Polyurethane is the latest polymer broken down by designer enzymes

Given a dozen hours, the enzyme can turn a foam pad into reusable chemicals.

See full article...

 

[HalfAHero]

What are the byproducts from this reaction, other than the building blocks of the polyurethane? What becomes of the enzyme when it no longer reacts with the polyurethane?

This all sounds very hopeful, but it must create some new problem to deal with, right? Hopefully a more manageable one.

 

[Molybdenum]

What are the byproducts from this reaction, other than the building blocks of the polyurethane? What becomes of the enzyme when it no longer reacts with the polyurethane?

This all sounds very hopeful, but it must create some new problem to deal with, right? Hopefully a more manageable one.

It might create some new problem, but it's possible that it's as close to an ideal solution as we'll ever get. Enzymes are catalysts, first and foremost, and are often highly selective for a certain specific reaction.

 

[skraft]

Would be interesting to see if sending the newly designed enzyme back to the protein databases / AlphaFold step results in an even better new enzyme. How many loops can we make through the process before the method starts to become unproductive?

 

[Don Reba]

Nice. The next step is to genetically engineer organisms that would produce this enzyme.

That’s the story behind a completely new enzyme that researchers developed to break down polyurethane, the polymer commonly used to make foam cushioning, among other things.

It is also commonly used as a wood finish and for waterproofing cement, such as in pools and parking garages.

 

[hillspuck]

What are the byproducts from this reaction, other than the building blocks of the polyurethane? What becomes of the enzyme when it no longer reacts with the polyurethane?

Isn't the point of an enzyme/catalyst that it doesn't get used up? So you'd have a puddle of those building blocks and enzyme.

 

[chaos215bar2]

Isn't the point of an enzyme/catalyst that it doesn't get used up? So you'd have a puddle of those building blocks and enzyme.

If it's perfectly, 100% stable. In reality there will be some rate of breakdown depending on conditions, but barring a freak coincidence, you're eventually going to wind up with a slurry of useless protein fragments and amino acids, which is about as benign as things get.

 

[LordSlinkySupreme]

Nice to see neural networks used for something potentially beneficial to everyone for once.

 

[JohnDeL]

Would be interesting to see if sending the newly designed enzyme back to the protein databases / AlphaFold step results in an even better new enzyme. How many loops can we make through the process before the method starts to become unproductive?

Typically, that tends to send the NN into a death spiral where it only focuses on the one solution. It is better/more effective to create variations of the solution and send them in with the original to see what drives the process to be more efficient/effective.

Sidenote: Thank you, John Timmer, for making the clear distinction between this type of (sometimes useful) AI and the (rarely if ever useful) LLM AIs!

 

[Mental Gear Reduction]

One thought that occurs is that polyurethane is really cheap, so recycling it might be too expensive without some kind of external intervention. Typically, bulk waste is considered someone else's problem, so this is the sort of thing that might need to be handled by government. You might, for instance, need regional facilities run by government entities.

Which Republicans will just instantly hate, on principle, so it's not likely to actually happen, most places.

 

[Jack56]

This is certainly interesting. Some observations.
First, the NN was useful for searching a large data space but the brain work was done by humans.
As for the utility of the enzyme, the key question is how much at what cost is needed to dispose of a given weight of polyurethane. Sounds expensive to use it as described. Short of engineering it into microorganisms or insects in the environment, will it ever be cheaper to dispose of polyurethane enzymatically rather than burn or bury it?

 

[twilightomni]

If it's perfectly, 100% stable. In reality there will be some rate of breakdown depending on conditions, but barring a freak coincidence, you're eventually going to wind up with a slurry of useless protein fragments and amino acids, which is about as benign as things get.

Enzymes and proteins have a natural lifecycle in cells and are broken down all the time. Garbage collecting old proteins is a major part of a cell’s internal upkeep.

A inorganic catalyst can be theoretically stable and completely reusable with reactivation, but an enzyme is a protein-based catalyst, and not guaranteed to be so.

