1 00:00:01,286 --> 00:00:05,317 The most important gift your mother and father ever gave you 2 00:00:05,341 --> 00:00:08,061 was the two sets of three billion letters of DNA 3 00:00:08,085 --> 00:00:09,649 that make up your genome. 4 00:00:10,014 --> 00:00:12,491 But like anything with three billion components, 5 00:00:12,515 --> 00:00:13,915 that gift is fragile. 6 00:00:14,815 --> 00:00:18,355 Sunlight, smoking, unhealthy eating, 7 00:00:18,379 --> 00:00:21,371 even spontaneous mistakes made by your cells, 8 00:00:21,395 --> 00:00:23,318 all cause changes to your genome. 9 00:00:24,942 --> 00:00:28,220 The most common kind of change in DNA 10 00:00:28,244 --> 00:00:32,473 is the simple swap of one letter, or base, such as C, 11 00:00:32,497 --> 00:00:35,738 with a different letter, such as T, G or A. 12 00:00:36,744 --> 00:00:40,117 In any day, the cells in your body will collectively accumulate 13 00:00:40,141 --> 00:00:44,977 billions of these single-letter swaps, which are also called "point mutations." 14 00:00:46,147 --> 00:00:48,678 Now, most of these point mutations are harmless. 15 00:00:48,702 --> 00:00:49,860 But every now and then, 16 00:00:49,884 --> 00:00:53,877 a point mutation disrupts an important capability in a cell 17 00:00:53,901 --> 00:00:57,256 or causes a cell to misbehave in harmful ways. 18 00:00:58,099 --> 00:01:01,098 If that mutation were inherited from your parents 19 00:01:01,122 --> 00:01:03,782 or occurred early enough in your development, 20 00:01:03,806 --> 00:01:06,772 then the result would be that many or all of your cells 21 00:01:06,796 --> 00:01:08,708 contain this harmful mutation. 22 00:01:09,153 --> 00:01:12,423 And then you would be one of hundreds of millions of people 23 00:01:12,447 --> 00:01:14,058 with a genetic disease, 24 00:01:14,082 --> 00:01:17,085 such as sickle cell anemia or progeria 25 00:01:17,109 --> 00:01:20,230 or muscular dystrophy or Tay-Sachs disease. 26 00:01:22,225 --> 00:01:25,407 Grievous genetic diseases caused by point mutations 27 00:01:25,431 --> 00:01:27,424 are especially frustrating, 28 00:01:27,448 --> 00:01:30,352 because we often know the exact single-letter change 29 00:01:30,376 --> 00:01:34,576 that causes the disease and, in theory, could cure the disease. 30 00:01:35,268 --> 00:01:38,117 Millions suffer from sickle cell anemia 31 00:01:38,141 --> 00:01:41,212 because they have a single A to T point mutations 32 00:01:41,236 --> 00:01:43,597 in both copies of their hemoglobin gene. 33 00:01:45,529 --> 00:01:48,661 And children with progeria are born with a T 34 00:01:48,685 --> 00:01:50,853 at a single position in their genome 35 00:01:50,877 --> 00:01:52,276 where you have a C, 36 00:01:53,125 --> 00:01:56,564 with the devastating consequence that these wonderful, bright kids 37 00:01:56,588 --> 00:02:00,564 age very rapidly and pass away by about age 14. 38 00:02:02,358 --> 00:02:04,041 Throughout the history of medicine, 39 00:02:04,065 --> 00:02:07,125 we have not had a way to efficiently correct point mutations 40 00:02:07,149 --> 00:02:08,918 in living systems, 41 00:02:08,942 --> 00:02:12,142 to change that disease-causing T back into a C. 42 00:02:13,482 --> 00:02:15,450 Perhaps until now. 43 00:02:15,474 --> 00:02:19,664 Because my laboratory recently succeeded in developing such a capability, 44 00:02:19,688 --> 00:02:21,488 which we call "base editing." 45 00:02:23,277 --> 00:02:25,301 The story of how we developed base editing 46 00:02:25,325 --> 00:02:27,999 actually begins three billion years ago. 47 00:02:29,055 --> 00:02:31,715 We think of bacteria as sources of infection, 48 00:02:31,739 --> 00:02:35,053 but bacteria themselves are also prone to being infected, 49 00:02:35,077 --> 00:02:36,984 in particular, by viruses. 