Seeking testers for a practical protein production powder

Written By Kenneth Kostenbader
 
 

Recent AI-related breakthroughs in protein design have brought us closer to an era of Extremely Easy Enzyme Engineering (EEEE). This could be huge; imagine a completely carbon-neutral or even carbon-negative chemical industry free from petrochemicals, easy materials recycling and waste remediation, or environmentally-benign mining and mineral processing without harsh acids. I recently finished a PhD in enzyme chemistry, and I’m very inspired by the potential impact that enzyme design could have but I am finding that enzyme engineering is prohibitively expensive today. I’ve started Princeton Enzymes LLC to develop and sell a Practical Protein Production Powder (PPPP) that will make EEEE accessible to entrepreneurs and researchers who want to engineer their own enzymes for novel impactful application ideas.

For background I’ve illustrated some enzymes that can be found in nature and have been lightly modified for commercial applications:

 
 

Historically, enzymes have been too complex to design from scratch, so researchers and entrepreneurs have always used existing enzymes from nature as starting points for their desired biocatalyst applications. However, the recent surge of AI-related breakthroughs in protein design dramatically changes the field, and puts us on the doorstep of Extremely Easy Enzyme Engineering. There’s no shortage of impressive examples:

Promising enzyme engineering ideas

So now, with this sudden ability to design functioning enzymes from scratch, a huge number of commercial applications are within reach. Here’s my list of possible engineered enzymes that I’m excited about:

  • Carbonic anhydrase for carbon capture: American, Belgian, Canadian, and Italian researchers (and a healthy helping of companies) say that a more thermostable carbonic anhydrase would enable more environmentally-friendly carbon capture than temperature-swing approaches.

  • Oxidoreductases, transferases, and synthases for petroleum-free platform chemicals: Since a well-engineered enzyme can make several kilograms of product per gram of enzyme used, and Solugen says they make their enzymes for $10/kg, we are sure to see CO2-derived bio-based commodity chemicals that are cost competitive with petrochemicals across the board. Now consider the strong environmental incentives of using biocatalysts, and a major shift in the chemical industry becomes a no-brainer.

  • Carbonic anhydrase for an environmentally-friendly alternative to concrete: Researchers at Worcester Polytechnic Institute made a proof-of-concept "Enzymatic Construction Material" that converts CO2 from the surrounding air into bicarbonate, which precipitates into calcite and makes a material “as strong as commercial cement mortar”. It’s even capable of self-healing as long as the enzymes stay active.

  • Oxidases for mining with bioleaching: Copper ore can be processed by selectively extracting copper ions into water by flowing harsh aqueous acids over the ore to oxidize the sulfide. This was a standard way to process copper ore until the 1950’s when it became cheaper and more profitable to replace the acids with microbes that contain sulfide-oxidizing enzymes. Today this technique (called heap bioleaching) is very common, but I’m sure that a sulfide-oxidizing enzyme could be engineered specifically for heap applications in a purified form and outperform the microbes in speeds, yields, and condition compatibility. In any case, I expect a biomining boom in the wake of EEEE.

  • Epimerases and oxidases for vitamin cascades: China uses an enormous amount of coal to make over 90% of the world’s supply of vitamin C. Most of the coal is used to achieve well over 100 °C for the first step (hydrogenation of D-glucose to sorbitol) as well as the last step (dehydration and esterification of 2-ketogulonic acid to ascorbic acid) in both of China’s two main synthetic routes (the Reichstein process and the two-step fermentation process). I believe that enzymes could be engineered for a more efficient glucose-to-vitamin C cascade that doesn’t require enormous amounts of energy. It would be hard to beat $5/kg, but I’m optimistic that it’s possible, and that it would be worth an attempt to reduce all those emissions. (A separate blog post on this topic with a pending patent coming soon.)

  • Enzymes of all kinds for drug synthesis: 99% of pharmaceutical feedstocks and reagents are derived from petrochemicals, which is a sustainability concern. Luckily most, if not all, major pharma companies have recently formed their own enzyme engineering groups (following Merck’s Januvia success in 2009) to enable more efficient and environmentally-friendly drug production. Soon, as enzyme engineering improves, more and more synthetic steps will adopt biocatalysts beyond the palette of existing natural enzymes. For example, the Arnold Group evolved a new enzyme that can zap an amine onto aliphatic carbons like magic.

  • Halogenases, acylases, and more for the green synthesis of fine chemicals: Aside from vitamins and drugs, the list of fine chemicals that can be made by enzyme cascades keeps growing.

  • Nitrification enzymes for urine recycling: In 2020, a project under the European Space Agency prototyped a more efficient urine recycling system that will generate fresh drinking water on the International Space Station more efficiently. The main improvement was to use bacteria to oxidize urine’s ammonia into more easily-removed nitrates, instead of super-hazardous chromium(IV) trioxide (which the existing ISS urine recycling system uses). I believe this system could be further improved by replacing the bacteria with purified nitrification enzymes, which would be more reliable and more compatible with a wider range of process conditions.