 

[Fatesrider]

If we test the enzymes and only feed the useful one back, I guess the system is actually gaining in some sense “new information.” So, the loop maybe could go on indefinitely, right?

Proteins break down over time, so there's no continued assurance of function without renewing the supply. Since that would dilute the number and eventually replace the original proteins over time, no, it couldn't go on in a closed loop indefinitely. I'd GUESS only a comparative handful of cycles (as was mentioned in the article) before they wore out and needed to be replenished.

And if thats the case, then it's POSSIBLE, maybe, to filter out the proteins that are still viably useful and reuse them. but how much that would cost, the energy needed for it, and the effectiveness of that are all probably worse than just adding fresh, new proteins.

As long as the disposal of the worn out ones isn't polluting and can be done safely and cheaply, this is an interesting development. If nothing else, it can at least reduce the polyurethane to its components.

And has also been pointed out, whether that's a fiscally viable thing to do (considering how cheap the materials to create polyurethane are), recycling it back into new polyurethane may make polyurethane more expensive.

None of those questions have been answered, of course, so it's hard to say. But this could be a big thing, or just an interesting science experiment.

 

[DarthSlack]

This is certainly interesting. Some observations.
First, the NN was useful for searching a large data space but the brain work was done by humans.
As for the utility of the enzyme, the key question is how much at what cost is needed to dispose of a given weight of polyurethane. Sounds expensive to use it as described.

We already know how to make enzymes in bulk. It requires some level of specialized equipment, but generally if it can be cloned into a bacteria or yeast it's pretty easy to grow huge quantities in a bioreactor. Purification may add cost, but it's not clear that really pure enzymes would be needed for this.

Short of engineering it into microorganisms or insects in the environment, will it ever be cheaper to dispose of polyurethane enzymatically rather than burn or bury it?

In the absence of a carbon tax or some sort of government mandate, probably not. Externalizing all the disposal costs is cheap and easy on manufacturers, but hell on the rest of us.

 

[Eskimonster]

i hope this is a big thing, we need a win.

 

[SpaceCollector]

Nice to see neural networks used for something potentially beneficial to everyone for once.

Yes!

 

[Speaking from Europe...]

If it's perfectly, 100% stable. In reality there will be some rate of breakdown depending on conditions, but barring a freak coincidence, you're eventually going to wind up with a slurry of useless protein fragments and amino acids, which is about as benign as things get.

I want to believe... but I think people were this excited when Teflon came out as well...

 

[forkloo]

Nice to see neural networks used for something potentially beneficial to everyone for once.

One of the potential problems with this is chemical warfare. Neural Networks/AI are way better at developing chemical weapons than we are. It's already being used.

 

[breze]

Proteins break down over time, so there's no continued assurance of function without renewing the supply. Since that would dilute the number and eventually replace the original proteins over time, no, it couldn't go on in a closed loop indefinitely. I'd GUESS only a comparative handful of cycles (as was mentioned in the article) before they wore out and needed to be replenished.

And if thats the case, then it's POSSIBLE, maybe, to filter out the proteins that are still viably useful and reuse them. but how much that would cost, the energy needed for it, and the effectiveness of that are all probably worse than just adding fresh, new proteins.

As long as the disposal of the worn out ones isn't polluting and can be done safely and cheaply, this is an interesting development. If nothing else, it can at least reduce the polyurethane to its components.

And has also been pointed out, whether that's a fiscally viable thing to do (considering how cheap the materials to create polyurethane are), recycling it back into new polyurethane may make polyurethane more expensive.

None of those questions have been answered, of course, so it's hard to say. But this could be a big thing, or just an interesting science experiment.

I was responding to somebody talking about feeding the enzymes designs back into the ML training set. There are often concerns about models collapsing if they are fed too many other ML outputs as their inputs. I was pointing out that just adding enzyme designs that actually worked back to the training set might mitigate that concern because it adds a source of ground truth.

Sorry for the confusion.

