Engineering, Synthesis

Chemical Feedstock Production by Fermentation

“Metabolic engineering of Escherichia coli for direct production of 1,4-butanediol” Yim, H.; Haselbeck, R.; Niu, W.; Pujol-Baxley, C.; Burgard, A.; Boldt, J.; Khandurina, J.; Trawick, J. D.; Osterhout, R. E.; Stephen, R.; Estadilla, J.; Teisan, S.; Schreyer, H.B.; Andrae, S.; Yang, T. H.; Lee, S. Y.; Burk, M. J.; Van Dien, S.  Nature Chem. Bio. 2011. 7, 445-452. DOI: 10.1038/nchembio.580

The production of chemicals from biologically-derived feedstocks is a major goal of green chemistry research, but despite a lot of work that’s been done, it’s going to be hard to make the switch from petroleum-derived chemicals to bio-based ones.  This is especially true for high-volume commodity chemicals – many of these chemicals have been produced from petroleum for a hundred years, the processes have been optimized to work efficiently on enormous scale, and they are really, really cheap.  So the bar is set pretty high, and most papers from academic labs on microbial or enzymatic chemical production are too low-yielding to ever be commercialized (although to be fair, the same could be said for most synthetic chemistry papers).  That’s why I was a drawn to this paper published by Genomatica, a company based in San Diego, on the production of 1,4-butanediol by an engineered strain of E. coli – first they got the bug to produce 1,4-butanediol, then they engineered it to produce lots of the stuff.  Currently one million tons of 1,4-butanediol (BDO) are produced each year, virtually all of it derived from petroleum-based feedstock chemicals.

Apparently 40% of this is used in the production of Spandex, and the rest of it is used to make other polymers and THF.  If Genomatica’s BDO production works according to their plan, all those tons of spandex could be bio-based!

The future of spandex?

BDO isn’t produced naturally by any known organisms, so the first order of business was to design a biosynthetic pathway for BDO production.  Genomatica has their own software for generating and ranking hypothetical biochemical pathways in E. coli, which came up with over 10,000 possibilities for converting common metabolites into BDO.  90% of these were eliminated because of unfavorable thermodynamics or low theoretical yield, and the remaining were ranked based on things like the number of biochemical steps, the number of steps not native to E. coli, and the number of steps from central metabolism.  It’s worth pointing out that they focused on metabolites and enzymes from central metabolism – which is the collection of biochemical reactions that are absolutely essential for survival of all organisms: breakdown of sugars and fats, production of amino acids and nucleic acids, storage of energy, etc.   The majority of an organism’s carbon flux is going through these intermediates so there’s a greater stream of metabolites to divert, and higher yields are theoretically possible.   Additionally, these enzymes have been under constant evolutionary pressure over billions of years to operate very efficiently so they are less likely to become bottlenecks in the overall process.

Their most promising artificial biosynthesis starts with two intermediates from the citric acid cycle, succinate and alpha-ketoglutarate, and converges at succinyl semialdehyde, which is reduced with three hydride equivalents to BDO.  Here’s the first half of their pathway:

Enzymes that catalyze steps A and D are native to E. coli, while enzymes catalyzing steps B and C were known from other microorganisms.  So, their first experiments were aimed at transforming genes encoding those two enzymes into E. coli, and then looking for production of 4-hydroxybutyrate (4HB).  Transformation of genes encoding either alpha-ketoacid decarboxylase (step B) or succinate semialdehyde dehydrogenase (step C) into E. coli conferred the ability to produce 4HB, albeit at low levels.  The authors experimented with enzymes from different organisms, plasmids whose genes were expressed in E. coli at different levels, and duplication of the native E. coli  enzymes for steps A and D with similar enzymes from other organisms. After all of this, they ended up with an organism that could produce ~1 g/L of the intermediate 4HB.  Pretty modest quantities, but a step in the right direction.

Here’s the second half of their artificial biosynthesis, which reduces 4HB to BDO, their desired product:

None of the three final steps were known in E. coli, so genes encoding enzymes for those three reaction had to be identified in other organisms and then transformed into E. coli.  Step E is catalyzed by a CoA transferase in P. gingivalis.  Enzymes performing steps F and G were not known, but the overall transformation is known to occur in C. acetobutylicum.  So, they tested a bunch of enzymes from that organism with alcohol and aldehyde dehydrogenase activity (which, in this context, are performing the reverse reaction that their name implies – confusing!), and found an enzyme capable of catalyzing both steps F and G when expressed in E. coli.  Then they took all of the enzymes from these experiments and expressed them in a strain of E. coli, which conferred the ability to produce BDO from glucose.  The production of BDO at this point was only 100 mg/L, and many other compounds were being produced too, so the organism was definitely not ready for prime time.

The next phase of the project was to engineer the organism to channel all of its carbon compounds and reducing equivalents through the artificial BDO pathway instead of its native metabolic pathways.  Like the design of the BDO biosynthesis itself, these metabolic engineering efforts were guided by software that suggested the deletion of certain enzymes in primary metabolism, and identified other enzymes as being critical for production (but not present in the BDO biosynthesis itself).

This software recommended that they delete three NADH-dependent dehydrogenase enzymes (alcohol, lactate and malate) to funnel NADH into the BDO biosynthesis, which uses three equivalents of NADH.  All of this extra NADH sloshing around in the organism had a deleterious effect though; one of the crucial enzymes identified by their software, pyruvate dehydrogenase, is inhibited by high levels of NADH, and didn’t function in the knockout organism.  This enzyme is so crucial to the organism (it connects glycolysis to the citric acid cycle) that it was barely able to grow, let alone produce any BDO.  But, they were able to swap in a version of pyruvate dehydrogenase containing a mutation that reduces NADH sensitivity, which restored growth and gave them a four-fold increase in BDO production.

They ended up doing a lot more engineering – more deletions, more mutations to reduce NADH inhibition, swapping in enzymes from other organisms, deleting transcriptional inhibitors of their crucial enzymes, and they ended up with an organism that produced an order of magnitude more BDO than their original strain.  Improvements made to the fermentation conditions gave them an additional order of magnitude increase in product titer, bringing them up to 18 g/L, or nearly 2% by weight.  Compared to the percentages that are obtained from ethanol fermentations (>13%) there is probably room for improvement, but these results are sufficiently promising that Genomatica has started large-scale production.  Overall, I thought this was a very cool (and very dense) piece of work, one that highlights all the work that is necessary to take a project from the proof-of-concept stage to a point that is commercially feasible.


5 thoughts on “Chemical Feedstock Production by Fermentation

  1. You mention the potential for scalability at 18 g/L. How universal is this metric? How does it compare to other bio based process?

  2. In the paper it said that they’re looking for 3 to 6-fold increase in productivity as they scale up, which would put them at 50-100 g/L, which is apparently what they need for their process to be cost competitive. I think it depends on what you’re making though – the higher the titer, the cheaper the production costs, but if you’re making a high value product, you can get away with lower production levels. Keasling’s original reports on biobased artemisinin production were getting 100-200 mg/L, and they probably will never get to 20 g/L. But they probably don’t need to either.

  3. Great post Courtney. Do you have a sense of what the efficiency would need to be in order for this process to be competitive with existing petrochemical processes (ie what would the g/L need to reach)? Also, do the authors factor in cost of feedstock- food vs fuel competition, etc. I’m assuming that at this point in the process development, they’re more interested in getting an organism that simply does what they want it to. But do they address the challenges of switching a feedstock to something like cellulose as opposed to starch. This must be a consideration if they’re planning on scaling this to production levels.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s