1
00:00:01,066 --> 00:00:03,527
So this is a talk about gene drives,

2
00:00:03,527 --> 00:00:06,801
but I'm going to start by 
telling you a brief story.

3
00:00:06,801 --> 00:00:10,098
20 years ago, a biologist
named Anthony James

4
00:00:10,098 --> 00:00:12,211
got obsessed by the idea
of making mosquitos

5
00:00:12,211 --> 00:00:15,857
that didn't transmit malaria.

6
00:00:15,857 --> 00:00:20,803
It was a great idea,
but pretty much a complete failure.

7
00:00:20,803 --> 00:00:23,427
For one thing, it turned out to be
really hard

8
00:00:23,427 --> 00:00:26,469
to make a malaria resistant mosquito.

9
00:00:26,469 --> 00:00:29,766
James managed it, finally,
just a few years ago

10
00:00:29,766 --> 00:00:31,902
by adding some genes
that make it impossible


11
00:00:31,902 --> 00:00:35,757
for the malaria gene
to survive inside the mosquito.

12
00:00:35,757 --> 00:00:38,218
But that just created another problem.

13
00:00:38,218 --> 00:00:41,051
Now that you've got malaria-resistant
mosquito,

14
00:00:41,051 --> 00:00:46,322
how do you get it to replace
all the malaria-carrying mosquitos?

15
00:00:46,322 --> 00:00:48,133
There are a couple options,

16
00:00:48,133 --> 00:00:53,009
but plan A was basically to breed up
a bunch of the new genetically-engineered mosquotos,

17
00:00:53,009 --> 00:00:54,263
release them into the wild,

18
00:00:54,263 --> 00:00:56,863
and hope that they pass on their genes.

19
00:00:56,863 --> 00:00:58,721
The problem was that you'd 
have to release

20
00:00:58,721 --> 00:01:01,925
literally 10x the number of native
mosquitos to work.

21
00:01:01,925 --> 00:01:05,246
So in a village with 10,000 mosquitos,

22
00:01:05,246 --> 00:01:08,497
you release an extra 100,000.

23
00:01:08,497 --> 00:01:11,004
As you might guess, this was not
a very popular strategy

24
00:01:11,004 --> 00:01:12,514
with the villagers.

25
00:01:12,514 --> 00:01:15,137
(Laughter)

26
00:01:15,137 --> 00:01:18,829
Then, last January, Anthony James
got an email


27
00:01:18,829 --> 00:01:21,639
from a biologist named 
Ethan Bier.

28
00:01:21,639 --> 00:01:24,472
Bier said that he and his grad student,
Valentino Gantz,


29
00:01:24,472 --> 00:01:26,887
had stumbled on a tool that could not only
guarentee

30
00:01:26,887 --> 00:01:30,161
that a particular gene trait
would not be inherited,

31
00:01:30,161 --> 00:01:32,994
but that it would spread
incredibly quickly.

32
00:01:32,994 --> 00:01:35,223
If they were right, it would basically
solve the problem


33
00:01:35,223 --> 00:01:38,752
that he and James had been
working on for 20 years.

34
00:01:38,752 --> 00:01:41,562
As a test, they engineered
two mosquitos

35
00:01:41,562 --> 00:01:43,257
to carry the anti-malaria gene

36
00:01:43,257 --> 00:01:45,556
and also this new tool,
a gene drive,

37
00:01:45,556 --> 00:01:47,506
which I'll explain in a minute.

38
00:01:47,506 --> 00:01:49,712
Finally, they set it up so that
any mosquitos

39
00:01:49,712 --> 00:01:51,825
that had inherited the
anti-malaria gene

40
00:01:51,825 --> 00:01:53,845
wouldn't have the usual white eyes,
but would instead have red eyes.

41
00:01:53,845 --> 00:01:58,280
That was pretty much just 
for convenience

42
00:01:58,280 --> 00:02:02,413
so they could tell just at a glance
which was which.

