by Christoph
«Oh, sir,… just one more thing»
– Peter Falk as Columbo
While Roberto looked deep into the fascinating past of How E. coli's Rose to Prominence, plus An Addendum, in his recent posts, I will take a look into the future here in Just One More Addendum, and a wider look around in an upcoming post Again, One More Addendum. But that was it for the time being with this Ecolimania here in STC... I admit that I feel a bit like the actor Peter Falk in the TV crime series Columbo, in which the detective always began the crucial question to the main suspect with "Oh, sir,... just one more thing" shortly before the end of an episode. Today’s "one more thing" is the question: Do genome‑reduced variants still belong to the species E. coli ?
A note in advance: Two of the images in this post, the Figures 1.1+1.4, are mainly illustrative and a look at the overall picture is sufficient. Clicking on the images may make them easier to read when enlarged, but the huge number of data points in each cannot really be grasped.
Do genome‑reduced variants still belong to the species E. coli ?
E. coli lab strains like K-12 or B are inherently able to adapt quickly to various growth conditions and can be easily (genetically) manipulated to produce fair amounts of heterologous proteins (or nucleic acids, or cofactors, or...) as they grow. However, biotechnologists are not much interested in such strains as "production strains" – they prefer the term “chassis” – for any type of plasmid-based expression system for products. Firstly, lab strains are not genetically stable enough for long-term maintenance and, secondly, they consume most of the supplied nutrients for their own reproduction rather than for the synthesis of the target product. Genetic manipulation to remove all gene segments from the genome whose gene products are dispensable for the reproduction of the cells under defined growth conditions could theoretically lead to "production strains" as needed by biotechnologists. If their growth rates were comparable to those of the parent strain, such genome-reduced variants would have the additional advantage of requiring less energy-consuming precursors, dNTPs, for the replication of their genomes (chromosomes and plasmids).
Two decades ago, Kolisnychenko et al. (2002) deleted the "K islands" from the chromosome of the (plasmid-free) E. coli K‑12 strain MG1655, that is, those sequence segments that are unique to K‑12 but lacking in E. coli O157:H7 and/or E. coli CFT073, both of which belong to different E. coli phylotypes (see part 2). The 4.263 Mbp genome of the resulting strain MDS12 is 8.11% smaller than the MG1655 parent genome (4.640 Mbp). The 12 deletions of a total of 376 kb removed most of the potentially "selfish" DNA like cryptic prophages, phage remnants, insertion sequences (IS elements, transposons), along with 409 ORFs of genes with hypothetical/unknown functions, and 2 stable RNAs (redundant tRNAs) (Figure 1.1). The authors assume that the removal of (most of) the IS elements and transposons has contributed to an improved stability of MDS12, since these are known to cause a background of intra-genomic rearrangements, often accompanied by unwanted mutations (see next paragraph). The growth of MDS12 in rich medium (LB) and in standard and supplemented minimal medium (M9/glucose, M9/casamino acid (CAS)/glucose) was virtually indistinguishable from that of the parent strain MG1655 (Figure 1.2), but showed a ~twofold growth reduction for 15 of 379 assayed growth conditions (3/190 carbon sources, 1/59 phosphate source, 11/95 nitrogen sources).
In a follow-up study, the researchers achieved in strain MDS43 (3.931 Mbp) a further reduction of the MG1655 genome (4.640 Mbp) by 15.27%. The growth characteristics of MDS43 and that of the progenitor strains MDS41 (3.977 Mbp) and MDS42 (3.976 Mbp) were comparable to that of MDS12. Using a clever assay, they were able to confirm their assumption that genome rearrangements and mutations caused by the transposition of IS elements occur less frequently in their IS element-depleted "MDS strains": metabolism of salicin by E. coli requires activation of the bgl operon, which occurs primarily by IS insertion into the promoter region. When mutants of MDS41 and MG1655 were selected with salicin as the sole carbon source, the activation rate for MDS41 was less than 8% of that for MG1655. These strains would thus be stable enough for long-term maintenance as desired by biotechnologists.
Mizoguchi et al. (2008) chose a different E. coli K‑12 strain, W3110 (4.646 Mbp) as target for the stepwise deletion of a total of 1.03 Mbp genomic sequences. More importantly, these researchers used a completely different strategy than Kolisnychenko et al. (2002) for the choice of deletion targets: First, they compared the E. coli MG1655 genome (4.640 Mbp) with the reduced genome of the aphid endosymbiont Buchnera aphidicola APS (0.641 Mbp, acc. no. NC_002528.1) and tagged all E. coli‑unique genes for deletion (Buchnera is of known enterobacterial origin and was featured here and here in STC). Then they "un‑tagged" any of those genes that were reported as essential in the PEC database. Thirdly, they surveyed the annotations of the remaining tagged candidates in the ERGO databases to judge their necessity for good growth in M9 minimal medium. Lastly, they chose genome segments with more than 10 continuous "unnecessary" genes for deletion. The 50+ targeted deletions were introduced individually into their recombination-proficient "starter strain" W3110red, checked for good growth on M9 minimal medium, and finally assembled together in E. coli strain MGF-01 (3.616 Mbp). The genome of this strain is reduced by 22.2% as compared to W3110 (4.646 Mbp).
