Written by Hannah Edstrom
In our last post on this paper, we introduced the topic and went over the logic of their approach. Today we will be looking at what goes on in lab to begin tackling this challenge.
This paper showcases an attempt to make E. coli better at producing isobutyl acetate (IBA), a compound used in chemistry, pharmaceuticals, and cosmetics. It’s also naturally produced in wine ferments and contributes a nice fruity fragrance.
What did they do in lab to answer their questions?
Let’s start by sketching out the approach here:
- Design a pathway
- Screen for optimal acetyl-CoA pathway
- Test affect of pathway on growth
- Test the effect of feed mixes on IBA production
- Test pathway with isotope tracing
- Demonstrate technique to increase glucose consumption
The strain the authors started with was JCL260, a strain of E. coli previously engineered to produce IBA. Now, this strain produces acetyl-CoA via the PDHC pathway (PDHC is the enzyme that converts pyruvate, from glucose, into acetyl-CoA.) So they’re starting with a strain that makes the desired product, but at a lower carbon yield than they think is possible.
Table 1 (on page 3) shows all strains engineered in this paper. The “host” is the starting strain; mostly, this is JCL260. But later on, the authors make another strain by knocking out the PDHC enzyme from JCL260. They name this strain, rather unromantically, AL2045. These are the two hosts, then, they work with.
The table continues with all the plasmids engineered into these two hosts. These plasmids carry various renditions of the acetate → acetyl-CoA pathway (Fig. 1B).
This brings us to Step #2 of their methodology: screening for a pathway. Fig. 1B presents three possible pathways to get acetyl-CoA out of acetate. They’ve been selected for redox balance and carbon efficiency (none of these reactions have by-products like carbon dioxide). In Fig. 2A, OD600 is used as a measure of cell growth. The left graph in Fig. 2A shows growth of four strains on glucose; these four strains are each JCL260 with three separate acetate → acetyl-CoA pathways, and a negative control (NC). You might notice that NC is Strain 4, the JCL260 strain with a GFP plasmid built into it (GFP is a common marker in biology laboratories.)
The result of this study, of growing all four strains on glucose, demonstrated that no one edit impacted growth when grown on glucose. That’s good – glucose is the standard feed, so we don’t want an edit that will slow down E. coli‘s growth. The second test, on the right, shows growth (again, measured by OD600) when the strains were grown on acetate as a carbon source. Here, the ackA-pta pathway from acetate to acetyl-CoA is clearly a winner, as it significantly increases growth on acetate. Thus, this was chosen as the favorite pathway for its efficiency.
The authors wanted more information on this winning pathway, however. So they tested four separate strains on three different substrate mixtures, depicted in Fig. 2B and 2C.
First, the authors have two hosts: JCL260, and AL2045. These strains are identical, except that AL2045 does not have a PDHC enzyme. Thus, it is impossible for this strain to produce acetyl-CoA from pyruvate.
Second, of these two hosts, they built the same plasmid into both of them. That plasmid is, of course, the one containing the ackA-pta pathway from acetate to acetyl-CoA.
So what is the next test? The authors don’t want to grow their strains on just glucose or just acetate. They want a strain that is happy growing on a mixture of acetate and glucose. Each graph, then, represents a different strain, and the three colors denote three substrates. Red represents the strain grown on both acetate and glucose. In which strain do the red samples depict the most growth?
Only in AL2045+ackA/pta does the strain grow best in glucose + acetate. That’s a nice confirmation of the health of this strain and efficiency for our intentions.
You may also notice that the empty AL2045 strain did not grow much at all – except it grew a tiny bit under glucose, and even less under glucose + acetate. We know that both glucose and acetyl-CoA are vital to cell growth. How is that possible? How can these strains grow without acetyl-CoA available? Although AL2045 has no means of producing acetyl-CoA out of glucose, there is another pathway in E. coli that converts glucose to acetate. From there, the E. coli can convert acetate to acetyl-CoA. These pathways are both very slow and inefficient, however, hence the slow growth where glucose is available.
Moving on to Step 4. We are further optimizing the feed mixture. First we determined how glucose alone and how acetate alone affected growth, then we tested edits for how well they grew on a glucose + acetate mixture. Now, let’s optimize ratio of glucose and acetate in the fed substrate.
Looking at Fig. 1, we can see that there is a 1:1 molar ratio of glucose:acetate needed. If PDHC is knocked out, and all acetyl-CoA is coming from acetate via the ackA/pta pathway, then:
- 1 mole of glucose is converted to 1 mole of isobutanol
- 1 mole of acetate is converted to 1 mole of acetyl-CoA
- 1 mole isobutanol and 1 mole of acetyl-CoA produce 1 mole IBA
If you increase the concentration of acetate to more than 10g/L, you start to harm the E. coli and growth is affected. Thus, all mixtures have a maximum of 10g/L acetate in them. To get a 1:1 molar ratio with 10g acetate, you only need 30g glucose. However, the authors also tested a surplus of glucose for mixtures labeled 30Glu+10Ace, 40Glu+10Ace, and 50Glu+10Ace as depicted in Fig. 3.
Looking briefly over the parts of Fig. 3, we can see that the 50Glu+10Ace mixture produces the most IBA (Fig. 3A). Culture grown on pure glucose consumed the most glucose (Fig. 3B) and culture grown on pure acetate consumed the most acetate (Fig. 3D), obviously – but Fig. 3A shows that co-utilization of both glucose and acetate, at concentrations of 50g/L glucose and 10g/L acetate, produce the most IBA.
Growth was not affected by any particular substrate (Fig. 3E), and only in the batch fed pure glucose was there leftover isobutanol (Fig. 3F). In Fig. 3C, we get carbon yield, calculated by:
TCMY = CIBA/(CGlu + CAce)
So not only did the 50Glu+10Ace mixture produce the most IBA, it also produced the highest carbon yield.
Let’s skip down to Fig. 5 and talk more about this molar ratio thing. We know that we can’t increase acetate concentration without harming E. coli. But we also know that 50Glu+10Ace is the most productive substrate for IBA production, and there isn’t enough acetate to use up all the extra glucose. So that identifies another area to improve carbon yield for IBA production.
To solve this, the authors ran another experiment in which they monitored acetate concentration and added acetate every 24 hrs to the batch. We can see the addition of acetate every 24 hrs in Fig. 5D, and also note that after about 100 hrs, the acetate is no longer being consumed. That indicates we’ve used up all the glucose and now have excess acetate. Feeding this way resulted in more glucose consumption (Fig. 5B), less leftover isobutanol (Fig. 5F), and a final IBA titer of 19.7g/L (Fig. 5A). Once again, they ensured growth was left unharmed by charting OD in Fig. 5E). Finally, we see a carbon yield consistently hovering around 60% – throughout the course of the run.
Thus, adding acetate throughout the run is another technique to improve carbon yield and IBA production.
We’ll go over Step 5 in the next and final post for this paper.
Tashiro, Y., Desai, S. H., & Atsumi, S. (2015). Two-dimensional isobutyl acetate production pathways to improve carbon yield. Nature Communications, 6(May), 1–9. https://doi.org/10.1038/ncomms8488