43
00:02:02,413 --> 00:02:05,153
So they took their two 
anti-malarial, red eye mosquitos

44
00:02:05,153 --> 00:02:07,847
and put them in a box with 30
ordinary white-eyed ones

45
00:02:07,847 --> 00:02:09,797
and let them breed.

46
00:02:09,797 --> 00:02:14,534
In two generations, 
those had produced 38,000 grandchildren.

47
00:02:14,534 --> 00:02:16,972
That is not the surprising part.

48
00:02:16,972 --> 00:02:18,923
This is the surprising part:

49
00:02:18,923 --> 00:02:21,593
given that you started with just
two red-eyed mosquitos

50
00:02:21,593 --> 00:02:23,172
and 30 white-eyed ones,

51
00:02:23,172 --> 00:02:26,725
you expect mostly white-eyed
descendents.

52
00:02:26,725 --> 00:02:29,790
Instead, when James opened the box,

53
00:02:29,790 --> 00:02:33,598
all 38,000 mosquitos had red eyes.

54
00:02:33,598 --> 00:02:35,432
When I asked Ethan Bier
about this moment,

55
00:02:35,432 --> 00:02:40,053
he became so excited, tht he was
literally shouting into the phone.

56
00:02:40,053 --> 00:02:42,073
That's because getting only
red-eyed mosquitos

57
00:02:42,073 --> 00:02:45,045
violates a rule that is the
absolute cornerstone of biology,

58
00:02:45,045 --> 00:02:47,042
Mendelian genetics.

59
00:02:47,042 --> 00:02:50,873
I'll keep this quick, but Mendelian genetics
says when a male and female mate,

60
00:02:50,873 --> 00:02:53,753
their baby inherits half of its
DNA from each parent.

61
00:02:53,753 --> 00:02:57,352
So if our original mosquito was aa
and our new mosquito is aB,

62
00:02:57,352 --> 00:02:59,140
where B is the anti-malarial gene,

63
00:02:59,140 --> 00:03:01,415
the babies should come out
in four permutations:

64
00:03:01,415 --> 00:03:05,130
aa, aB, aa and Ba.

65
00:03:05,130 --> 00:03:07,197
Instead, with the new gene drive,

66
00:03:07,197 --> 00:03:10,123
they all came out aB.

67
00:03:10,123 --> 00:03:12,723
Biologically, that shouldn't
even be possible.

68
00:03:12,723 --> 00:03:14,906
So what happened?

69
00:03:14,906 --> 00:03:16,717
The first thing that happened
was the arrival

70
00:03:16,717 --> 00:03:18,645
of a gene-editing tool
known as CRISPR in 2012.

71
00:03:18,645 --> 00:03:22,638
Many of you have probably heard
about CRISPR,

72
00:03:22,638 --> 00:03:25,401
so I'll just say briefly that CRISPR
is a tool that allows researchers

73
00:03:25,401 --> 00:03:28,188
to edit genes very precisely,
easily and quickly.

74
00:03:28,188 --> 00:03:33,482
It does this by harnessing a mechanism
that already existed in bacteria.

75
00:03:33,482 --> 00:03:35,409
Basically, there's a protein
that acts like a scissors

76
00:03:35,409 --> 00:03:37,267
and cuts the DNA,

77
00:03:37,267 --> 00:03:39,217
and there's an RNA molecule
that directs the scissors

78
00:03:39,217 --> 00:03:41,237
to any point on the genome you want.

79
00:03:41,237 --> 00:03:44,651
The result is basically a word processor
of genes.

80
00:03:44,651 --> 00:03:47,135
You can take an entire gene out,
put one in,

81
00:03:47,135 --> 00:03:49,945
or even edit just a single letter
within a gene.

82
00:03:49,945 --> 00:03:53,544
And you can do it in nearly any species.

83
00:03:53,544 --> 00:03:58,351
Okay, remember how I said 
that gene drives

84
00:03:58,351 --> 00:03:59,094
originally had two problems?

