Genome-wide screen identifies curli amyloid fibril as a bacterial component promoting host neurodegeneration

Abstract

Growing evidence indicates that gut microbiota play a critical role in regulating the progression of neurodegenerative diseases such as Parkinson's disease. The molecular mechanism underlying such microbe-host interaction is unclear. In this study, by feeding Caenorhabditis elegans expressing human α-syn with Escherichia coli knockout mutants, we conducted a genome-wide screen to identify bacterial genes that promote host neurodegeneration. The screen yielded 38 genes that fall into several genetic pathways including curli formation, lipopolysaccharide assembly, and adenosylcobalamin synthesis among others. We then focused on the curli amyloid fibril and found that genetically deleting or pharmacologically inhibiting the curli major subunit CsgA in E. coli reduced α-syn-induced neuronal death, restored mitochondrial health, and improved neuronal functions. CsgA secreted by the bacteria colocalized with α-syn inside neurons and promoted α-syn aggregation through cross-seeding. Similarly, curli also promoted neurodegeneration in C. elegans models of Alzheimer's disease, amyotrophic lateral sclerosis, and Huntington's disease and in human neuroblastoma cells.

Keywords: Parkinson’s disease; curli; microbe–host interaction; neurodegeneration.

Fig. 1.

Genome-wide screen for proneurodegenerative genes in E. coli. (A) A flowchart of the screen using UM10 unkIs7[aex-3p::α-syn(A53T), dat-1p::gfp] and UM6 unkIs9 [dat-1p::α-syn(A53T), dat-1p::gfp] strains and the Keio library. (B) Representative images of uncoordinated movement and ADE neurodegeneration in UM10 animals fed with wild-type (WT) K12 E. coli and the restored locomotion and intact ADE neurons in animals fed with csgA(-) K12. (C) Percentage of UM10 animals with two ADE neurons when fed with WT, csgA(-), and csgB(-) K12 on various days into adulthood. (D) Locomotion rate of UM10 animals at L4 stage and day-2-adult stage when fed with WT, csgA(-), and csgB(-) K12. Mean ± SD were shown, and each dot represents one animal assayed. (E) Locomotion rate of WT N2 and UM6 animals on and off the bacteria lawn. UM6 animals exhibited impairment in food-induced basal slowing response when fed with WT K12, and the slowing response was restored when UM6 animals were fed with csgA(-) K12. Double asterisks indicate statistical significance (P < 0.01) in multiple comparisons using ANOVA analysis followed by a Tukey’s HSD post hoc test.

Fig. 2.

Bacterial curli production promotes α-syn–induced neurodegeneration. (A) WT and mutant K12 E. coli and WT and csgA(-) UTI2 E. coli were grown on Congo red indicator plates at 25 °C for 2 d to visualize curli production. Curli subunit mutants are in the black box, and mutants that were found in our screen and also showed reduced curli production are in the pink box. (B) Percentage of UM10 unkIs7[aex-3p::α-syn(A53T), dat-1p::gfp] animals with non-Unc phenotype at L4 stage or with two ADE neurons at day-2-adult stage when fed with either heat-killed WT, csgA(-), and csgB(-) K12 or a mixture of WT with csgA(-) or csgB(-) K12 at 1:99 ratio [indicated as 99% A(-) or B(-), respectively] or a mixture of csgA(-) and csgB(-) at 1:1 ratio [indicated as A(-):B(-) = 1:1]. Mean ± SD were shown, and each dot represents one independent experiment with 20∼30 animals scored. (C) Percentage of UM10 animals with non-Unc phenotype and two intact ADE neurons when fed with either K12 WT or csgA(-) at different developmental stages. The dotted lines indicate the timing of diet switch at specific stages from egg to day 1 adults (D1A) or day 2 adults (D2A). (D) WT K12 and UTI2 E. coli and their derivatives containing the csgA::3×FLAG genomic edits were grown on Congo red indicator plates with or without 200 μg/mL EGCG. Western blots of the bacteria lysate using anti-FLAG antibodies were used to confirm the inhibition of CsgA expression by EGCG. Anti-GAPDH blotting was used as a loading control. (E) Percentage of UM10 animals with non-Unc phenotype and two intact ADE neurons or the percentage of UM6 unkIs9 [dat-1p::α-syn(A53T), dat-1p::gfp] animals with two ADEs when grown on NGM plates that contained 200 μg/mL EGCG or empty vehicle and seeded with EGCG-treated or -untreated WT K12. For the treatment of K12 with EGCG, a single colony was cultured with LB medium containing 200 μg/mL EGCG overnight prior to seeding on NGM plate. The mean ± SD of three independent experiments (25 animals were scored for each experiment) is shown. Double asterisks indicate statistical significance (P < 0.01) in multiple comparisons using ANOVA analysis followed by a Tukey’s HSD post hoc test.