50 00:02:37,871 --> 00:02:40,022 So about three billion years ago, 51 00:02:40,046 --> 00:02:43,926 bacteria evolved a defense mechanism to fight viral infection. 52 00:02:45,649 --> 00:02:48,434 That defense mechanism is now better known as CRISPR. 53 00:02:49,008 --> 00:02:51,833 And the warhead in CRISPR is this purple protein 54 00:02:51,857 --> 00:02:55,635 that acts like molecular scissors to cut DNA, 55 00:02:55,659 --> 00:02:58,087 breaking the double helix into two pieces. 56 00:02:59,323 --> 00:03:03,299 If CRISPR couldn't distinguish between bacterial and viral DNA, 57 00:03:03,323 --> 00:03:05,562 it wouldn't be a very useful defense system. 58 00:03:06,315 --> 00:03:09,100 But the most amazing feature of CRISPR 59 00:03:09,124 --> 00:03:14,161 is that the scissors can be programmed to search for, 60 00:03:14,185 --> 00:03:16,608 bind to and cut 61 00:03:16,632 --> 00:03:19,370 only a specific DNA sequence. 62 00:03:20,911 --> 00:03:24,308 So when a bacterium encounters a virus for the first time, 63 00:03:24,332 --> 00:03:27,705 it can store a small snippet of that virus's DNA 64 00:03:27,729 --> 00:03:31,373 for use as a program to direct the CRISPR scissors 65 00:03:31,397 --> 00:03:34,933 to cut that viral DNA sequence during a future infection. 66 00:03:35,778 --> 00:03:40,691 Cutting a virus's DNA messes up the function of the cut viral gene, 67 00:03:40,715 --> 00:03:43,417 and therefore disrupts the virus's life cycle. 68 00:03:46,059 --> 00:03:50,860 Remarkable researchers including Emmanuelle Charpentier, George Church, 69 00:03:50,884 --> 00:03:53,537 Jennifer Doudna and Feng Zhang 70 00:03:53,561 --> 00:03:57,530 showed six years ago how CRISPR scissors could be programmed 71 00:03:57,554 --> 00:04:00,141 to cut DNA sequences of our choosing, 72 00:04:00,165 --> 00:04:02,534 including sequences in your genome, 73 00:04:02,558 --> 00:04:05,901 instead of the viral DNA sequences chosen by bacteria. 74 00:04:06,550 --> 00:04:09,084 But the outcomes are actually similar. 75 00:04:09,606 --> 00:04:12,074 Cutting a DNA sequence in your genome 76 00:04:12,098 --> 00:04:16,225 also disrupts the function of the cut gene, typically, 77 00:04:16,997 --> 00:04:21,464 by causing the insertion and deletion of random mixtures of DNA letters 78 00:04:21,488 --> 00:04:22,641 at the cut site. 79 00:04:24,625 --> 00:04:28,506 Now, disrupting genes can be very useful for some applications. 80 00:04:30,005 --> 00:04:34,306 But for most point mutations that cause genetic diseases, 81 00:04:34,330 --> 00:04:38,687 simply cutting the already-mutated gene won't benefit patients, 82 00:04:38,711 --> 00:04:42,679 because the function of the mutated gene needs to be restored, 83 00:04:42,703 --> 00:04:44,318 not further disrupted. 84 00:04:45,259 --> 00:04:48,141 So cutting this already-mutated hemoglobin gene 85 00:04:48,165 --> 00:04:50,688 that causes sickle cell anemia 86 00:04:50,712 --> 00:04:54,228 won't restore the ability of patients to make healthy red blood cells. 87 00:04:55,631 --> 00:04:59,972 And while we can sometimes introduce new DNA sequences into cells 88 00:04:59,996 --> 00:05:03,417 to replace the DNA sequences surrounding a cut site, 89 00:05:03,441 --> 00:05:07,765 that process, unfortunately, doesn't work in most types of cells, 90 00:05:07,789 --> 00:05:10,230 and the disrupted gene outcomes still predominate. 91 00:05:12,297 --> 00:05:14,479 Like many scientists, I've dreamed of a future 92 00:05:14,503 --> 00:05:17,277 in which we might be able to treat or maybe even cure 93 00:05:17,301 --> 00:05:18,672 human genetic diseases. 