85
00:03:59,094 --> 00:04:01,137
The first is that it was hard
to engineer a mosquito

86
00:04:01,137 --> 00:04:02,623
to be malaria resistant.

87
00:04:02,623 --> 00:04:05,363
That's basically gone now,
thanks to CRISPR.

88
00:04:05,363 --> 00:04:07,546
But the other problem was 
logistical.

89
00:04:07,546 --> 00:04:10,402
How do you get your trait to spread?

90
00:04:10,402 --> 00:04:13,165
This is where it gets clever.

91
00:04:13,165 --> 00:04:16,810
A couple years ago, a biologist
at Harvard named Kevin Esvelt

92
00:04:16,810 --> 00:04:19,086
wondered what would happen
if you made it so that

93
00:04:19,086 --> 00:04:21,803
CRISPR inserted not only
your new gene,

94
00:04:21,803 --> 00:04:25,448
but also the machinery 
that does the cutting and pasting.

95
00:04:25,448 --> 00:04:30,185
In other words, what if CRISPR
also copy and pasted itself.

96
00:04:30,185 --> 00:04:34,713
You'd end up with a perpetual
motion machine for gene editing.

97
00:04:34,713 --> 00:04:37,267
And that's exactly what happened.

98
00:04:37,267 --> 00:04:40,077
This CRISPR gene drive
that Esvelt created

99
00:04:40,077 --> 00:04:43,606
not only guarantees that a trait 
will get passed on,

100
00:04:43,606 --> 00:04:46,068
but if its used in the germline cell,

101
00:04:46,068 --> 00:04:48,668
it will automatically copy and paste
your new gene

102
00:04:48,668 --> 00:04:51,989
into both chromosomes of every
single individual.

103
00:04:51,989 --> 00:04:54,450
It's like a global search and replace,

104
00:04:54,450 --> 00:04:56,006
or in science terms,

105
00:04:56,006 --> 00:04:59,326
it makes a heterozygous trait
homozygous.

106
00:04:59,326 --> 00:05:01,927
So, what does this mean?

107
00:05:01,927 --> 00:05:04,574
For one thing, it means we have
a very powerful,

108
00:05:04,574 --> 00:05:08,846
but also somewhat alarming new tool.

109
00:05:08,846 --> 00:05:11,586
Up until now, the fact that gene drives
didn't work very well

110
00:05:11,586 --> 00:05:13,699
was actually kind of a relief.

111
00:05:13,699 --> 00:05:15,928
Normally when we mess around
with an organisms's genes,

112
00:05:15,928 --> 00:05:18,784
we make that thing less evolutionarily fit.

113
00:05:18,784 --> 00:05:20,944
So biologists can make all the mutant
fruit flies they want

114
00:05:20,944 --> 00:05:22,500
without worrying about it.

115
00:05:22,500 --> 00:05:27,004
If some escape, natural selection
just takes care of it.

116
00:05:27,004 --> 00:05:30,302
What's remarkable and powerful
and frightening about gene drives

117
00:05:30,302 --> 00:05:32,856
is that that will no longer be true.

118
00:05:32,856 --> 00:05:36,943
Assuming that your trait does not
have a big evolutionary handicap,

119
00:05:36,943 --> 00:05:39,009
like a mosquito that can't fly,

120
00:05:39,009 --> 00:05:42,492
the CRISPR-based gene drive
will spread the change relentlessly

121
00:05:42,492 --> 00:05:47,322
until it is in every single individual
in the population.

122
00:05:47,322 --> 00:05:50,178
Now, it isn't easy to make
a gene drive that works that well,

123
00:05:50,178 --> 00:05:53,568
but James and Esvelt think
that we can.

124
00:05:53,568 --> 00:05:57,469
The good news is that this opens
the door to some remarkable things.

125
00:05:57,469 --> 00:06:00,116
If you put an anti-malarial gene drive
in just 1 percent

126
00:06:00,116 --> 00:06:02,183
of anopheles mosquitos,
the species that transmits malaria.

