Engineering microbial biofuel tolerance and export using efflux pumps

Abstract

Many compounds being considered as candidates for advanced biofuels are toxic to microorganisms. This introduces an undesirable trade-off when engineering metabolic pathways for biofuel production because the engineered microbes must balance production against survival. Cellular export systems, such as efflux pumps, provide a direct mechanism for reducing biofuel toxicity. To identify novel biofuel pumps, we used bioinformatics to generate a list of all efflux pumps from sequenced bacterial genomes and prioritized a subset of targets for cloning. The resulting library of 43 pumps was heterologously expressed in Escherichia coli, where we tested it against seven representative biofuels. By using a competitive growth assay, we efficiently distinguished pumps that improved survival. For two of the fuels (n-butanol and isopentanol), none of the pumps improved tolerance. For all other fuels, we identified pumps that restored growth in the presence of biofuel. We then tested a beneficial pump directly in a production strain and demonstrated that it improved biofuel yields. Our findings introduce new tools for engineering production strains and utilize the increasingly large database of sequenced genomes.

Figure 1

Competition assay efficiently identifies efflux pumps that provide biofuel tolerance. (A) Plasmids containing the operons for individual pumps were transformed into cells. These strains were grown independently and then pooled in equal proportion. The pooled culture was grown both with and without biofuel for 96 h, with dilution into fresh medium every 10–14 h. At selected time points, cultures were saved, plasmids isolated, and the relative levels of each plasmid were quantified. If certain plasmids provide a growth advantage, they become overrepresented in the culture. (B) Simulations of competitive exclusion. Grown separately, the growth curve of a strain is dependent only on its growth rate. When grown together, as in the competition assay, strains with higher growth rates dominate the population.

Figure 2

When grown with biofuel, strains with beneficial pumps dominate the culture. (A) The pooled culture shows performance of all pumps to be similar in the absence of an inhibitor. Each row represents the quantity of a different pump-containing plasmid; columns show time points in the competition assay. For each pump, the host and GenBank accession number (or protein name, if annotated) of the inner membrane protein are given (Supplementary Table S1). (B) When the pooled culture was grown with 2% geranyl acetate, some pumps improved survival. Positive and negative hybridization controls label all pump plasmids and a pump-free plasmid, respectively (Materials and methods). All data are shown in log scale with arbitrary units (a.u.).

Figure 3

Efflux pumps can improve tolerance and production yields for selected biofuels. (A) Plasmid levels for each of the efflux pumps in the library after 72 h of competition. The pool culture was grown without an inhibitor. For all others, biofuel was added at an inhibitory level (Materials and methods). Data from three biological replicates and their average are shown. (B) Inhibition experiments with the top-performing pump (shown with an arrow in (A)) showed that for n-butanol and isopentanol the pump provides no growth advantage over the negative control. Taken together, (A) and (B) demonstrate that none of the efflux pumps we tested improved tolerance to n-butanol or isopentanol. (C) Competition results for biofuels where efflux pumps provided a growth advantage. (D) Top-performing pumps (shown with an arrow in (C)) showed improvements over the negative control strain. Measurements shown in (B) and (D) were taken in triplicate and averaged; error bars show standard deviation. Note that all efflux pumps that survived the α-pinene competition showed good performance when grown individually with α-pinene (Supplementary Figures S3, S4). (E) Limonene yield in a production strain expressing an efflux pump compared with an identical strain without the pump. Error bars show standard deviation between biological replicates.