
00:23:37.28	of physiologic and metabolic studies in Pygmies,
00:23:42.00	but nearly all of these are pointing towards
00:23:44.18	disruptions of the growth hormone/IGF1 pathway,
00:23:47.19	so this is in contrast to what we're seeing in European populations.
00:23:52.05	However, there's been quite a bit of dispute of
00:23:54.28	where along this pathway these disruptions are occurring.
00:24:00.04	So, in order to try to address these questions,
00:24:03.06	we genotyped one million single nucleotide polymorphisms
00:24:08.06	in 67 pygmy individuals
00:24:10.23	and 58 of the neighboring Bantu individuals.
00:24:14.14	And here we can see a plot,
00:24:17.10	similar to what I've shown you before,
00:24:19.09	based on structure analysis.
00:24:21.09	And to remind you,
00:24:23.02	this is composed of a series of lines,
00:24:24.23	and each line represents a person,
00:24:26.16	and they can have ancestry
00:24:28.08	from different ancestral populations,
00:24:31.01	represented by the different colors.
00:24:33.04	So here in orange
00:24:34.29	are individuals who speak the Bantu language
00:24:38.00	and practice agriculture,
00:24:40.08	and in dark green are individuals who self-identify as Pygmies.
00:24:44.19	And what you can see is that there's been
00:24:46.22	a lot of admixture between the Pygmies
00:24:49.29	and the neighboring Bantu people.
00:24:52.03	Now, interestingly, this tends to be unidirectional,
00:24:54.28	and it tends to be gene flow between males
00:24:57.27	from the Bantu population
00:25:00.03	with females of the Pygmy population.
00:25:02.22	This is largely due to socioeconomic factors.
00:25:06.23	Now, when we look at a correlation
00:25:08.26	between ancestry and height,
00:25:11.03	we observed a very strong and significant positive correlation.
00:25:15.04	So, we can see that Pygmies who have more of the Bantu ancestry
00:25:19.23	tend to be taller.
00:25:21.19	And, so this is showing
00:25:22.25	that there's a strong genetic component to this trait.
00:25:26.17	We've also worked with collaborators
00:25:28.11	to develop methods
00:25:30.19	to infer tracts of Pygmy and Bantu ancestry
00:25:35.11	across the chromosome.
00:25:36.29	So here, these are the different chromosomes,
00:25:38.18	starting with chromosome 1
00:25:40.05	and going up to chromosome 22,
00:25:42.25	and here I'm showing you an example from chromosome 3.
00:25:46.04	And in blue is showing tracts of the genome
00:25:49.03	that are Pygmy ancestry,
00:25:50.24	and in red are tracts of the genome that are Bantu ancestry,
00:25:54.23	and what we tend to see are very, very short tracts of Bantu ancestry.
00:25:58.28	And that's reflected in the fact that admixture
00:26:01.08	has been occurring over thousands of years.
00:26:06.11	Now, the next question that we wanted to address
00:26:08.17	is how do the genomes of the Pygmy hunter-gatherers
00:26:12.04	differ from the genomes of the Bantu agriculturalists
00:26:17.00	and from other groups, such as the Maasai pastoralists
00:26:20.28	from east Africa.
00:26:22.28	And to do that,
00:26:25.03	we use a number of scans of natural selection
00:26:27.28	across the genome.
00:26:29.29	Without getting into detail about the methods,
00:26:32.26	I'll just point out that you can see by the different colors here
00:26:37.00	across the different chromosomes,
00:26:39.00	here's chromosome 22 and going down to chromosome 1,
00:26:42.04	that we found a number of regions of the genome
00:26:44.22	that are targets of selection.
00:26:47.05	But there was one region in particular,
00:26:49.26	on chromosome 3,
00:26:52.04	where we saw a cluster of targets of natural selection.
00:26:57.01	And this was over about a 15 million basepair region.
00:27:01.14	Now, given our small sample size,
00:27:03.20	we have very little power
00:27:05.15	to detect a genome-wide association.
00:27:09.04	And so what we did is,
00:27:10.26	under the hypothesis that this is an adaptive trait,
00:27:13.17	we just focused on the regions of the genome
00:27:16.07	that are targets of selection, shown here,
00:27:19.11	and then we looked for an association with height.
