Discovery of antibacterial cyclic peptides that inhibit the ClpXP protease

Abstract

A method to rapidly screen libraries of cyclic peptides in vivo for molecules with biological activity has been developed and used to isolate cyclic peptide inhibitors of the ClpXP protease. Fluorescence activated cell sorting was used in conjunction with a fluorescent reporter to isolate cyclic peptides that inhibit the proteolysis of tmRNA-tagged proteins in Escherichia coli. Inhibitors shared little sequence similarity and interfered with unexpected steps in the ClpXP mechanism in vitro. One cyclic peptide, IXP1, inhibited the degradation of unrelated ClpXP substrates and has bactericidal activity when added to growing cultures of Caulobacter crescentus, a model organism that requires ClpXP activity for viability. The screen used here could be adapted to identify cyclic peptide inhibitors of any enzyme that can be expressed in E. coli in conjunction with a fluorescent reporter.

Figure 1.

Selection for cyclic peptide inhibitors of ClpXP. (A) Schematic diagram of SICLOPPS library construction. Three fixed and five randomized codons were cloned in the coding region between the intein I_C and I_N domains. When this gene is expressed, the I_C and I_N domains promote circular ligation of the intervening peptide, resulting in 8-mer cyclic peptides with five random amino acids. (B) Schematic diagram of the GFP-tag reporter. The tmRNA peptide tag sequence was encoded at the 3′-end of the gfp gene under control of an IPTG-inducible promoter. All GFP produced from this gene will have the tmRNA peptide tag at the C terminus. (C) Result of expression of the GFP-tag reporter and inhibitory cyclic peptides in E. coli. Production of GFP-tag was induced in wild-type E. coli (wt), a strain lacking the clpX gene (ΔclpX), and a strain that was also producing the IXP1 cyclic peptide; the cells were imaged by immunofluorescence (fluor.) to see fluorescent cells and differential interference contrast microscopy (DIC) to see all cells. In wild type, SspB binds to the tmRNA peptide at the C terminus of GFP-tag and tethers the protein to the ClpXP protease, resulting in rapid degradation and no fluorescent cells. In the ΔclpX strain, the absence of active ClpXP protease results in stabilization of GFP-tag and highly fluorescent cells. In wild-type cells producing IXP1, the cyclic peptide inhibits degradation of GFP-tag, resulting in fluorescent cells. DIC images show that the ΔclpX cells and wild-type cells producing IXP1 are also slightly filamentous.

Figure 2.

Cyclic IXP1 inhibits ClpXP in vitro. GFP-tag was incubated with ClpXP, and proteolysis was monitored using a continuous fluorometric assay. Representative assays without inhibitor and with IXP1 are shown. The assays were repeated using different concentrations of substrate to determine the apparent kinetic parameters. Eadie-Hofstee plots (inset) for proteolysis with no inhibitor (solid line), 50 μM IXP1 (long dashes), and 100 μM IXP1 (short dashes), which are consistent with an uncompetitive inhibition model.

Figure 3.

Interaction of IXP1 with ClpX and ClpP in vitro. (A) IXP1 was incubated with ClpXP for 60 min, and samples before (0 min) and after (60 min) incubation were analyzed by reverse-phase HPLC. Plots of the absorbance at 280 nm versus time after injection are shown with arrows indicating the retention time for cyclic IXP1 and linear IXP1 as determined from control assays without ClpXP. The area under the cyclic peptide peaks was unchanged after 60 min. (B) The effects of IXP1 on the ATPase activity of ClpX with and without GFP-tag, and on the peptidase activity of ClpP, were measured. Each assay was normalized to the activity in the absence of IXP1. Error bars indicate the standard deviation at each IXP1 concentration.

Figure 4.

Cyclic IXP1 inhibits degradation of λ O by ClpXP. λ O protein was incubated with ClpXP in the presence or absence of IXP1, and the loss of intact substrate was monitored by SDS-PAGE. Representative SDS-polyacrylamide gels stained with Coomassie blue showing the amount of λ O protein at various times after addition of ClpXP are shown. The amount of λ O protein remaining was plotted versus time and fit with a single exponential function to determine the substrate half-life. The average half-life for degradation of λ O was 35 ± 2 min in the absence of IXP1, and 73 ± 8 min in the presence of 100 μM IXP1.