As a brilliant confirmation – yes, "brilliant" is no hyperbole here – of their chosen genome‑reduction strategy, the researchers were then able to determine in growth tests that E. coli MGF-01 not only grows just as well on M9 minimal medium during the exponential growth phase as E. coli W3110red, but ultimately even achieves significantly higher cell numbers while having an almost identical cell size (Figure 1.3). Mizoguchi et al. (2008) consider it plausible that together the deletions made in the genome led to a “rewiring” of the central glucose metabolism allowing for more efficient glucose utilization, avoidance of a slide into overflow metabolism, and to improved glucose uptake (an intriguing finding for systems biologists). For biotechnologists, MGF-01 would make an attractive "chassis" despite the fact that its long-term stability had not been thoroughly explored at the time of its publication.
Two recent studies describe E. coli strains genomes had been reduced by ~24% or even ~39%, respectively, in each case relative to the MG1655 "starting" strain. Both strains E. coli eMS57 (~3.50 Mbp) by Choe et al. (2019) and E. coli Δ41c (~2.60 Mbp) by Kotaka et al. (2023) were obtained from heavily genome-reduced "intermediate strains" exhibiting poor growth that were selected for better growth by one or two rounds of adaptive laboratory evolution (ALE). ALE is basically an application of the LTEE strategy designed by Richard Lenski: Let cells adapt to constant growth conditions over many generations, and check throughout how they have managed to do this, which mutations they have accumulated along the way (STC had a post on the LTEE experiment here; Rich Lenski recently announced that generation 80,000 has been exceeded, and the strains are still adapting to their growth medium albeit considerably slower than in the beginning).
Choe et al. (2019) and Kotaka et al. (2023) sequenced their respective strains and both found a whole series of very interesting mutations. To present these here in full or even just by way of example would completely exceed my word count (and your attention span). So just a rough summary: analysis of the mutations found, including transcriptomics, suggests that they have all led to extensive transcriptional "rewiring" in both strains, and that there were no significant mutations in metabolic enzymes.
With a cursory comparison of the genomic positions of the deletions in E. coli MDS12 (Figure 1.1) and in E. coli Δ33a and Δ41c (Figure 1.4), you will notice that they are distributed fairly equally across both chromosome halves ("replichores") between origin (oriC) and terminus (dif). This implies that replication of both replichores starting at oriC can proceed in an almost balanced manner up to the terminus region. And indeed, no problems with chromosome replication have been observed for these strains so far. This is certainly not insignificant for all attempts to further reduce the genome of E. coli (says the "replicationist" me).
You will have noticed that I have casually referred to all the strains with reduced genomes mentioned above as "E. coli". They still "smell" like E. coli (their cultures do too, by the way, I had the questionable pleasure of smelling a MGF-01 culture). The smallest genome whose carrier could still be issued a passport as E. coli would probably have a size of approximately 1.5 Mbp (~1.500 protein coding genes), and a visa would be limited to a journey in standard minimal medium (M9/glucose) and rich media like LB or CSL. This assumption is based on the finding that the so far smallest known genomes of free-living bacteria fall within this size range: P. ubique HTCC1062 (1.309 Mbp) and P. marinus MED4 (1.658 Mbp). Both species Pelagibacter (Alphaproteobacteria) and Prochlorococcus (Cyanobacteriota) thrive as numerous strains, all in huge numbers of planktonic bacteria, in the planet's oceans (an estimated 2.9×1027 Prochlorococcus). However, both bacteria can so far not be grown to high cell‑density in liquid culture in the lab, they achieve ~1 doubling/day, and to form colonies (high cell-density!) on plates they need "helper bacteria" providing metabolites and enzymes, like catalase, for example, to remove excess ROS. This is not in the interest of biotechnologists, who prefer high cell-densities in their fermenters in order to achieve an economically attractive product/biomass/reactor volume ratio.
There is still a lot of tinkering to be done to reduce a genome of 2.60 Mbp as in E. coli Δ41c to the size of a Prochlorococcus (~1.6 Mbp) or Pelagibacter (1.3 Mbp) genome. If researchers pursue the stepwise deletion of one megabasepair (1 Mbp!) from the genome and continue to restore growth capacity by adaptive laboratory evolution (ALE) of impaired intermediate strains, fascinating insights into the intricacies of gene regulation In E. coli will be gained along the way. That's for sure.
Do you want to comment on this post? We would be happy about it! Please comment on Mastodon, Bluesky, or on 𝕏 (formerly Twitter).
Comments