94 00:05:19,135 --> 00:05:22,936 But I saw the lack of a way to fix point mutations, 95 00:05:22,960 --> 00:05:25,984 which cause most human genetic diseases, 96 00:05:26,008 --> 00:05:28,396 as a major problem standing in the way. 97 00:05:29,434 --> 00:05:32,102 Being a chemist, I began working with my students 98 00:05:32,126 --> 00:05:37,061 to develop ways on performing chemistry directly on an individual DNA base, 99 00:05:37,085 --> 00:05:42,704 to truly fix, rather than disrupt, the mutations that cause genetic diseases. 100 00:05:44,522 --> 00:05:47,070 The results of our efforts are molecular machines 101 00:05:47,094 --> 00:05:48,482 called "base editors." 102 00:05:49,618 --> 00:05:55,093 Base editors use the programmable searching mechanism of CRISPR scissors, 103 00:05:55,117 --> 00:05:58,053 but instead of cutting the DNA, 104 00:05:58,077 --> 00:06:01,018 they directly convert one base to another base 105 00:06:01,042 --> 00:06:03,295 without disrupting the rest of the gene. 106 00:06:04,674 --> 00:06:08,832 So if you think of naturally occurring CRISPR proteins as molecular scissors, 107 00:06:08,856 --> 00:06:11,642 you can think of base editors as pencils, 108 00:06:11,666 --> 00:06:15,162 capable of directly rewriting one DNA letter into another 109 00:06:16,098 --> 00:06:19,901 by actually rearranging the atoms of one DNA base 110 00:06:19,925 --> 00:06:22,259 to instead become a different base. 111 00:06:23,513 --> 00:06:25,689 Now, base editors don't exist in nature. 112 00:06:26,683 --> 00:06:29,913 In fact, we engineered the first base editor, shown here, 113 00:06:29,937 --> 00:06:31,294 from three separate proteins 114 00:06:31,318 --> 00:06:33,548 that don't even come from the same organism. 115 00:06:34,151 --> 00:06:39,248 We started by taking CRISPR scissors and disabling the ability to cut DNA 116 00:06:39,272 --> 00:06:43,811 while retaining its ability to search for and bind a target DNA sequence 117 00:06:43,835 --> 00:06:45,369 in a programmed manner. 118 00:06:46,351 --> 00:06:49,188 To those disabled CRISPR scissors, shown in blue, 119 00:06:49,212 --> 00:06:51,720 we attached a second protein in red, 120 00:06:51,744 --> 00:06:56,045 which performs a chemical reaction on the DNA base C, 121 00:06:56,069 --> 00:06:59,402 converting it into a base that behaves like T. 122 00:07:00,958 --> 00:07:04,100 Third, we had to attach to the first two proteins 123 00:07:04,124 --> 00:07:05,474 the protein shown in purple, 124 00:07:05,498 --> 00:07:09,098 which protects the edited base from being removed by the cell. 125 00:07:10,466 --> 00:07:13,308 The net result is an engineered three-part protein 126 00:07:13,332 --> 00:07:17,450 that for the first time allows us to convert Cs into Ts 127 00:07:17,474 --> 00:07:19,637 at specified locations in the genome. 128 00:07:21,490 --> 00:07:24,522 But even at this point, our work was only half done. 129 00:07:24,546 --> 00:07:27,172 Because in order to be stable in cells, 130 00:07:27,196 --> 00:07:30,855 the two strands of a DNA double helix have to form base pairs. 131 00:07:32,125 --> 00:07:35,783 And because C only pairs with G, 132 00:07:35,807 --> 00:07:38,809 and T only pairs with A, 133 00:07:39,752 --> 00:07:44,598 simply changing a C to a T on one DNA strand creates a mismatch, 134 00:07:44,622 --> 00:07:47,471 a disagreement between the two DNA strands 135 00:07:47,495 --> 00:07:51,763 that the cell has to resolve by deciding which strand to replace. 136 00:07:53,149 --> 00:07:57,490 We realized that we could further engineer this three-part protein 137 00:07:58,649 --> 00:08:02,515 to flag the nonedited strand as the one to be replaced 138 00:08:02,539 --> 00:08:04,450 by nicking that strand. 