Bacterial production and direct functional screening of expanded molecular libraries for discovering inhibitors of protein aggregation

Abstract

Protein misfolding and aggregation are associated with a many human disorders, including Alzheimer's and Parkinson's diseases. Toward increasing the effectiveness of early-stage drug discovery for these conditions, we report a bacterial platform that enables the biosynthesis of molecular libraries with expanded diversities and their direct functional screening for discovering protein aggregation inhibitors. We illustrate this approach by performing, what is to our knowledge, the largest functional screen of small-size molecular entities described to date. We generated a combinatorial library of ~200 million drug-like, cyclic peptides and rapidly screened it for aggregation inhibitors against the amyloid-β peptide (Aβ42), linked to Alzheimer's disease. Through this procedure, we identified more than 400 macrocyclic compounds that efficiently reduce Aβ42 aggregation and toxicity in vitro and in vivo. Finally, we applied a combination of deep sequencing and mutagenesis analyses to demonstrate how this system can rapidly determine structure-activity relationships and define consensus motifs required for bioactivity.

Fig. 1. Construction and characterization of a combinatorial cyclic heptapeptide library with expanded diversity.

(A) Left: Representation of the pSICLOPPS-NuX₁X₂X₃X₄X₅X₆ vector library encoding the combinatorial heptapeptide library cyclo-NuX₁X₂X₃X₄X₅X₆. Nu: Cys, Ser, or Thr; X: any of the 20 natural amino acids; NNS: randomized codons, where N = A, T, C, or G and S = G or C; I_C: C-terminal domain of the Ssp DnaE split intein; I_N: N-terminal domain of the Ssp DnaE split intein. Right: Peptide cyclization using the SICLOPPS construct. Upon interaction between the two intein domains I_C and I_N, the encoded I_C-NuX₁X₂X₃X₄X₅X₆-I_N-CBD fusions undergo intein splicing and peptide cyclization, leading to the production of the cyclo-NuX₁X₂X₃X₄X₅X₆ library. (B) Western blot analysis of 12 randomly picked individual clones from the combinatorial heptapeptide library cyclo-NuX₁X₂X₃X₄X₅X₆, showing the expression and processing of the precursor fusion protein I_C-peptide-I_N-CBD. The 25-kDa band corresponds to the unprocessed precursor and the 20-kDa band to the processed I_N-CBD construct, and indicates, wherever present, successful intein splicing and peptide cyclization. Clone 10, for which the precursor is not expressed, was to contain a stop codon in its peptide-encoding region. (C) Heatmap representation of the amino acid distribution at each position of the constructed cyclo-CysX₁X₂X₃X₄X₅X₆ (left), cyclo-SerX₁X₂X₃X₄X₅X₆ (middle), and cyclo-ThrX₁X₂X₃X₄X₅X₆ (right) sublibraries, as demonstrated by the deep sequencing analysis results.

Fig. 2. Biosynthesis and ultrahigh-throughput screening of a combinatorial cyclic heptapeptide library with expanded diversity for discovering inhibitors of protein aggregation.

(A) Schematic of the used bacterial platform for discovering inhibitors of protein aggregation and for the high-throughput analysis of the selected hits. pMisP-GFP: plasmid encoding a misfolded protein-GFP fusion; pSICLOPPS-NuX₁X₂X₃X₄X₅X₆: vector library encoding the combinatorial heptapeptide library cyclo-NuX₁X₂X₃X₄X₅X₆; Nu: Cys, Ser, or Thr; X:, any of the 20 natural amino acids; FSC-H: forward scatter; SSC-H: side scatter; P: sorting gate. (B) FACS of E. coli Tuner (DE3) cells overexpressing Aβ42-GFP and the combined cyclic heptapeptide library. M: mean GFP fluorescence in arbitrary units. FITC-A: filter for fluorescein isothiocyanate. (C) Relative fluorescence of E. coli Tuner (DE3) cells overexpressing Aβ42-GFP and 10 randomly selected cyclic heptapeptide clones isolated after the seventh round of FACS shown in (B) and using either the wild-type split Ssp DnaE intein (green bars) or the splicing-deficient variant H24L/F26A (white bars) (25). Two randomly picked cyclic peptide sequences (random 1 and 2) previously shown to have no effect on Αβ42-GFP fluorescence and aggregation (18) were used as a negative control. The fluorescence of the bacterial population producing cyclic peptide random 1 was arbitrarily set to 100. Mean values ± SEM are presented (n = 3 independent experiments, each performed in three replicates). (D) Top: Western blot analysis of total (left) and soluble (right) lysates of E. coli Tuner (DE3) cells overexpressing Aβ42-GFP and the 10 individual cyclic peptide sequences tested in (C). The predicted molecular mass of the Aβ42-GFP fusion is ~32 kDa. Bottom: Western blotting using the anti-Aβ antibody 6E10 (left) and in-gel fluorescence (right) analyses of total lysates following native PAGE of E. coli Tuner (DE3) cells coexpressing Aβ42-GFP and the 10 individual cyclic peptide sequences tested in (C). (E) Emission spectra of E. coli Tuner (DE3) cells overexpressing Aβ42 along with four of the selected cyclic heptapeptide sequences tested in (C) and stained with ThS. The maximum fluorescence of the bacterial population producing cyclic peptide random 1 was arbitrarily set to 100. Mean values ± SEM are presented (n = 1 experiment performed in three replicates).