Fig. 3.

Bacterial curli promotes α-syn aggregation. (A) Representative images of α-syn::YFP aggregates in the muscle of NL5901 pkIs2386[unc-54p::α-syn::YFP; unc-119(+)] animals fed with different bacteria. Aggregates at the head region of day-1 adults were shown. (Scale bar: 20 μm.) For image quantification, the number of fluorescent aggregates in the same head area were quantified for 15 worms per group. Mean ± SD were shown. (B) Anti–α-syn antibody staining of UM10 unkIs7[aex-3p::α-syn(A53T), dat-1p::gfp] animals at L2 stage fed with WT or csgA(-) K12. (Scale bar: 20 μm.) (C) Anti–α-syn staining showed the α-syn aggregates (arrows) in the DA neurons of day-2 adults in UM6 unkIs9 [dat-1p::α-syn(A53T), dat-1p::gfp] animals. Animals fed with csgA(-) K12 showed less aggregation. GFP expressed from the dat-1 promoter labels the DA neurons. (Scale bar: 10 μm.) (D) Sequential fractionation of the lysate of NL5901 and UM10 animals fed with WT or csgA(-) K12 and different fractions were blotted by anti–α-syn antibodies in Western blot assays. Relative intensity of different fractions was quantified using ImageJ and normalized to whole animal lysate (WL). Anti-tubulin and anti-GFP blotting were used as internal controls. Arrow and asterisk indicate intact and degraded α-syn::YFP, respectively. (E) Western blot of α-syn and tubulin in the lysate of UM10 animals fed with WT K12 and treated with proteasome inhibitors MG132 (11 μM) and bortezomib (13 μM). (F) Percentage of UM10 animals with non-Unc phenotype and two intact ADE neurons under the treatment of ubiquitination inhibitor PYR-41 (1.4 mM) or proteasome inhibitors MG132 (11 μM) and bortezomib (13 μM). The mean ± SD of three independent experiments with 20∼30 animals is shown. Double asterisks indicate statistical significance (P < 0.01) in multiple comparisons using ANOVA analysis followed by a Tukey’s HSD post hoc test.

Fig. 4.

CsgA colocalizes with α-syn. (A) Day-5 adults of NL5901 pkIs2386[unc-54p::α-synuclein::YFP; unc-119(+)] animals fed with UTI2-csgA-3×FLAG bacteria were stained with anti-FLAG (red) antibodies. Insets show the enlarged region outlined by the dashed box and indicate the colocalization of CsgA and α-syn. (B) Day-5 adults of CGZ512 unkEx109[myo-3p::α-syn(A53T); dpy-5(+)] animals fed with UTI2-csgA-3×FLAG bacteria were stained with anti–α-syn (green) and anti-FLAG (red) antibodies. Insets are enlarged images of the boxed regions. (C) Day-1 adults of UM6 unkIs9[dat-1p::α-syn(A53T), dat-1p::gfp] animals fed with UTI2-csgA-3×FLAG bacteria were stained with both anti–α-syn (cyan) and anti-FLAG (red) antibodies. GFP signal indicates the position of the DA neurons. Colocalization of CsgA and α-syn were observed in all three types of DA neurons, CEP, ADE, and PDE neurons. Inserts are enlarged images of the boxed regions showing the colocalization in ADE and PDE neurons. (Scale bars: 20 μm.) Images were processed using the Leica THUNDER imaging system. The raw images can be found in SI Appendix, Fig. S5.

Fig. 5.