139 00:08:05,276 --> 00:08:07,805 This little nick tricks the cell 140 00:08:07,829 --> 00:08:12,776 into replacing the nonedited G with an A 141 00:08:12,800 --> 00:08:15,125 as it remakes the nicked strand, 142 00:08:15,149 --> 00:08:19,180 thereby completing the conversion of what used to be a C-G base pair 143 00:08:19,204 --> 00:08:21,500 into a stable T-A base pair. 144 00:08:24,585 --> 00:08:26,136 After several years of hard work 145 00:08:26,160 --> 00:08:30,141 led by a former post doc in the lab, Alexis Komor, 146 00:08:30,165 --> 00:08:33,347 we succeeded in developing this first class of base editor, 147 00:08:33,371 --> 00:08:37,037 which converts Cs into Ts and Gs into As 148 00:08:37,061 --> 00:08:39,220 at targeted positions of our choosing. 149 00:08:40,633 --> 00:08:45,863 Among the more than 35,000 known disease-associated point mutations, 150 00:08:45,887 --> 00:08:49,672 the two kinds of mutations that this first base editor can reverse 151 00:08:49,696 --> 00:08:55,839 collectively account for about 14 percent or 5,000 or so pathogenic point mutations. 152 00:08:56,593 --> 00:09:01,363 But correcting the largest fraction of disease-causing point mutations 153 00:09:01,387 --> 00:09:05,022 would require developing a second class of base editor, 154 00:09:05,046 --> 00:09:09,132 one that could convert As into Gs or Ts into Cs. 155 00:09:10,846 --> 00:09:14,573 Led by Nicole Gaudelli, a former post doc in the lab, 156 00:09:14,597 --> 00:09:17,719 we set out to develop this second class of base editor, 157 00:09:17,743 --> 00:09:23,870 which, in theory, could correct up to almost half of pathogenic point mutations, 158 00:09:23,894 --> 00:09:27,805 including that mutation that causes the rapid-aging disease progeria. 159 00:09:30,107 --> 00:09:33,274 We realized that we could borrow, once again, 160 00:09:33,298 --> 00:09:37,366 the targeting mechanism of CRISPR scissors 161 00:09:37,390 --> 00:09:42,551 to bring the new base editor to the right site in a genome. 162 00:09:43,543 --> 00:09:46,635 But we quickly encountered an incredible problem; 163 00:09:47,896 --> 00:09:50,324 namely, there is no protein 164 00:09:50,348 --> 00:09:54,400 that's known to convert A into G or T into C 165 00:09:54,424 --> 00:09:55,585 in DNA. 166 00:09:56,760 --> 00:09:58,926 Faced with such a serious stumbling block, 167 00:09:58,950 --> 00:10:01,482 most students would probably look for another project, 168 00:10:01,506 --> 00:10:03,246 if not another research advisor. 169 00:10:03,270 --> 00:10:04,434 (Laughter) 170 00:10:04,458 --> 00:10:06,400 But Nicole agreed to proceed with a plan 171 00:10:06,424 --> 00:10:09,091 that seemed wildly ambitious at the time. 172 00:10:09,966 --> 00:10:12,305 Given the absence of a naturally occurring protein 173 00:10:12,329 --> 00:10:14,490 that performs the necessary chemistry, 174 00:10:14,514 --> 00:10:17,950 we decided we would evolve our own protein in the laboratory 175 00:10:17,974 --> 00:10:21,809 to convert A into a base that behaves like G, 176 00:10:21,833 --> 00:10:26,660 starting from a protein that performs related chemistry on RNA. 177 00:10:27,230 --> 00:10:31,164 We set up a Darwinian survival-of-the-fittest selection system 178 00:10:31,188 --> 00:10:35,180 that explored tens of millions of protein variants 179 00:10:35,204 --> 00:10:37,222 and only allowed those rare variants 180 00:10:37,246 --> 00:10:40,467 that could perform the necessary chemistry to survive. 