  • Defluorinases and lyases to break down persistent organic pollutants: Enzymes already exist that split C-P and C-F bonds, which are the two main bonds that form the so-called “forever chemicals” class of pollutants. Since those enzymes were naturally evolved for cellular environments, they should be engineered for cheaper application at spill sites and waste treatment facilities to make a broad impact. This could become a practical way to turn those fluorocarbon “forever chemicals” into “hardly knew you hydrocarbons”.


But why are enzymes so expensive to make in a lab?

Solugen says they make their industrial enzymes for $1-10 per kilogram, while Enginzyme says that theirs cost $0.10-1 per kilogram. But those prices are only possible at huge industrial scales and after months (or even years) of optimization around a specific enzyme. The price tag for a new enzyme at the test tube scale, however, is closer to $1,000 per milligram (which equals $1B/kg), whether you make that milligram with your own equipment or you outsource to a company that specializes in making engineered enzymes. And it’ll take you (or that company) at least a month starting from your gene synthesis order. Let’s take the CRISPR/Cas9 nuclease as an example; It’s a niche enzyme that people only need a few micrograms of at time, so the commercial production never had to scale beyond a few milligrams at a time, and the pricing ends up matching the $1,000/mg range of a new lab-made enzyme. That price is the status quo for lab-scale enzymes.

The reason for this high cost is simple: production of heterologous enzymes in vivo has so many failure modes:

  • Inefficient transformation (plasmid insertion)

  • Inappropriate induction timing or IPTG concentration

  • Error in the plasmid’s ribosome binding site or reading frame

  • Ineffective transcription (DNA->mRNA) due to secondary structures or weak promoter

  • Ineffective translation (mRNA->protein) due to frame shift truncation or weak ribosome binding site

  • Poor expression yield due to suboptimal codon usage or mRNA instability

  • Aggregation or incorrect protein folding

  • Toxicity of the over-expressed protein to the E. coli host

  • Protease sensitivity (like for T7 RNA Polymerase)

  • Incorrect expression temperature for protein solubility

  • Problematic protein purification due to concealed his-tag or lack of unique isoelectric point for ion exchange

  • Redox sensitivity (also like T7 RNA Polymerase)

These problems are very common, rarely predictable, and require a high degree of researcher oversight to identify and solve, even in the easiest and most reliable protein production platform (E. coli). This is what I suspect is largely responsible for the $1,000/mg price tag for engineered enzymes.


Cell-free is the way to free up your day!

At the risk of sounding like an infomercial, I need to sing my praises for Cell-Free Protein Synthesis (CFPS). It cuts the two-week E. coli expression protocol down to a few hours, requiring only one step: Just add the DNA sequence that codes for the enzyme of interest to the mix. In a few hours you can either test your enzyme right from within the CFPS mix, or purify it out for testing. CFPS is quick & easy because it separates the complexities of organism cultivation from all the variables of protein expression. It’s a huge convenience to be able to grow your bacteria the same exact way for every enzyme you want to make, rather than customize recipes and protocols every time. With CFPS, you simply cultivate and lyse the bacterial cells in bulk, add some reaction ingredients, and store the mix for later. When you’re ready to make your engineered enzyme, you drop in the DNA sequence and wait a few hours.

Luckily in the last few years the price point of CFPS has become better than the in vivo status quo in most cases. For example, cell-free.com introduced a CFPS protein production and purification service very recently and the pricing and turnaround are extremely favorable relative to all the in vivo status quo services linked above. They’ll ship you 10-100 micrograms of your purified protein in only two weeks including gene synthesis, and you only pay about $200. The services I linked above are more like $2,000 for half a milligram shipped to you in 4-6 weeks. So it should stand to reason, then, that if you acquire CFPS mix online and use it to produce and screen your own engineered enzymes in-house, you’ll save a lot of money and time. Let’s look at some pricing examples on the milligram (easy-to-purify) scale.

Enzyme cost with CFPS:

  • $271/mg if you get the ideal 1.4 mg out of Liberum Biotech’s $379 CFPS mix (1.4 milligrams from one 1 milliliter reaction). Six-part solution, four vials stored at -80 C, two vials stored at -20 C.

  • $481/mg if you get the advertised 480 micrograms out of Bioneer’s $231 CFPS mix (20 micrograms from each 45 microliter reaction). Three-part solution, one vial stored at -80 C, two vials stored at -20 C.

  • $573/mg if you get the advertised 480 micrograms out of Daicel Arbor Biosciences’s $275 CFPS mix (20 micrograms from each 12 microliter reaction). One-part solution stored at -80 C.

  • $904/mg if you get the advertised 250 micrograms out of NEB’s $226 CFPS mix (25 micrograms from each 50 microliter reaction). Four-part solution, two vials stored at -80 C, two vials stored at -20 C.

The Princeton Enzymes plan:

  • $35/mg if you get the expected 4 mg out of Princeton Enzyme’s $140 CFPS mix (4 milligrams from one 20 milliliter reaction). One simple powder stored at room temperature. Available for free beta testing in 2023.