127
00:06:02,183 --> 00:06:06,502
Researchers estimate that it would
spread to the entire population

128
00:06:06,502 --> 00:06:08,406
in a year.

129
00:06:08,406 --> 00:06:11,401
So in a year, you could virtually 
eliminate malaria.

130
00:06:11,401 --> 00:06:14,350
In practice, we're still a few years out
from being able to do that,

131
00:06:14,350 --> 00:06:18,344
but sitll, a 1,000 children a day
die of malaria.

132
00:06:18,344 --> 00:06:21,293
In a year, that number could be
almost zero.

133
00:06:21,293 --> 00:06:25,426
The same goes for dengue fever,
chicken genuang (?), yellow fever.

134
00:06:25,426 --> 00:06:27,469
And it gets better.

135
00:06:27,469 --> 00:06:29,512
Say you want to get rid 
of an invasive species,

136
00:06:29,512 --> 00:06:32,670
like get Asian Carp out of
The Great Lakes.

137
00:06:32,670 --> 00:06:34,412
All you have to do is release 
a gene drive

138
00:06:34,412 --> 00:06:37,570
that makes the fish produce
only male offspring.

139
00:06:37,570 --> 00:06:41,889
In a few generations, there'll be
no females left, no more carp.

140
00:06:41,889 --> 00:06:44,164
In theory, this means that we 
could restore hundreds

141
00:06:44,164 --> 00:06:47,601
of native species that have been
pushed to the brink.

142
00:06:47,601 --> 00:06:50,968
Okay, that's the good news,

143
00:06:50,968 --> 00:06:53,313
this is the bad news.

144
00:06:53,313 --> 00:06:55,380
Gene drives are so effective,

145
00:06:55,380 --> 00:06:59,234
that even an accidental release 
could change an entire species,

146
00:06:59,234 --> 00:07:01,440
and often very quickly.

147
00:07:01,440 --> 00:07:03,530
Anthony James took the precautions.

148
00:07:03,530 --> 00:07:05,736
He breed his mosquitos in
a bio-containment lab

149
00:07:05,736 --> 00:07:08,197
and he also used a species
that's not native to the US

150
00:07:08,197 --> 00:07:09,706
so that even if some did escape,

151
00:07:09,706 --> 00:07:11,564
they'd just die off,

152
00:07:11,564 --> 00:07:12,725
there'd be nothing for them
to mate with.

153
00:07:12,725 --> 00:07:15,651
But it's also true that if 
a dozen Asian Carp

154
00:07:15,651 --> 00:07:19,482
with the all-male gene drive accidentally
got carried from The Great Lakes

155
00:07:19,482 --> 00:07:20,968
back to Asia,

156
00:07:20,968 --> 00:07:25,519
they could potentially wipe out 
the native Asian Carp population.

157
00:07:25,519 --> 00:07:29,048
And that's not so unlikely, given
how connected our world is.

158
00:07:29,048 --> 00:07:32,230
In fact, it's why we have 
an invasive species problem.

159
00:07:32,230 --> 00:07:33,646
And that's fish.

160
00:07:33,646 --> 00:07:37,826
Things like mosquitos and fruit flies,
there's literally no way to contain them.

161
00:07:37,826 --> 00:07:41,982
They cross borders and oceans
all the time.

162
00:07:41,982 --> 00:07:44,467
Okay, the other piece of bad news
is that

163
00:07:44,467 --> 00:07:48,809
a gene drive might not stay confined
to what we call the target species.

164
00:07:48,809 --> 00:07:50,039
That's because of gene flow,

165
00:07:50,039 --> 00:07:52,454
which is a fancy way of saying
that neighboring species

166
00:07:52,454 --> 00:07:54,103
sometimes inter-breed.

167
00:07:54,103 --> 00:07:56,727
If that happens, it's possible
a gene drive could cross over,

168
00:07:56,727 --> 00:07:59,861
like Asian Carp could infect
some other kind of Carp.

