Coupling toxin proteolysis with cell survival.

A) ATP-dependent proteases can initiate proteolysis by recognizing substrate degrons. B) Sequence-based recognition of substrates is poorly understood in bacteria. E. coli’s five ATP-dependent proteases are depicted. C) Overview of the DEtox screening approach, which couples degron-directed toxin proteolysis to degron enrichment in liquid culture.

Proteolysis rescues cells from toxicity of VapC expression.

A) Expression of untagged VapC with 1% arabinose arrests cell growth in wild-type E. coli and in a ΔclpA ΔclpX strain. B) Appendage of the full-length ssrA tag to the C-terminus of VapC restores cell growth in wild-type but not ΔclpA ΔclpX cells. A minimal ssrA tag (YALAA) is sufficient to rescue toxicity in wild-type but not ΔclpA ΔclpX cells. D) No rescue is observed when the terminal Ala is substituted with Asp (YALAD).

Degron strength correlates with growth rate.

A) Ec°ClpXP (0.25 μM) degrades GFPYALAA with a KM of 8.3 ± 0.9 μM, GFPYALAS with a KM of 65 ± 18 μM, and GFPYALAD with a KM of 820 ± 670 μM. Data were fit to a Michaelis-Menten equation. B) Expression of VapCYALAA, VapCYALAS, VapCYALAD in E. coli results in growth rates that correlate with each tag’s respective KM. C) A VapC library randomizing the last position of the minimal ssrA tag (VapCYALAX) was transformed into wild-type E. coli and grown under inducing conditions in liquid culture. Sanger sequencing revealed random codon composition at 0 h but strong enrichment of Ala-encoding codons (GCT, GCG) after 6 h of growth.

Expression of VapC5X library on plates and in liquid culture.

A) A VapC library bearing five NNK randomized C-terminal codons (VapC5X) was transformed into wild-type E. coli and plated on 0% (-) and 1% (+) arabinose. Approximately 1% of library transformants support growth under inducing conditions, producing colonies of varying size. B) Cell density of the transformed library was monitored over time in the absence and presence of inducer. Samples of induced culture were harvested at the indicated time points (red arrows) to assess library composition.

DEtox screening in wild-type E. coli.

A) A heat map illustrates the fold-enrichment of unique sequences compared to their abundance at 0 h. The top ∼1% of sequences were enriched at least 100-fold at 6 h. B) The apparent growth rate correlates with fold-enrichment on a semi-log plot. C) Heat map of amino acid log fold-enrichment at each tag position over time. (Raw abundance values are reported in Fig. S3C). A GFP model substrate bearing the mostly strongly enriched tag, GFPFKLVA, is degraded by EcoClpXP (0.25 μM) with a similar KM to GFPYALAA (13.6 ± 1.7 μM and 8.3 ± 0.9 μM, respectively). E) Heat maps plot sequence fold-enrichment against net charge, hydrophobicity, or similarity to the ssrA tag (as average per-residue BLOSUM62 score against YALAA). Boxes indicate the most enriched tags, which are comparatively hydrophobic and more similar to the ssrA sequence than the bulk population.

Terminal Ala-Ala motifs are abundant among enriched 5-mer tags.

A) Sequence logos of the full library, the top 1000, and the top 100 highest fold-enrichment tags. B) Heat map illustrating the average fold enrichment of 5-mer, 4-mer or 3-mer tags ending in Ala-Ala, bearing the indicated amino acid at the indicated position.

Fewer strongly enriched degrons appear in strains lacking components of ClpXP.

A) Fold-enrichment of tags observed across replicates in the indicated E. coli strain were cross-compared to identify consistently enriched tags. Tags enriched after 6 hours but not observed in the pre-induction samples were set to an initial occurrence of 1. Tags enriched <5-fold in at least one replicate are colored gray, and make up the majority of observed sequences in each strain. B) Charge, hydrophobicity and ssrA similarity of consistently enriched tags are plotted against fold-enrichment.