Landscape of protein-protein interactions in Drosophila immune deficiency signaling during bacterial challenge

Abstract

The Drosophila defense against pathogens largely relies on the activation of two signaling pathways: immune deficiency (IMD) and Toll. The IMD pathway is triggered mainly by Gram-negative bacteria, whereas the Toll pathway responds predominantly to Gram-positive bacteria and fungi. The activation of these pathways leads to the rapid induction of numerous NF-κB-induced immune response genes, including antimicrobial peptide genes. The IMD pathway shows significant similarities with the TNF receptor pathway. Recent evidence indicates that the IMD pathway is also activated in response to various noninfectious stimuli (i.e., inflammatory-like reactions). To gain a better understanding of the molecular machinery underlying the pleiotropic functions of this pathway, we first performed a comprehensive proteomics analysis to identify the proteins interacting with the 11 canonical members of the pathway initially identified by genetic studies. We identified 369 interacting proteins (corresponding to 291 genes) in heat-killed Escherichia coli-stimulated Drosophila S2 cells, 92% of which have human orthologs. A comparative analysis of gene ontology from fly or human gene annotation databases points to four significant common categories: (i) the NuA4, nucleosome acetyltransferase of H4, histone acetyltransferase complex, (ii) the switching defective/sucrose nonfermenting-type chromatin remodeling complex, (iii) transcription coactivator activity, and (iv) translation factor activity. Here we demonstrate that sumoylation of the IκB kinase homolog immune response-deficient 5 plays an important role in the induction of antimicrobial peptide genes through a highly conserved sumoylation consensus site during bacterial challenge. Taken together, the proteomics data presented here provide a unique avenue for a comparative functional analysis of proteins involved in innate immune reactions in flies and mammals.

Keywords: IMD interactome; functional proteomics; small ubiquitin-like modifier.

Fig. 1.

Overview of the functional proteomics analysis of the IMD pathway. (A) Work flow of the experiment. Light green nodes represent the bait proteins. The size of nodes is in relation to the numbers of interactants. (B) Number of identified proteins per bait. Black columns indicate the numbers of proteins found at all time points, and white columns indicate those found at some time points only. (C–G) Fold-change of numbers of identified proteins over time in each bait. Numbers of identified proteins at 0 min is considered the base.

Fig. 2.

Functional clusters common between humans and flies and their associated genes. GO terms of cellular components (A), biological processes (B), and molecular functions (C) shared between human and fly annotation databases are shown. Black and gray bars indicate fold-enrichment values for humans and flies, respectively. Four major parental GO terms are indicated with an asterisk.

Fig. 3.

Sumoylation cascade and its role in the defense against bacteria. (A) Ontology term enrichment in molecular function in the IMD interactome using the Drosophila database. (B) The survival of DmUbc9 (lwr) mutants (MUT) during bacterial challenge (WT, n = 43; MUT, n = 50). Statistical significance: *P < 0.0001 by log-rank test. (C) Bacterial growth over time postinfection (n = 3–4 per dot). Statistical significance: *P < 0.0001 by two-tailed Student t test (WT vs. MUT at day 10). Individual experiments are shown in dots; bars indicate the average. (D) Attacin A induction 6 h postinfection. The average and SD of three independent experiments are shown. Attacin A expression is normalized to that of RpL32 that encodes ribosomal protein. Statistical significance: *P = 0.0143 by Student t test (WT vs. MUT during infection). (E) SMT3 is covalently linked to IRD5 but not to KEY. His-tagged SMT3 was immunoprecipitated from cells challenged by administration of heat-killed E. coli at 0 min (control), 10 min, 2 h, and 8 h and HA-tagged IRD5 and KEY were detected by an anti-HA antibody. Five percent of input is shown as total.

Fig. 4.

K152 sumoylation of IRD5 is induced and critical for AMP induction. (A) Amino acid sequence alignment of the IRD5 homologs among 12 Drosophila species and four mammalian species. The sumoylation consensus motif is shown, Ψ-K-x-D/E: Ψ, hydrophobic amino acids; K, lysine; x, any amino acid; D/E, aspartate/glutamate. (B) Three-dimensional view of Xenopus IKKβ (PDB ID code 3RZF; visualized by CueMol). Orange-colored side chain (Inset) indicates lysine at amino acid position 152 (20). (C) K152 of IRD5 is sumoylated in a stimulation-dependent manner. His-tagged SMT3 was immunoprecipitated from cells transfected with WT vs. K152A IRD5 constructs and challenged by administration of heat-killed E. coli at 0 min (control), 10 min, 2 h, and 8 h; HA-tagged IRD5 WT and IRD5 K152A were detected by anti-HA antibody. Five percent of input is shown as total. (D) A time-dependent analysis of the levels of sumoylation in cells transfected with WT vs. K152A IRD5 constructs and challenged by administration of heat-killed E. coli at 0 min (control), 10 min, 2 h, and 8 h. The value observed at 0 min for WT constructs was set as 1 on the scale. The average and SD from three experiments are shown. One-way ANOVA was performed (P = 0.0009) and followed by Dunnett’s multiple comparison test. Asterisks indicate statistical significance (P < 0.05) compared with WT at 0 min. (E) IRD5 WT constructs restore Attacin A induction, but not the SUMO-mutant and kinase-dead forms. S2 cells were transfected with either GFP dsRNA as negative control or 3′ UTR ird5 dsRNA and various forms of ird5 (K152A, SUMO mutant; K50A, kinase dead) were individually expressed. Firefly luciferase activities of the reporter Attacin A firefly luciferase were measured and normalized to Renilla luciferase activities of Act5C-Renilla luciferase. The relative value against nonstimulated cells is shown. The value represents the average and SD from three independent experiments. One-way ANOVA was performed (P = 0.0003) and followed by Dunnett’s multiple comparison test. Asterisks indicate statistical significance (P < 0.05) compared with Vec/GFP:GFP dsRNA knockdown; empty vector transfection. (F) In vivo rescue experiment. Ird5 WT or K152A transgenic flies in an ird5-deficient background were established using ΦC31 transgenesis. Flies were challenged by live E. coli. The Attacin A mRNA level was measured by quantitative RT-PCR and normalized to RpL32. The relative values are indicated against nontransgenic ird5 mutants. Each value represents the average with SD, from three independent experiments. A pool of 5–7 adult flies per genotype was collected in each experiment. One-way ANOVA was performed (P < 0.0001) and followed by Dunnett’s multiple comparison test. Asterisk indicates statistical significance (P < 0.05) compared with nontransgenic ird5 mutants.