Liberum’s, Bioneer’s, Arbor’s, and NEB’s mixes are great because they yield a lot of protein per milliliter of CFPS reaction, but their pricing makes them more suitable for the microtiter screening step of enzyme design. I plan on using their products in 96-well plates when I need to screen sequence libraries for my next engineered enzyme. But after that step I still need a way to make the most successful variants at a multi-milligram scale for more testing, ideally with all the convenience of CFPS. I can’t find anything that fits my needs online, so I am drawing on these recent insights from literature to develop a solution:

  • Warfel et al. found that maltodextrin can serve as both an energy source and a lyoprotectant for CFPS reactions. This means we can freeze-dry the CFPS mix, which enables shipping and storing at room temperature instead of -80 C.

  • Nagappa et al. found that hydrolysates like peptone and tryptone can replace defined amino acid standards to support CFPS. This brings the price down.

  • Failmezger et al. found that E. coli harvested in the stationary phase yielded even more CFPS activity than the more common mid-log phase when heat stress was applied. This greatly simplifies scale-up.

The Princeton Enzymes plan is to combine and test these findings, and eventually bring a Practical Protein Production Powder to market. This product will enable a more straightforward scale-up of the most promising enzyme candidates to obtain several milligrams at a time, or simply replenish lab stocks of enzymes that you have plasmids on hand for.

 

  1. 16 liters of cultivation media is autoclaved inside the 20-liter bioreactor vessel (sans electronics).

  2. The motor platform is attached to the lid with four bolts, stuff is hooked up, and fed-batch begins.

  3. After growth is finished, the batch is centrifuged in 1-liter bottles to collect and wash the cells.

  4. The cell pellet is sonicated and the resulting lysate is clarified in a second, smaller centrifuge.

  5. Ingredients are added to the clarified lysate before it is freeze dried into a shelf-stable powder.

 

I will use the purpose-built bioreactor pictured above to produce a large number of test batches of PPPP this summer, but each of these will result in far more powder than I could reasonably test on my own. Please contact me if you work in a research lab or at a startup and can assay the yields of several proteins beyond GFP in-house.

Stay tuned for a video next week where I describe this bioreactor in detail. I hope it can be broadly useful for researchers and entrepreneurs that would like to enter the field of biomanufacturing but don’t yet have six figures of capital to spend on a proper (super well-engineered) 20 to 30 liter bioreactor. This Princeton Enzymes construction is a mostly-McMaster bioreactor that only uses off-the-shelf parts, with about 20 to 30 hand-drilled holes. I won’t patent any configuration or construction details, but I do have a potentially patentable idea for a novel way to use it to make CFPS mix much more efficiently, and I’ll apply for a patent if that idea works. Finally, this brings me to…



The CFPS patent landscape

Most people are familiar with breakthroughs in biotech being protected by big famous patents like gene recombination, polymerase chain reaction, and CRISPR. But the type of CFPS that I’ve been describing in this blog post doesn’t seem to have any patent protection…. CFPS itself is a very broad term, because protein synthesis can always occur without the need for intact cells. In principle, the crude lysate of any organism will produce your protein if you add in that protein’s coding DNA along with an RNA polymerase, amino acids, and an energy supply. But the dominant CFPS format today, which uses clarified E. coli lysate and T7 RNA Polymerase to achieve high protein yields, evolved in many labs slowly over time, and the core technology was never patented.

These CFPS-related patents are worth being aware of:

  • US6399323B1 (1990): This patent has expired for a continuous-format CFPS that achieves higher yields by using an ultrafiltration membrane for continuous, passive exchange of small-molecule substrates and byproducts throughout the reaction.

  • US9528137B2 (2014): Northwestern University holds the patent for CFPS reactions that use yeast cell lysate, which has the advantage of simplifying post-translational modifications that are often required in protein therapeutics. Most of the relevant developments were made by the Jewett Lab.

  • US10590456B2 (2016): University of Illinois and Northwestern University own the rights to using ribosomes whose large and small subunits are tethered together for CFPS.

  • I’m still searching, so please comment any others below. That’s where I’ll put additional patents I find.

And some that are pending:

  • JP2013009623A (2011): Tohoku University and Shimadzu Corp are applying to own the rights in Japan to adding chelators to a CFPS mix based on insect cells to control the magnesium concentration.

  • US20200255881A1 (2020): LLNS LLC is applying to own the rights in the US to use ribosomes that have been immobilized on a solid support for CFPS.

  • US20230015505A1 (2020): Northwestern University and Lanzatech Inc are applying to own the rights in a long list of countries to the use of plasmids in CFPS that bear golden gate cloning sites.

Protein Powder Party: Ready for beta testers in August!

Let’s bring lab-scale protein production to all the bright researchers and entrepreneurs that want to benefit the human condition with engineered enzymes! If you’re interested in participating in this journey, please reach out to receive samples from PPPP test batches this summer. I appreciate your help! Also, feel free to leave feedback as a comment below, or a tweet, a LinkedIn comment, or HackerNews comment, and stay tuned for the next few blog post topics:

  • Results of small-scale CFPS tests in the coming weeks

  • A video detailing the design and construction of the bioreactor pictured above

  • What’s an enzyme? (An intro for people outside the field)

  • How to use the Practical Protein Production Powder and follow up with purification


Kenneth Kostenbader