 

[Kyuu]

But can it break down reporters getting butthurt and promoting their own snarky comments insulting people who make legitimate constructive criticism of their articles?

The irony is strong here.

 

[Jharm]

I know this process to get rid of the product requires money. What is difficult for me is that you are allowed to sell stuff at production cost without thinking about the downstream costs.

Somehow, if we want live in a relatively clean world, it has to be a part of the product costs. How to implement without creating a bureaucratic monster I don’t know.

But if waste facilities need to process these products then the producer needs to transfer money to that facility.

That we are still allowed to put products on the market we know will be a nightmare later, shows that we still have far to go.

 

[McTurkey]

I know this process to get rid of the product requires money. What is difficult for me is that you are allowed to sell stuff at production cost without thinking about the downstream costs.

Somehow, if we want live in a relatively clean world, it has to be a part of the product costs. How to implement without creating a bureaucratic monster I don’t know.

But if waste facilities need to process these products then the producer needs to transfer money to that facility.

That we are still allowed to put products on the market we know will be a nightmare later, shows that we still have far to go.

A straightforward solution includes a tax on plastic which goes into a cleanup fund that is used to administer recycling facilities. There would also need to be incentives for using plastics which are proven to be recyclable and for designing your products in a way that makes separating the plastic for recycling easier. Probably also fines for any business which knowingly fails to recycle their plastics, and free recycling for all business up to a certain threshold, with gradually higher costs by volume. All of that would pay for government-operated recycling facilities.

 

[aleph_nought]

Typically, that tends to send the NN into a death spiral where it only focuses on the one solution. It is better/more effective to create variations of the solution and send them in with the original to see what drives the process to be more efficient/effective.

Sidenote: Thank you, John Timmer, for making the clear distinction between this type of (sometimes useful) AI and the (rarely if ever useful) LLM AIs!

There's not much information in the article about the Pythia model being used other than it being a neural network. The full code is available on Github.

 

[FranzJoseph]

And they suggest that these approaches may tell us more about how to get functional proteins by focusing on forming a similar 3D structure.

Nice. Though I wonder about intrinsically disordered enzymes (proteins). These have no well‑defined 3D structure at all and there is no single functional pocket that pythia‑pocket could optimise for. Yet IDPs form a lot of the functional proteome (which in itself doesn't say much about them being the best possible solution, of course – could be just a local valley in the evolutionary space).

Is there any way NN could be used to look for and optimise these? Because just modelling them requires a pretty hefty physics simulation, as I understand it.

 

[RichyRoo]

One thought that occurs is that polyurethane is really cheap, so recycling it might be too expensive without some kind of external intervention. Typically, bulk waste is considered someone else's problem, so this is the sort of thing that might need to be handled by government. You might, for instance, need regional facilities run by government entities.

Which Republicans will just instantly hate, on principle, so it's not likely to actually happen, most places.

The rest of the world isn't run by Republicans, we will happily use it

 

[DarthSlack]

Do those kinds of libertarians exist anymore? I'm sure there's one or two still kicking around, but the vast majority seem to have migrated into a "FUCK YOU, GOT MINE" philosophy. And the only compensation they're good with are the ones that come out of someone else's pocket into theirs.

 

[paulfdietz]

What would be interesting would be to see the status of such an enzyme after it had been subjected to extensive directed evolution. This has been done in the drug industry to make custom enzymes for particularly troublesome steps in synthesis of drugs.

https://www.science.org/content/blog-post/chemical-biology-engineering-enzymes

 

[Veritas super omens]

I know this process to get rid of the product requires money. What is difficult for me is that you are allowed to sell stuff at production cost without thinking about the downstream costs.

Somehow, if we want live in a relatively clean world, it has to be a part of the product costs. How to implement without creating a bureaucratic monster I don’t know.

But if waste facilities need to process these products then the producer needs to transfer money to that facility.

That we are still allowed to put products on the market we know will be a nightmare later, shows that we still have far to go.

