Fig. 6: Combinatorial CRISPRa library applied to an HMO biosynthesis pathway identifies high-producing strains and pathway bottlenecks.

a The LNT pathway consists of lacY, lgtA, and wbgO overexpression controlled by the J3, J5, and J6 promoters, respectively. The substrates UDP-GlcNAc and UDP-Gal come from endogenous metabolism. b HPLC analysis of supernatant from singlicate cultures indicates LNT production levels across the scRNA library. The highest producing strain (#17, black arrow) was used in the galactosyltransferase comparison in e. The x-axis heatmap is color coded to indicate the encoded promoter expression for each strain, as described on the bottom right. The no-pathway culture carries an empty vector. For comparison, LNT II levels are shown in Supplementary Fig. 12. c Dependence of LNT (top) and LNT II (bottom) production on lgtA and wbgO activation highlights sensitivity to wbgO activation and accumulation of LNT II. Only medium-lacY strains are shown here, due to their rich variance across the subset (box plot in b: center line, median; box limits, upper and lower quartiles; whiskers, range). The arrow again indicates strain #17. d Computational strain recommendations from the Automated Recommendation Tool (ART) and their predicted LNT titers. Strains are defined by their scRNA spacer lengths (measured in nucleotides), which determine degree of CRISPR activation (lower right). The 20 strains with highest predicted titer are highlighted in color on each subgraph, with the rest shown in gray. The same 32 strains are shown on each subgraph. Spacer lengths defined as high, medium, and low expression in experimental data are indicated as vertical lines. Points in d represent each recommended strain’s specific truncation for that scRNA, while error bars indicate the 95% credible interval of the predictive posterior distribution. See Supplementary Fig. 15 for how recommendations are combined within each strain. e A more active enzyme from C. violaceum⁷³ (right) resolves accumulation of LNT II (left), at various initial feedstock concentrations. The horizontal line indicates LNT titer from WbgO and 2 g/L initial lactose; CvGalT achieves similar titer using only 0.05–0.2 g/L initial lactose. Bar values in e represent the average ± standard deviation calculated from n = 3 biologically independent samples. Source data are provided as a Source Data file.