00:27:22.10	And one of the strongest, most significant associations
00:27:25.09	was exactly in that same 15 million basepair region
00:27:29.19	of chromosome 3.
00:27:31.23	And indeed, it encompassed several genes,
00:27:34.15	one of which is DOCK3,
00:27:36.18	which has been shown to be associated with height
00:27:39.09	in non-African populations,
00:27:41.08	so we replicated that finding.
00:27:43.20	But nearby was another gene called CISH,
00:27:47.09	which is a member of the cytokine signaling family,
00:27:50.10	plays a very important role in regulating
00:27:52.28	IL-2 cytokine signaling pathway,
00:27:56.18	and studies have shown that it's associated
00:27:58.26	with resistance to a number of infectious diseases
00:28:01.18	in Africa.
00:28:04.01	Now, interestingly,
00:28:05.29	CISH also directly inhibits
00:28:07.14	human growth hormone receptor action
00:28:10.06	by blocking the STAT5 phosphorylation pathway.
00:28:13.15	And so we know that studies in mice
00:28:15.17	show that when this gene is overexpressed,
00:28:18.06	the mice are short statured.
00:28:20.23	Now, this led me to the hypothesis that,
00:28:24.14	could it be that there could actually be selection
00:28:26.19	for immune function
00:28:28.11	that is indirectly resulting
00:28:30.05	in short stature in Pygmies,
00:28:32.05	because that gene plays an important role in both.
00:28:35.29	And we need to do further functional studies,
00:28:38.20	and look at differences in gene expression
00:28:40.13	to test this hypothesis.
00:28:44.04	The last study I wanna tell you about is a study
00:28:46.20	in which we sequenced the entire genomes,
00:28:49.15	at high coverage,
00:28:51.07	of 15 African hunter-gatherers,
00:28:53.22	including 5 Pygmies,
00:28:55.28	5 Hadza,
00:28:57.10	and 5 Sandawe.
00:28:59.26	We identified over 13 million variants,
00:29:02.29	3 million of which are completely novel;
00:29:05.29	they have never previously been identified.
00:29:08.13	And that's just from 15 individuals,
00:29:10.14	so you can imagine how much variation is out there.
00:29:13.16	Many of these are novel variants...
00:29:15.27	many of these novel variants are in known regulatory sites.
00:29:21.04	So now, combining the two studies,
00:29:24.08	we wanted to ask the question,
00:29:26.03	which pathways are enriched for genes near targets of selection?
00:29:29.16	And these enriched pathways
00:29:31.25	include genes involved in neuro-endocrine signaling,
00:29:35.01	reproduction,
00:29:36.06	metabolism,
00:29:37.11	and immune function,
00:29:38.22	and interestingly, based on the whole genome sequencing study,
00:29:42.08	we saw an enrichment for genes
00:29:44.06	that play a role in pituitary function in Pygmies,
00:29:47.13	including follicle-stimulating hormone receptor,
00:29:50.13	growth hormone receptor,
00:29:52.11	HESX1, which I'll tell you more about in a moment,
00:29:55.11	and thyrotropin-releasing hormone receptor.
00:29:58.15	In fact, TRHR was one of the biggest hits
00:30:02.13	that we saw in terms of these studies of selection.
00:30:05.17	And what's interesting is that this gene
00:30:08.22	plays an important role in the hypothalamic-pituitary-thyroid axis,
00:30:12.28	influencing a number of traits that could potentially
00:30:15.14	be of adaptive significance in Pygmies.
00:30:18.26	And also of interest was that anthropologists
00:30:21.18	have noted that there is a significant difference
00:30:24.23	in the prevalence of Goiter
00:30:27.00	among Pygmies and neighboring Bantu groups.
00:30:29.24	So the Pygmies have a much lower frequency of Goiter
00:30:33.16	compared to the neighboring Bantu populations,
00:30:36.16	and this could reflect a biological adaptation in Pygmies
00:30:41.20	to a low iodine environment.
00:30:43.24	It's very deleterious to get Goiter
00:30:46.22	because it can also lead to a diseased called Cretinism,
00:30:49.27	which of course is going to be very deleterious.
00:30:52.18	So again, here's an example
00:30:54.10	where something like adaptation to diet
00:30:56.13	could indirectly influence growth
00:30:58.28	or other phenotypes in the Pygmy population.