Ah... externalities...the joy of capitalists and the bane of civilization.

 

[iollmann]

What are the byproducts from this reaction, other than the building blocks of the polyurethane? What becomes of the enzyme when it no longer reacts with the polyurethane?

This all sounds very hopeful, but it must create some new problem to deal with, right? Hopefully a more manageable one.

It depends on the solution in which the reaction is done. Enzymes function in water naturally. If you did that, you’d probably end up with CO2 and the amino alcohol monomer precursor for the plastic. If the enzyme was rock stable and immobilized on a solid support then you could in principle put the polyurethane in one end of a filter bed with the enzyme attached and get monomer and CO2 out. A problem is the polyurethane is a solid, and resistant to dissolution or flow through a matrix because it is a polymer. Excessive heating or chemical cracking would fairly quickly denature the enzyme. So it is not clear to me how one would get that to work but humans can be creative.

The diethylene glycol method seems to result in a mess of polyols, mostly various low molecular weight polyethylene glycols, CO2, and presumably the afore mentioned amino alcohols unless these further decompose, though I couldn’t find any mention of that. You can read one description of that sort of work here:

https://www.sciencedirect.com/science/article/pii/S1026918525000095

 

[Orakle]

Cue the dystopian/apocalyptic SF novel regarding engineered plastic-eating bacteria (from 1971 no less):
https://astrofella.wordpress.com/2018/07/27/mutant-59-the-plastic-eater-kit-pedler-gerry-davis/

 

[DDopson]

I have a horrible (and somewhat joking) fear of this gene escaping into the wild, getting merged with some common soil bacteria, and then my floors needing be refinished every 6 months.

Even so, congrats on this discovery. It makes one wonder how far they will be able to take this approach. Can we make cost-effective industrial bio-reactors that digest foams back into nutrients? What's the maximum recoverable economic value and is the total value dominated by that or by the waste disposal savings? Can we engineer other enzymes to break down other materials that are currently hard to recycle? Can we develop a mix of enzymes such that we can shred whole items without separating their constituent parts and feed the shredded bits into the bioreactor for "recycling"? And then engineer some microbe to sustainable manufacture that suite of enzymes, ideally while being fed in the very bioreactor doing the recycling.

I once invested in a company doing something similar for food waste (which is infinitely simpler to digest), breaking it down to produce a liquid fertilizer. The process worked fairly well, but the margins on that sort of thing are fairly narrow and it's very sensitive to little details such as how much labor is required to remove boxes, wrappers, or other packing materials from the waste food before it can be pulverized and digested.

 

[Shavano]

One thought that occurs is that polyurethane is really cheap, so recycling it might be too expensive without some kind of external intervention. Typically, bulk waste is considered someone else's problem, so this is the sort of thing that might need to be handled by government. You might, for instance, need regional facilities run by government entities.

Yeah, I imagine synthesizing the enzyme is quite expensive now. If the ratio of enzyme needed to polyurethane it can degrade is high, it might still be viable. But as you say the trouble is getting people do to it when it's cheaper to chuck the used plastic in a landfill.

 

[paulfdietz]

If the goal is just to get the plastics to degrade, mixing them with wood chips or sawdust may be adequate. White rot fungi break down wood by producing active chemicals (like hydrogen peroxide and OH radicals) that attack a wide range of polymers, as they also attack the lignin holding together wood.

There are fungi that can grow on polyurethane as the sole carbon source.

https://en.wikipedia.org/wiki/Pestalotiopsis

 

[nosh]

What becomes of the enzyme when it no longer reacts with the polyurethane?

Enzymes are made from amino acids. And getting those amino acids back from the enzymes is something that life has gotten really good at the last few billion years...

 

[JohnDeL]

I have a horrible (and somewhat joking) fear of this gene escaping into the wild, getting merged with some common soil bacteria, and then my floors needing be refinished every 6 months.

Well, we've been doing this with insulin since 1978 and there haven't been any escapes yet.