169
00:07:59,861 --> 00:08:03,437
That's not so bad if your drive
just promotes a trait, like eye color.

170
00:08:03,437 --> 00:08:05,782
In fact, there's a decent chance that
we'll see a wave

171
00:08:05,782 --> 00:08:09,614
of very weird fruit flies 
in the near future.

172
00:08:09,614 --> 00:08:11,703
But it could be a disaster
if your drive is deigned

173
00:08:11,703 --> 00:08:14,629
to eliminate the species entirely.

174
00:08:14,629 --> 00:08:17,903
The last worrisome thing is that
the technology to do this,

175
00:08:17,903 --> 00:08:21,618
to genetically engineer an organism
and include a gene drive

176
00:08:21,618 --> 00:08:25,543
is something basically any lab
in the world can do.

177
00:08:25,543 --> 00:08:27,423
An undergraduate can do it.

178
00:08:27,423 --> 00:08:32,160
A talented high schooler
with some equipment can do it.

179
00:08:32,160 --> 00:08:35,527
Now I'm guessing that this sounds 
terrifying.

180
00:08:35,527 --> 00:08:41,355
Interestingly though, nearly 
every scientist I talk to

181
00:08:41,355 --> 00:08:43,747
seems to think that gene drives
were not actually

182
00:08:43,747 --> 00:08:44,583
that frightening or dangerous.

183
00:08:44,583 --> 00:08:46,162
Partly because they believe that
scientists will be

184
00:08:46,162 --> 00:08:48,554
very cautious and responsible
about using them.

185
00:08:48,554 --> 00:08:50,318
(Laughter)

186
00:08:50,318 --> 00:08:52,524
So far, that's been true.

187
00:08:52,524 --> 00:08:55,380
The gene drives also have
some actual limitations.

188
00:08:55,380 --> 00:08:58,399
For one thing, they work in only
sexually reproducing species.

189
00:08:58,399 --> 00:09:01,882
So thank goodness, they can't be used
to engineer viruses or bacteria.

190
00:09:01,882 --> 00:09:05,643
Also, the trait spreads only with each
succesive genertion.

191
00:09:05,643 --> 00:09:08,732
So changing or eliminating a population
is practical only if that species

192
00:09:08,732 --> 00:09:11,100
has a fast reproductive cycle,


193
00:09:11,100 --> 00:09:14,722
like insects or maybe small vertebrates
like mice or fish.

194
00:09:14,722 --> 00:09:16,534
In elephants or people, it would
take centuries

195
00:09:16,534 --> 00:09:20,342
for a trait to spread widely enough
to matter.

196
00:09:20,342 --> 00:09:26,332
Also, even with CRISPR, it's not that easy
to engineer a truly devastating trait.

197
00:09:26,332 --> 00:09:27,679
Say you wanted to make a fruit fly


198
00:09:27,679 --> 00:09:30,303
that feeds on ordinary fruit instead
of rotten fruit

199
00:09:30,303 --> 00:09:33,298
with the aim of sabotaging 
American agriculture.

200
00:09:33,298 --> 00:09:35,481
First you'd have to figure out
which gene controls

201
00:09:35,481 --> 00:09:37,594
what the fly wants to eat,

202
00:09:37,594 --> 00:09:40,636
which is already a very long 
and complicated project.

203
00:09:40,636 --> 00:09:43,817
Then you'd have to alter those genes
to change the fly's behavior

204
00:09:43,817 --> 00:09:45,442
to whatever you'd want it to be,

205
00:09:45,442 --> 00:09:48,507
which is an even longer 
and complicated project.

206
00:09:48,507 --> 00:09:51,224
And it might not even work because
the genes that control behavior are complex.