00:31:04.01	The last thing we wanted to do
00:31:06.01	was to look for regions of the genome,
00:31:08.08	using the whole genome sequencing data,
00:31:10.13	that are specific to Pygmies,
00:31:12.20	and those are shown in green here.
00:31:16.02	Now, we identified 25 clusters in the genome,
00:31:19.23	and the largest cluster
00:31:22.27	was right in that same region of chromosome 3
00:31:25.14	that we had previously identified.
00:31:28.00	But we had missed it in the prior study,
00:31:30.11	and the reason why is because
00:31:32.17	it contains these Pygmy-specific variants,
00:31:35.08	that were not captured by the SNP array that we used,
00:31:39.17	and thus demonstrating the great importance
00:31:42.00	of doing resequencing for identifying novel
00:31:44.24	and potentially functionally important variation
00:31:47.15	in ethnically diverse populations.
00:31:50.28	Now, this cluster consisted of
00:31:55.10	44 SNPs in 100% association with each other
00:31:59.16	over 170,000 nucleotide,
00:32:03.06	shown here,
00:32:05.24	and it contained a very interesting candidate gene called HESX1.
00:32:10.10	HESX1 codes for a transcription factor
00:32:13.05	that plays a very important role
00:32:15.04	in regulating the development
00:32:17.15	at the anterior pituitary in the brain,
00:32:20.14	and that's the site of production of growth hormone,
00:32:22.23	as well as other reproductive hormones.
00:32:25.11	Now, interestingly,
00:32:27.06	we identified a non-synonymous,
00:32:29.28	so an amino acid change, basically,
00:32:33.23	in this gene
00:32:36.03	that had been previously associated
00:32:38.13	with idiopathic short stature in humans.
00:32:41.26	But it turns out that this varian
00:32:44.01	t is present at about a 20% frequency in other Africans.
00:32:47.12	So what we hypothesize is that
00:32:49.13	there's something about this region
00:32:51.22	that may be altering gene expression of HESX1
00:32:55.07	or other genes in that region.
00:32:58.01	Upstream, we found another cluster
00:33:01.18	near this gene POU1F1, also known at Pit-1 in mouse,
00:33:07.13	and again this codes for a transcription factor
00:33:09.18	that plays a critical role in regulating growth hormone expression.
00:33:14.23	So another excellent candidate gene.
00:33:17.28	Now, what is interesting is that
00:33:19.27	both of these clusters, or genes,
00:33:23.18	are amongst the most differentiated regions
00:33:26.27	of the Pygmy genomes,
00:33:28.27	compared to genomes from elsewhere in Africa.
00:33:31.29	So we then picked out some of the SNPs in these regions
00:33:37.13	and genotyped them in a larger set
00:33:39.19	of western and eastern Pygmies,
00:33:41.26	and we showed that they are statistically
00:33:44.02	associated with short stature in Pygmies.
00:33:47.29	So the next step is going to be
00:33:49.24	to try to make transgenic models
00:33:52.01	that express these variants using transgenic mouse models,
00:33:56.06	and see what the phenotype looks like.
00:34:00.19	So that leads us to a number of hypotheses.
00:34:03.19	One, is that alterations in the growth hormone/IGF1 pathway
00:34:07.15	play a role in the short stature trait in Pygmies.
00:34:13.01	Two, is that anterior pituitary hormones
00:34:15.10	may play a central role in the Pygmy phenotype,
00:34:18.09	influencing growth, reproduction,
00:34:20.15	metabolism, and immunity.
00:34:24.00	And thirdly, that short stature
00:34:26.16	could be a byproduct of selection
00:34:28.11	acting on pleiotropic loci.
00:34:31.04	So if we look here,
00:34:32.21	one of the candidate loci that we identified is HESX1.
00:34:36.13	That's going to influence expression and development
00:34:39.20	of the anterior pituitary,
00:34:42.02	site of production of growth hormone.
00:34:44.20	Growth hormone expression is also regulated
00:34:46.23	by this other gene we found, POU1F1.
00:34:50.04	And this CISH regulates growth hormone receptor.
00:34:54.17	Now, if we look at the downstream effects
00:34:56.24	of growth hormone,
00:34:59.07	growth hormone, when it binds to growth hormone receptor,
00:35:02.18	will trigger off expression of IGF1,
00:35:06.12	predominantly from the liver, but from other tissues as well.