181 00:10:41,883 --> 00:10:44,271 We ended up with a protein shown here, 182 00:10:44,295 --> 00:10:47,152 the first that can convert A in DNA 183 00:10:47,176 --> 00:10:49,268 into a base that resembles G. 184 00:10:49,292 --> 00:10:50,895 And when we attached that protein 185 00:10:50,919 --> 00:10:53,490 to the disabled CRISPR scissors, shown in blue, 186 00:10:53,514 --> 00:10:55,522 we produced the second base editor, 187 00:10:55,546 --> 00:10:58,641 which converts As into Gs, 188 00:10:58,665 --> 00:11:02,506 and then uses the same strand-nicking strategy 189 00:11:02,530 --> 00:11:04,450 that we used in the first base editor 190 00:11:04,474 --> 00:11:09,939 to trick the cell into replacing the nonedited T with a C 191 00:11:09,963 --> 00:11:11,638 as it remakes that nicked strand, 192 00:11:11,662 --> 00:11:15,833 thereby completing the conversion of an A-T base pair to a G-C base pair. 193 00:11:16,845 --> 00:11:18,892 (Applause) 194 00:11:18,916 --> 00:11:20,086 Thank you. 195 00:11:20,110 --> 00:11:23,467 (Applause) 196 00:11:23,491 --> 00:11:25,826 As an academic scientist in the US, 197 00:11:25,850 --> 00:11:27,997 I'm not used to being interrupted by applause. 198 00:11:28,021 --> 00:11:31,172 (Laughter) 199 00:11:31,196 --> 00:11:35,601 We developed these first two classes of base editors 200 00:11:35,625 --> 00:11:38,399 only three years ago and one and a half years ago. 201 00:11:39,267 --> 00:11:40,815 But even in that short time, 202 00:11:40,839 --> 00:11:44,561 base editing has become widely used by the biomedical research community. 203 00:11:45,776 --> 00:11:50,141 Base editors have been sent more than 6,000 times 204 00:11:50,165 --> 00:11:54,036 at the request of more than 1,000 researchers around the globe. 205 00:11:55,475 --> 00:11:58,991 A hundred scientific research papers have been published already, 206 00:11:59,015 --> 00:12:02,743 using base editors in organisms ranging from bacteria 207 00:12:02,767 --> 00:12:04,901 to plants to mice to primates. 208 00:12:07,950 --> 00:12:09,557 While base editors are too new 209 00:12:09,581 --> 00:12:12,466 to have already entered human clinical trials, 210 00:12:12,490 --> 00:12:17,612 scientists have succeeded in achieving a critical milestone towards that goal 211 00:12:17,636 --> 00:12:20,485 by using base editors in animals 212 00:12:20,509 --> 00:12:24,418 to correct point mutations that cause human genetic diseases. 213 00:12:25,815 --> 00:12:26,966 For example, 214 00:12:26,990 --> 00:12:30,783 a collaborative team of scientists led by Luke Koblan and Jon Levy, 215 00:12:30,807 --> 00:12:33,220 two additional students in my lab, 216 00:12:33,244 --> 00:12:37,363 recently used a virus to deliver that second base editor 217 00:12:37,387 --> 00:12:39,577 into a mouse with progeria, 218 00:12:39,601 --> 00:12:43,458 changing that disease-causing T back into a C 219 00:12:43,482 --> 00:12:47,588 and reversing its consequences at the DNA, RNA and protein levels. 220 00:12:48,880 --> 00:12:51,626 Base editors have also been used in animals 221 00:12:51,650 --> 00:12:54,574 to reverse the consequence of tyrosinemia, 222 00:12:55,642 --> 00:12:59,260 beta thalassemia, muscular dystrophy, 223 00:12:59,284 --> 00:13:02,974 phenylketonuria, a congenital deafness 224 00:13:02,998 --> 00:13:04,937 and a type of cardiovascular disease -- 225 00:13:04,961 --> 00:13:09,823 in each case, by directly correcting a point mutation 226 00:13:09,847 --> 00:13:12,400 that causes or contributes to the disease. 227 00:13:13,688 --> 00:13:15,744 In plants, base editors have been used 228 00:13:15,768 --> 00:13:19,840 to introduce individual single DNA letter changes 229 00:13:19,864 --> 00:13:21,832 that could lead to better crops. 230 00:13:22,253 --> 00:13:26,842 And biologists have used base editors to probe the role of individual letters 231 00:13:26,866 --> 00:13:29,683 in genes associated with diseases such as cancer. 