Rationally reduced libraries for combinatorial pathway optimization minimizing experimental effort

Rational flux design in metabolic engineering approaches remains difficult since important pathway information is frequently not available. Therefore empirical methods are applied that randomly change absolute and relative pathway enzyme levels and subsequently screen for variants with improved performance. However, screening is often limited on the analytical side, generating a strong incentive to construct small but smart libraries. Here we introduce RedLibs (Reduced Libraries), an algorithm that allows for the rational design of smart combinatorial libraries for pathway optimization thereby minimizing the use of experimental resources. We demonstrate the utility of RedLibs for the design of ribosome-binding site libraries by in silico and in vivo screening with fluorescent proteins and perform a simple two-step optimization of the product selectivity in the branched multistep pathway for violacein biosynthesis, indicating a general applicability for the algorithm and the proposed heuristics. We expect that RedLibs will substantially simplify the refactoring of synthetic metabolic pathways.


Supplementary Figure 3 | Manipulation of absolute pigment levels of the violacein biosynthesis pathway.
(a) The rationally reduced RBS library exhibits a large phenotypic variety with respect to absolute pigment levels. Within the limited screening effort (372 clones) mutants with higher levels for violacein, deoxyviolacein and crude pigment as compared to the library average (solid line) as well as the parent clone (dashed line) were identified. (b) The five best mutants for violacein, deoxyviolacein and crude pigment production respectively were characterized by quantitative pigment extraction. Data represents the average of four independent replicate cultures with standard deviation.
Supplementary Figure 4 | Operon prediction for clones with high selectivity for deoxyviolacein. The TIR's for clones with the highest deoxyviolacein selectivity (clones %dVio2-5; prediction value according to 1 ) are given. Please note that clone %dVio1 was omitted due to an in-frame stop codon mutation in vioD and can therefore be considered a knock-out mutant for this gene.

Supplementary Figure 5 | Illustration of the library evaluation performed by RedLibs.
The distribution of TIR values assigned to the input data set (NNNNNNNN) of 65'536 sequences predicted 2 for mCherry (see also Results section, Fig. 2) as well as for two reduced sub-libraries with a combinatorial size of 36 (BVRGGSGG and BVRGGRGG) are depicted in the form of histograms (top) and their cumulative distribution functions (bottom). The Kolmogorov-Smirnov distance (dKS) between the library cdf (solid line) and the cdf of a uniform target distribution (dashed line) is highlighted (double arrow) and represents the criterion for library quality: high values of dKS (as for NNNNNNNN or BVRGGSGG) represent unsatisfactory library distributions whereas low dKS values (as for BVRGGRGG) indicate a good resemblance between actual and target distribution. The calculation of cdf's and dKS is described in the Methods section. Figure 6 | Schematic depiction of the cloning procedure for the XFP RBS libraries. In order to introduce the respective degeneracy into the RBS regions of mCherry and sfGFP two overlapping degenerate oligonucleotides are fused together by extension PCR 3 and the resulting double stranded PCR product is inserted into the target vector pMJ1 Lib by conventional restriction digest (SphI & XbaI) and ligation procedures. The process is exemplified for oligonucleotides 1+2 resulting in an N8xN8 library and was carried out accordingly for oligonucleotides 3+4 (N6xN6 library), 5+6 (4x4 library), 7+8 (12x12 library), and 9+10 (24x24 library). Figure 7 | Schematic cloning procedure for the first violacein RBS library. In order to introduce the respective degeneracy into the RBS regions of vioC, vioD and vioE initially three fragments are created by PCR using oligonucleotide pairs 19+20, 21+22, and 23+24 and pMJ3 as template. Subsequently these fragments are joined in a sewing PCR making use of the homologous end regions 3 . The resulting full-length double stranded PCR product is then inserted into the target vector pMJ3 Lib by conventional restriction digest (BglII & KpnI) and ligation procedures. A similar procedure was carried out in order to produce the second violacein RBS library but using oligonucleotide pairs 25+26, 21+22, and 27+24 to generate the three PCR fragments in the first step.

Supplementary Table 1 | Sequences specified for prediction by the RBS Library Calculator.
To retrieve input data sets for the respective genes the degenerate RBS region containing eight consecutive N's (underlined) was specified in the RBS library calculator 2 . Additionally 20 bp upstream of the RBS (pre-sequence) as well as the first 50 bp of the coding sequence (start codon highlighted in bold) were used. The 16S-rRNA sequence was chosen for E. coli strain DH10B.

Chemicals and Reagents
If not stated otherwise all chemicals and reagents were obtained from Sigma Aldrich (Buchs, Switzerland). Restriction enzymes were obtained from New England Biolabs (Ipswich, MA, USA).