William Shih | Self Assembling DNA Nanostructures @ Molecular Systems Design Workshop 2023_transcription

[00:00] Thank you so much again for joining Will.

[00:02] And it's always a pleasure to have you at these workshops.

[00:05] And we're very excited for your talk.

[00:06] Thanks.

[00:07] I'm going to be telling you about our work

[00:09] in crisscross polymerization, led by talented junior

[00:13] investigators.

[00:14] Two of them are here, Chris Wintersinger and Anastasia

[00:17] Irsheva.

[00:19] Third key contributor, Dio Mena, was a former foresight fellow.

[00:22] And then we also have Jie Dong, who just took a faculty

[00:25] position in Wuhan.

[00:26] And I'll be telling you about a reminder of what

[00:28] I talked about last year of building

[00:31] what we call megastructures from building blocks that

[00:34] are DNA origami.

[00:36] And then I'll be sharing very briefly some exciting advances

[00:39] that Anastasia led in trying to get exponential growth

[00:43] from single-stranded slots.

[00:46] And the starting point is thinking about DNA origami.

[00:49] And what I love the most about DNA origami

[00:51] is that you have absolute control

[00:53] over the number of structures that form

[00:56] based on the number of scaffold strands that you add.

[01:00] But what if you wanted to build structures that

[01:02] are far bigger than a scaffold?

[01:06] Because the problem is that with DNA origami, half the mass

[01:09] has to be the scaffold.

[01:10] And so that's the problem we set out to solve.

[01:12] Why would you want to do that?

[01:13] Well, we can envision a future where

[01:15] we can program the self-assembly of squishy robots

[01:18] the size of a cell that might be a million times bigger

[01:22] than an individual DNA origami.

[01:23] We don't have a scaffold half the size of a cell.

[01:27] Contrarily, another point is we would

[01:29] like to be able to build ultra-sensitive enzyme-free

[01:32] diagnostics.

[01:33] We're trying to convert a single analyte

[01:36] into vast polymerization of single-stranded building blocks

[01:39] into double-stranded products.

[01:40] So build a lot of material for amplification.

[01:43] And again, this is something beyond the capability

[01:45] of conventional DNA origami.

[01:48] So I'm going to introduce this concept of crisscross

[01:51] polymerization, where the goal is

[01:52] to program some kind of building block that's

[01:56] thermodynamically specified to assemble into desired shape,

[01:59] but is completely blocked kinetically from ever doing so,

[02:03] except when you add a small seed that is much tinier

[02:06] than the final assembly.

[02:08] And so the basic ingredient is specifications

[02:10] you have a building block through these elongated slats

[02:13] with many specific glues on them.

[02:15] In this case, these are DNA origami slats.

[02:17] Each one of these long things is one DNA origami.

[02:19] It has 32 sequence-specific glues on it.

[02:23] We program them so that they can make interactions

[02:25] with each other at roughly 90 degrees.

[02:28] And it's programmed in the final assembly to have 32 neighbors.

[02:32] But it can only make one contact with any one neighbor.

[02:37] Importantly, we operate this at a temperature

[02:39] where you need close to half of the bonds, or 16 in this case,

[02:43] for stable assembly.

[02:44] So if you only made eight bonds or nine bonds,

[02:47] it'll transiently come together.

[02:49] And then it'll fall apart.

[02:51] And that's really the key of designing a process that

[02:54] will never ever start kinetically,

[02:56] even though it's favored thermodynamically.

[02:58] So we can imagine a growth trajectory where we have two

[03:00] slats come together.

[03:01] The energy deficit is represented.

[03:03] One represents the cost.

[03:04] One unit is the cost of fixing one of those building blocks

[03:08] into exactly the right position entropically.

[03:11] And so just bringing two things together,

[03:12] it's making only one bond.

[03:14] Well, that's not favored.

[03:15] So that's very unstable.

[03:18] But what's even more unstable is trying to add two more slats.

[03:21] Because every time you add another slat,

[03:22] you're making fewer than those 16 bonds.

[03:24] So you're just writing up the energy landscape,

[03:28] climbing Mount Improbable, you might call it,

[03:31] to make successively more preposterously unlikely

[03:34] structures.

[03:35] Until you finally get to this structure

[03:37] we call the critical nucleus, where finally once you achieve

[03:39] this mountain peak, you can start adding building blocks

[03:42] where each added building block is making 16 bonds

[03:45] and therefore is stable.

[03:47] But the situation is to get to this peak,

[03:49] we need to climb an energy barrier of 16 units.

[03:53] In other words, the entropic price of fixing 16 building

[03:56] blocks in exactly the right orientation

[03:58] without anything to hold them together.

[04:00] Which will basically, you could fill the ocean with these things

[04:02] and then wait the age of the universe.