00:35:10.06	IGF1 will have an effect on muscle growth
00:35:13.14	and also on bone growth and height,
00:35:16.02	but the other impact, or the other role of growth hormone
00:35:20.12	is that it also influences insulin metabolism,
00:35:24.06	it influences fat metabolism.
00:35:28.01	And then we know that infectious disease
00:35:30.01	alters immune response and cytokine levels,
00:35:33.08	and that these can influence gene expression from CISH,
00:35:36.11	or other genes that are in this pathway.
00:35:40.09	So, when we go back to Africa to study the Pygmies,
00:35:42.28	what we would ultimately like to do next
00:35:45.16	is to measure all of the phenotypes,
00:35:48.01	because if you want to understand something
00:35:50.04	like the evolution of short stature in Pygmies,
00:35:52.19	I think you can't just be looking at stature
00:35:55.09	because the growth hormone pathway
00:35:58.25	plays a role in all of these different traits,
00:36:01.01	so we need to be looking at this as an integrative picture.
00:36:06.01	And in fact, our approach in the future
00:36:08.26	is to use an integrative genomics approach
00:36:11.24	combining whole genome data,
00:36:14.15	data on protein variation from blood,
00:36:17.25	epigenetic variation,
00:36:19.21	which can be influenced by diet and environment,
00:36:22.12	gene expression,
00:36:24.10	we're starting to look at the microbiome,
00:36:27.16	which is the spectrum of bacteria in the gut,
00:36:32.05	because that can not only be influenced by diet,
00:36:35.16	it can also have an influence on the metabolome,
00:36:38.12	or the set of all the metabolites, for example,
00:36:40.27	in blood.
00:36:42.20	And we want to combine that information
00:36:44.25	together with information on diet
00:36:46.22	and other environmental factors,
00:36:48.29	to try to identify genetic and environmental factors
00:36:52.15	that play a role in short stature
00:36:55.05	and in other anthropometric,
00:36:56.25	cardiovascular,
00:36:58.01	and metabolic traits.
00:37:00.20	One of the other approaches we can take
00:37:02.20	to distinguish the role of genetics and environment is, for example,
00:37:06.00	to look at individuals of the same or similar ethnic background,
00:37:10.29	but living in an urban versus a rural environment.
00:37:16.21	We can also take a different...
00:37:18.14	the opposite approach.
00:37:20.00	We can look at individuals who have
00:37:22.06	very different genetic ancestries,
00:37:25.03	but live in similar environments.
00:37:27.13	So for example,
00:37:29.20	this is a girl who is from the Fulani population,
00:37:33.17	and here's a neighboring...
00:37:35.19	an individual from the Tupuri population.
00:37:38.26	So they are genetically very differentiated,
00:37:41.20	but live in a similar environment,
00:37:43.16	yet the Fulani seem to have some innate resistance
00:37:47.06	to malaria infection.
00:37:50.03	By contrast, in the San,
00:37:53.09	from southern Africa,
00:37:54.29	are very differentiated from the Bantu,
00:37:57.15	but the San seem to have an innate susceptibility
00:38:01.09	to TB infection.
00:38:03.20	So again, by contrasting populations with different ancestry,
00:38:07.26	and living in different environments,
00:38:09.11	we may identify clues about the genetic basis
00:38:12.10	of differences in phenotypic variation
00:38:14.26	and disease susceptibility.
00:38:17.23	So in conclusion,
00:38:20.20	Africans have the highest levels of genetic diversity
00:38:23.04	within and among populations.
00:38:26.28	The demographic history of Africans
00:38:29.00	and local adaptation to different environments
00:38:31.04	has resulted in population
00:38:33.01	or region specific genetic variation.
00:38:36.25	And we need to be including
00:38:38.21	ethnically diverse Africans in genomic studies
00:38:41.17	to better identify both unique rare, and common variants
00:38:45.28	which may be of functional importance,
00:38:47.28	including those that play a role in disease risk
00:38:50.13	in these populations.
00:38:52.14	And I will just end by thanking
00:38:54.04	the many individuals
00:38:55.25	who contributed to these studies,
00:38:57.29	and my funding agencies,
00:39:00.16	and particular thanks to the Africans
00:39:02.20	who have contributed to these studies.