232 00:13:31,046 --> 00:13:35,613 Two companies I cofounded, Beam Therapeutics and Pairwise Plants, 233 00:13:35,637 --> 00:13:39,462 are using base editing to treat human genetic diseases 234 00:13:39,486 --> 00:13:41,092 and to improve agriculture. 235 00:13:41,953 --> 00:13:43,919 All of these applications of base editing 236 00:13:43,943 --> 00:13:47,037 have taken place in less than the past three years: 237 00:13:47,061 --> 00:13:49,425 on the historical timescale of science, 238 00:13:49,449 --> 00:13:50,731 the blink of an eye. 239 00:13:52,657 --> 00:13:53,910 Additional work lies ahead 240 00:13:53,934 --> 00:13:56,966 before base editing can realize its full potential 241 00:13:56,990 --> 00:14:00,604 to improve the lives of patients with genetic diseases. 242 00:14:01,244 --> 00:14:04,024 While many of these diseases are thought to be treatable 243 00:14:04,048 --> 00:14:05,897 by correcting the underlying mutation 244 00:14:05,921 --> 00:14:09,437 in even a modest fraction of cells in an organ, 245 00:14:09,461 --> 00:14:12,437 delivering molecular machines like base editors 246 00:14:12,461 --> 00:14:14,228 into cells in a human being 247 00:14:14,252 --> 00:14:15,421 can be challenging. 248 00:14:16,962 --> 00:14:20,335 Co-opting nature's viruses to deliver base editors 249 00:14:20,359 --> 00:14:22,557 instead of the molecules that give you a cold 250 00:14:22,581 --> 00:14:25,268 is one of several promising delivery strategies 251 00:14:25,292 --> 00:14:26,951 that's been successfully used. 252 00:14:28,268 --> 00:14:30,633 Continuing to develop new molecular machines 253 00:14:30,657 --> 00:14:32,525 that can make all of the remaining ways 254 00:14:32,549 --> 00:14:35,441 to convert one base pair to another base pair 255 00:14:35,465 --> 00:14:39,845 and that minimize unwanted editing at off-target locations in cells 256 00:14:39,869 --> 00:14:41,069 is very important. 257 00:14:41,782 --> 00:14:46,488 And engaging with other scientists, doctors, ethicists and governments 258 00:14:46,512 --> 00:14:51,303 to maximize the likelihood that base editing is applied thoughtfully, 259 00:14:51,327 --> 00:14:53,708 safely and ethically, 260 00:14:53,732 --> 00:14:55,732 remains a critical obligation. 261 00:14:57,525 --> 00:14:59,136 These challenges notwithstanding, 262 00:14:59,160 --> 00:15:02,815 if you had told me even just five years ago 263 00:15:02,839 --> 00:15:04,490 that researchers around the globe 264 00:15:04,514 --> 00:15:08,053 would be using laboratory-evolved molecular machines 265 00:15:08,077 --> 00:15:11,074 to directly convert an individual base pair 266 00:15:11,098 --> 00:15:12,280 to another base pair 267 00:15:12,304 --> 00:15:14,923 at a specified location in the human genome 268 00:15:14,947 --> 00:15:18,772 efficiently and with a minimum of other outcomes, 269 00:15:18,796 --> 00:15:19,964 I would have asked you, 270 00:15:19,988 --> 00:15:22,462 "What science-fiction novel are you reading?" 271 00:15:23,706 --> 00:15:27,166 Thanks to a relentlessly dedicated group of students 272 00:15:27,190 --> 00:15:31,650 who were creative enough to engineer what we could design ourselves 273 00:15:31,674 --> 00:15:34,599 and brave enough to evolve what we couldn't, 274 00:15:34,623 --> 00:15:39,663 base editing has begun to transform that science-fiction-like aspiration 275 00:15:39,687 --> 00:15:41,544 into an exciting new reality, 276 00:15:42,250 --> 00:15:45,481 one in which the most important gift we give our children 277 00:15:45,505 --> 00:15:48,530 may not only be three billion letters of DNA, 278 00:15:48,554 --> 00:15:51,664 but also the means to protect and repair them. 279 00:15:52,339 --> 00:15:53,490 Thank you. 280 00:15:53,514 --> 00:15:58,016 (Applause) 281 00:15:58,040 --> 00:15:59,190 Thank you.