[04:04] It's just not going to happen as long as you're

[04:05] close to that reversible temperature.

[04:07] And then that will buy us the opportunity

[04:09] to control this process by putting in a little deus

[04:13] ecmakina, a single denier origami seed that's

[04:15] using very strong bonds to position those first 16

[04:19] horizontal structures.

[04:21] It pays that entropic price and now the process

[04:23] can go off and running, where we can imagine starting

[04:26] with this bidirectional growth front,

[04:28] we're adding 16 slats at a time, and then we

[04:31] can towel around into some desired structure.

[04:34] And so we can see a transmission electron micrograph here

[04:36] of an object approaching a micrometer in scale

[04:39] with 192 origami.

[04:41] We published this work in Nature and Nanotechnology

[04:43] and the station designed this beautiful cover.

[04:46] So these structures are made from over 1,000 denier origami.

[04:50] Each are different.

[04:51] Overall dimensions of two microns by two microns.

[04:55] And right now we're working on trying

[04:57] to extend this to make it in more order magnitude bigger,

[05:00] or maybe two orders magnitude bigger, try to make 3D.

[05:02] I don't have time to go into those details right now.

[05:04] Happy to talk about it later.

[05:06] I just want to briefly point out I stuck this

[05:08] into one of the working documents,

[05:09] this idea of how to use crisscross for implosion.

[05:12] So I'm not really going to talk about it here.

[05:14] I'm just going to briefly mention that the point of this

[05:17] is that it's how do you build pre-assembled three

[05:20] to five nanometer building blocks into much bigger

[05:22] structures if that's what you want.

[05:24] This is agnostic about how you make those smaller building

[05:26] blocks.

[05:27] It's really about that next step.

[05:29] And so now I'd like to briefly talk about crisscross

[05:32] with single-stranded building blocks

[05:33] for the purpose of signal amplification for diagnostics.

[05:38] And you can repeat this crisscross process

[05:41] with the same advantages of kinetic blockade,

[05:43] but with now tiny, tiny building blocks that

[05:45] are just oligonucleotides.

[05:47] And the oligonucleotides are just binding to each other

[05:49] with half a turn.

[05:51] That's one binding site.

[05:52] And the same process where you specify that at equilibrium,

[05:56] all the binding sites are occupied

[05:57] by a bunch of different binding partners,

[06:01] but that kinetically it never gets off the starting point

[06:04] because of this entropic barrier.

[06:06] So you're writing up this climbing non-improbable,

[06:09] and it's so rare to get to the peak

[06:10] that this process basically never happens.

[06:13] But now we have single-stranded building blocks,

[06:14] so the assembly can happen much more quickly.

[06:16] We have these building blocks at very high concentration.

[06:19] And again, if we introduce a seed that somehow

[06:21] is linked to the analyte that we're trying to detect,

[06:23] in this case, Sedinia ergami, then we

[06:25] can trigger the process in a controlled fashion.

[06:27] So it nucleates the assembly, and then it grows.

[06:30] We have ideas of how to convert any analyte into such a seed.

[06:34] Happy to talk about it later.

[06:36] And really, the brand new idea that Anastasia

[06:39] has been developing that we're very excited about

[06:41] is how do we actually get this to grow not just linearly,

[06:45] but exponentially?

[06:46] It turns out linear growth is too slow in order

[06:48] to get truly in-fold amplification.

[06:50] And so the scheme, we came up with a strand displacement

[06:53] based scheme in order to cut these thick filaments.

[06:55] And that might sound very difficult,

[06:57] but we figured out a scheme that works.

[07:00] And it's based on this notion of toe-hole mediated strand

[07:02] displacement that many of you might be familiar with,

[07:04] that you can think of this as how do we

[07:06] break the blue and the black strands apart?

[07:09] That's kind of cutting this into two pieces.

[07:11] And the way that we do this is by introducing

[07:14] a single-stranded toe-hold that the displacing strand can

[07:16] grab onto to nucleate, undergo branch migration,

[07:19] and kick the black off.

[07:20] This is how we split the blue and black into two strands.

[07:24] And then in this case, in the middle,

[07:26] is taking this analogy with one of our crisscross structures.

[07:29] So here we're asking the green strand in this middle thing

[07:33] to come in and displace the top slot away.

[07:36] And so it kind of comes in.

[07:37] And I'm not going to describe this in great detail,

[07:40] but hopefully you can kind of see

[07:42] the analogy between the middle and the top.

[07:44] I'll have you talk about it later.

[07:45] And then here's kind of the mind bender,

[07:48] is we can extend the process to now involve

[07:50] coordinated strand displacement by a set of these invader

[07:53] strands that coordinate in a way to basically separate

[07:57] the bottom block from the northeast block.