Bacterial curli promotes α-syn–induced mitochondrial dysfunction. (A) Venn diagram of genes that are down-regulated in UM10 unkIs7[aex-3p::α-syn(A53T), dat-1p::gfp] animals compared to N2 control and genes that are up-regulated in UM10 animals fed with csgA(-) K12 compared to animals fed with WT K12. Gene ontology analysis for down-regulated genes between N2 and UM10 using the Database for Annotation, Visualization and Integrated Discovery (DAVID) tools. (B) RT-qPCR measurement of mRNA level of mitochondrial genes alh-13, mel-32, ech-6, and hphd-1 in day-1 adults of UM10 animals fed with WT or csgA(-) K12 bacteria. Three biological replicates were performed, and mean ± SD were shown. (C) Basal respiration, ATP-linked respiration, and nonmitochondrial respiration (measured as oxygen consumption rate) of day-1 adults of N2, MQ887 isp-1(qm150), and UM10 animals fed with WT or csgA(-) K12 bacteria. Representative results of 6∼18 repeats for each condition were shown as mean ± SD. A total of 70 to 160 animals were added to each microplate well. (D) Representative images of the ALM neurons in CGZ833 unkIs7; twnEx8[mec-7p::tomm-20::mCherry; myo-2p::GFP] animals fed with WT or csgA(-) K12. (Scale bar: 2 μm.) Quantification shows the percentages of neurons with highly fragmented mitochondria in the soma in day-2 and day-6 adults. At least 20 animals were examined. (E) Representative images of zcIs13[hsp-6::GFP] and zcIs39[dve-1p::GFP] reporter expression in animals carrying unkIs7[aex-3p::α-syn(A53T), dat-1p::gfp] and fed with WT or csgA(-) K12 bacteria. Insets are enlarged images of the boxed region showing nuclear localization of DVE-1. (Scale bar: 25 μm.) Mean ± SD were shown for the quantification of the hsp-6p::GFP intensity and the number of intestinal cells showing DVE-1::GFP nuclear puncta (20 animals were analyzed for each experiment). Double asterisks indicate statistical significance (P < 0.01) in multiple comparisons using ANOVA analysis followed by a Tukey’s HSD post hoc test.

Fig. 6.

Bacterial curli promotes neurodegeneration in C. elegans models of ALS, AD, and HD. (A) Representative images of ALM neurons and ventral nerve cord (VNC) neurons in ALS strains carrying the transgene snb-1p::SOD1(G85R)::YFP or snb-1p::SOD1(WT)::YFP and fed with WT or csgA(-) K12. At least 30 day-2 adults were assessed. (B) The percentage of animals with visible ASER neurons in FDX25 sesIs25[flp-6p::Aβ1–42; gcy-5p::GFP; rol-6(D)] animals fed with WT or csgA(-) K12. For each condition, 50 animals were scored. (C) Learning index of day-1 adults of CL2355 smg-1(cc546) dvIs50 [snb-1p::Aβ1-42::3′UTR; mtl-2::GFP] animals grown on WT and csgA(-) K12 and WT and csgA(-) UTI2 bacteria in an associative learning assay. A total of 2 to 400 animals were used in each experiment; three biological replicates and three technical replicates were performed. Mean ± SD is shown. (D) Morphological changes of the ALM neurons and the alteration of Huntingtin (Htt) aggregation pattern in PLM and PVD neurons in TU6295 uIs115[mec-17p::TagRFP]; igIs5[mec-3p::htt57-128Q::GFP; lin-15(+)] animals fed with csgA(-) K12 compared to animals fed with WT K12. A total of 20 to 30 day-1 adults were assessed for each condition. (E) Harsh touch sensitivity of TU6295 animals fed with WT or csgA(-) K12 bacteria.

Fig. 7.

CsgA-derived amyloidogenic peptides cross-seed α-syn and induce neuronal death in human cells. (A) Representative anti–α-syn immunofluorescent images of SH-SY5Y cells transfected with α-syn(WT or A53T)–expressing constructs and then treated with CsgA-derived amyloidogenic hexapeptides, nonamyloidogenic control, or empty vehicle. (Scale bars: 20 μm.) (B) Corrected total cell fluorescence (shown as mean ± SD) from the experiments shown in A. Amyloidogenic peptides significantly enhanced α-syn expression and accumulation in the SH-SY5Y cells. (C) SH-SY5Y cells transfected with α-syn(WT or A53T)–expressing constructs and treated with rhodamine-conjugated QYGGNN peptides (red) were then stained with anti–α-syn antibodies (green) to show colocalization of CsgA-derived peptides and α-syn. Insets are enlarged images of the boxed regions. Images were processed using the Leica THUNDER imaging system. The raw images can be found in SI Appendix, Fig. S7B. (D) Intensity profile for the orange dashed line in C. (E) Representative results of cell viability assays using α-syn(WT or A53T)–expressing SH-SY5Y cells treated with amyloidogenic hexapeptides, nonamyloidogenic control, or empty vehicle. Mean ± SD is shown, and the single asterisk indicates P < 0.05 in a one-way ANOVA followed by a Tukey’s post hoc test.