207
00:09:51,224 --> 00:09:53,941
So if you're a terrorist and have 
to choose between

208
00:09:53,941 --> 00:09:55,682
starting a grueling basic research program

209
00:09:55,682 --> 00:09:57,726
that will require years 
of meticulous lab work

210
00:09:57,726 --> 00:09:59,398
and still might not pan out,

211
00:09:59,398 --> 00:10:00,768
or just blowing stuff up?

212
00:10:00,768 --> 00:10:03,717
You'll probably choose the later.

213
00:10:03,717 --> 00:10:05,621
This is especially true because
at least in theory,

214
00:10:05,621 --> 00:10:09,011
it should be pretty easy to build
what's called a reversal drive.

215
00:10:09,011 --> 00:10:13,028
That's one that basically overwrites
the change made by the first gene drive.

216
00:10:13,028 --> 00:10:15,303
So if you don't like the effects 
of a change,

217
00:10:15,303 --> 00:10:18,159
you can just release a second drive
that will cancel it out,

218
00:10:18,159 --> 00:10:21,573
at least in theory.

219
00:10:21,573 --> 00:10:24,847
Okay, so where does this leave us?

220
00:10:24,847 --> 00:10:29,351
We now have the ability to change
entire species at will.

221
00:10:29,351 --> 00:10:30,814
Should we?

222
00:10:30,814 --> 00:10:34,251
Are we gods now?

223
00:10:34,251 --> 00:10:36,317
I'm not sure I'd say that.

224
00:10:36,317 --> 00:10:38,663
But I would say this:

225
00:10:38,663 --> 00:10:41,426
First, some very smart people
are even now debating

226
00:10:41,426 --> 00:10:43,841
how to regulate gene drives.

227
00:10:43,841 --> 00:10:46,418
At the same time, some other
very smart people

228
00:10:46,418 --> 00:10:48,601
are working hard to create
safeguards,

229
00:10:48,601 --> 00:10:52,687
like gene drives that self regulate
or petter out after a few generations.

230
00:10:52,687 --> 00:10:56,101
But this technology still requires
a conversation.

231
00:10:56,101 --> 00:10:59,723
And given the nature of gene drives,

232
00:10:59,723 --> 00:11:02,370
that conversation has to be global.

233
00:11:02,370 --> 00:11:05,528
What if Kenya wants to use a drive
that Tanzania doesn't?

234
00:11:05,528 --> 00:11:11,287
Who decides whether to release
a gene drive that can fly?

235
00:11:11,287 --> 00:11:13,841
I don't have the answer to that question.

236
00:11:13,841 --> 00:11:15,884
All we can do going forward, I think,

237
00:11:15,884 --> 00:11:18,671
is talk honestly about 
the risks and benefits

238
00:11:18,671 --> 00:11:21,945
and take responsibility for our choices.

239
00:11:21,945 --> 00:11:25,962
By that I mean, not just the choice
to use a gene drive,

240
00:11:25,962 --> 00:11:29,421
but also the choice not to use one.

241
00:11:29,421 --> 00:11:32,347
Humans have a tendency to assume
that the safest option

242
00:11:32,347 --> 00:11:35,064
is to preserve the status quo.

243
00:11:35,064 --> 00:11:37,711
But that's not always the case.

244
00:11:37,711 --> 00:11:41,356
Gene drives have risks and those
need to be discussed,

245
00:11:41,356 --> 00:11:45,211
but malaria exists now and kills
1,000 people a day.

246
00:11:45,211 --> 00:11:48,856
To combat it, we spray pesticides
that do grave damage to other species,

247
00:11:48,856 --> 00:11:51,875
including amphibians and birds.

248
00:11:51,875 --> 00:11:54,336
So when you hear about gene drives
in the coming months,

249
00:11:54,336 --> 00:11:57,193
and trust me, you will be 
hearing about them,

250
00:11:57,193 --> 00:11:58,632
remember that.

251
00:11:58,632 --> 00:12:00,675
It can be frightening to act,

252
00:12:00,675 --> 00:12:04,925
but sometimes, not acting
is worse.

253
00:12:04,925 --> 00:12:12,518
(Applause)