[08:00] And in that way, we can have coordinated attack

[08:02] by many invader strands and then split this very thick object

[08:06] into two.

[08:07] And then we can incorporate that into our crisscross growth

[08:10] model where we're making these linear ribbons.

[08:13] We introduce these coordinated toeholds

[08:15] so you have a bunch of invader strands

[08:16] that can eat in from the north and from the east.

[08:20] And if you have invasion on two fronts,

[08:22] then you can basically get this thing to split

[08:24] in a stage of midnight animation showing this process.

[08:28] Again, you can download it from the Google Doc

[08:31] if you want to look at this in more detail.

[08:34] And in that way, we can get exponential amplification.

[08:39] And it kind of works,

[08:41] but we're still trying to improve on its performance.

[08:44] The latest is we have some kind of variant where we can get

[08:46] down to subatomol or limited detection,

[08:48] which we're very excited about.

[08:49] So in summary, this is a scheme with

[08:51] the single-stranded building blocks that we think will be

[08:53] extremely useful for enzyme-free,

[08:55] ultra-sensitive detection of things like pathogens,

[08:58] especially in the developing world.

[09:00] And I'll conclude by just reminding again

[09:03] that here specifically we're looking at how do we

[09:05] implement exponential conversion of

[09:07] single-stranded building blocks into

[09:09] double-stranded products for the purpose of diagnostics.

[09:12] But on another scale, we're longitudinally interested,

[09:14] how do we advance the field of

[09:16] biomelecular nanotechnology to sustain

[09:19] an exponential trend in the maximum complexity

[09:22] of the structures that we can construct over time.

[09:25] And then another thing I'd just like to point out is that

[09:27] nothing in our crisscross cartoons

[09:29] has anything to do with DNA.

[09:30] We implemented this single-stranded DNA.

[09:33] We implemented with DNA origami building blocks.

[09:36] We challenge people in the community to

[09:38] repeat this algorithm with other materials such as

[09:40] proteins or maybe non-biological polymers as well.

[09:43] So, thanks.

[09:44] Interesting. Thank you.

[09:48] Okay. We have time for one question.

[09:54] Really, really simple question,

[09:56] particularly related to the last point.

[09:58] So, how does one of your deficit units compare to KT?

[10:02] So, you had your energy deficit units,

[10:06] how do they compare to KT?

[10:08] Yeah. So, if you imagine, let's say your building blocks

[10:10] were at one micromolar concentration,

[10:13] and then the effective concentration within the ribbon,

[10:16] let's just say for argument's sake it's one molar.

[10:18] So, that's a six order of magnitude concentration to

[10:22] pull it out of micromolar bulk solution into the ribbon.

[10:26] So, each order of magnitude corresponds to 2.3 KT.

[10:30] So, in the case of the origami,

[10:33] there's 16 units,

[10:36] 16 times 2.3 is like 38 KT.

[10:39] That's huge. That's like two ATP's worth of energy.

[10:42] It's never going to happen. It's not going to happen.

[10:47] So, how do you think that

[10:49] your work could fit into their respective architecture?

[10:52] Yeah. So, I have this figure that I embedded in the work,

[10:55] the workpiece and also it's here.

[10:58] You can download it.

[10:59] So, again, this is a scheme for making

[11:02] an exploded view of pre-assembled 3-5 nanometer building blocks

[11:07] that each would fit on the node on the crisscross grid.

[11:11] Then the advantage here is that you could

[11:14] hierarchically construct each slot with,

[11:17] let's say, 32 guests on it.

[11:18] So, first you have some kind of mechanism,

[11:20] your first level printer makes

[11:22] your 3-5 nanometer building blocks.

[11:24] Then your second level process fabricates the slats

[11:27] with 32 of them loaded up in a linear fashion.

[11:30] This is the third step where we now assemble

[11:34] those 32 subunits together into

[11:36] a much bigger structure in an exploded view.

[11:39] Then we can also do funny,

[11:40] we have more access in this exploded view.

[11:42] Then when we're ready, we could program it to

[11:45] compact along the y-axis.

[11:47] So, this is just like a little grid structure.

[11:49] Hopefully, we've all seen these kind of

[11:51] the boxing glove thing that expands and contracts.

[11:53] So, contract in the y-axis.

[11:56] Then what we do is we remove the DNA scaffold and we've

[11:59] also inserted some rubber bands along the orthogonal axis.

[12:04] So, in the second step, we can take the structure that's only

[12:06] cross-linked along the y-axis and then we

[12:10] can compact it along the x-axis as well.

[12:13] It's probably confusing, but happy to talk about it more later.

[12:16] Wonderful. Thank you so much, Will.