Fig. 3. Sequence analysis of the selected cyclic heptapeptide pool.

(A) Left: Frequency of appearance of the 20 natural amino acids at each position of the heptapeptide sequences selected after the seventh round of sorting (Fig. 2B). Right: Enrichment of the 20 natural amino acids at each position of the heptapeptide sequences selected after the seventh round of sorting (Fig. 2B). Values represent the log₂-fold change of the amino acid frequency of appearance of the peptides from the sorted pool compared to the initial library. (B) Visualization of the main clusters formed by the selected cyclic heptapeptides according to their sequence similarities. Nodes represent different cyclic peptide sequences, and solid lines connect pairs of peptides that share at least 70% sequence identity. The sequences of the members of the two most dominant clusters (clusters I and II) are shown in the corresponding dendrograms.

Fig. 4. AβC7-1 and AβC7-14 inhibit Aβ42 aggregation in vitro.

(A) Chemical structures of the selected cyclic heptapeptides AβC7-1 and AβC7-14. (B) Kinetic profiles of the aggregation of 2 μM Aβ42 in the absence and presence of AβC7-1 at different molar ratios (left) and the normalized t_1/2, t_lag, and t_growth values of the corresponding aggregation reactions (right). (C) As in (B) for AβC7-14. In (B) and (C), mean values ± SEM are presented (n = 1 experiment performed in three replicates).

Fig. 5. AβC7-1 and AβC7-14 inhibit Aβ42 aggregation in vivo.

(A) Normalized motility (left) and normalized speed of movement (right) of Aβ42 and wild-type worms in the absence and presence of 40 μΜ AβC7-1 and 5 μΜ AβC7-14 during days 5 to 10 of adulthood. (B) Motility (left) and speed (right) of individual Aβ and wild-type worms in the absence and presence of AβC7-1 and AβC7-14 at day 7 of adulthood. (C) Total fitness (51) of the worms as in (B). (D) Relative fluorescence of Aβ42 and wild-type worms at day 7 of adulthood showing a 50 to 60% decrease in Aβ42 aggregate formation in the presence of AβC7-1 and AβC7-14. (E) Representative images from (D). In (A) to (C), ~200 worms were analyzed on average, while in (D), 25 worms were analyzed in total. In all panels, mean values ± SEM are presented (n = number of worms tested in one experiment). Statistical significance is denoted by *P ≤ 0.05 and ****P ≤ 0.0001, for differences to the “No peptide Aβ worms” sample.

Fig. 6. Structure-activity analysis of the selected heptapeptides ΑβC7-1 and AβC7-14.

(A) Relative fluorescence of E. coli Tuner (DE3) cells overexpressing Aβ42-GFP and AβC7-1 (left) or AβC7-14 (right) or the indicated variants thereof as measured by flow cytometry. The fluorescence of the bacterial population coproducing the random cyclic peptide was arbitrarily set to 100. Experiments were carried out in triplicate (n = 1 experiment), and the reported values correspond to the mean value ± SEM. (B) Western blotting using the anti-Aβ antibody 6E10 (top) and in-gel fluorescence (bottom) analyses following native PAGE of total lysates of E. coli Tuner (DE3) cells coexpressing Aβ42-GFP and AβC7-1 (left) or AβC7-14 (right) along with the indicated variants thereof. (C) Heatmap representation of the amino acid distribution at each position of the peptide sequences corresponding to cluster I (Fig. 3B), as demonstrated by the deep sequencing analysis results. The total (left) or the unique (right) heptapeptide sequences were included in the analysis. (D) As in (C) for cluster II.