(A) LPS induces an increase of about 10-fold in the number of Dipt-lacZ cells that stain positively for β-galactosidase. Ecdysone sensitizes the cells and promotes the response.
(B) Dipt-lacZ induction by LPS requires known Imd signaling components, but not Tl pathway members. The fraction of β-galactosidase-positive cells was normalized to the induced control (normalized %), and influence of RNAi of Tl pathway members (dif, spz, and tub) or Imd pathway members (PGRP-LC, Imd, Ird5, and Dredd) is shown.
(C–H) Activity stain (X-Gal) for β-galactosidase.
(C) Untreated cells.
(D) Cells treated with ecdysone alone.
(E) Cells treated with ecdysone and LPS. About 10% of cells express detectable β-galactosidase.
(F) RNAi against the DDRi sick reduces Dipt-lacZ expression in response to LPS.
(G) RNAi of a representative EDRi, the Ras signaling pathway component Cnk, enhances Dipt-lacZ induction by LPS.
(H) RNAi of a representative CDRi, the actin regulator SCAR induces Dipt-lacZ in the absence of LPS.
(I–J) Immunofluorescence of S2 cells with actin in red, tubulin in green, and DNA in blue. Scale bars in (I) and (J) indicate 10 μm.
(I) Wild-type cells have a characteristic rounded morphology.
(J) RNAi against many CDRi genes disrupts morphological features of wild-type S2 cells. S2 cells are shown treated with MESR4 dsRNA. Cells are significantly larger in appearance and less round, with irregular tubulin and actin networks.

The genes identified as DDRi (A), EDRi (B), and CDRi (C) are listed, and the colored bars show the influence of the corresponding dsRNA on Dipt-lacZ expression. The top two entries in (A), (B), and (C) show control cells (no dsRNA) without and with LPS, respectively. The scales for the false colors are given at the bottom left. Dipt-lacZ levels are given in terms of percent positive cells. For exact Dipt-lacZ expression values for each dsRNA refer to accompanying supplemental tables. In (B), the color scale (right) is compressed and extended compared to (A), and an asterisk indicates genes that also caused a CDRi phenotype. In (C), the pound sign indicates morphological defects and an asterisk indicates genes that also caused an EDRi phenotype, and the division of the genes into epistatic groups is shown. To the immediate left a false-color bar (coded as in [B]) indicates the effect of the dsRNAs on Dipt-lacZ expression without LPS addition. The block of colored columns shows the results of epistasis tests. Here, we set the undisturbed level of CDRi activation to 100% (as indicated in the left column in this group and the color code below), and to the right we represent reduction of this activation by prior RNAi of different Imd pathway genes. Five epistatic clusters (I–V) were identified (indicated by the lines to the left).

(A–B) Immunofluorescence of GFP-Relish cells with GFP-Relish in green, DNA in blue, and actin in red. Relish is predominantly cytoplasmic in untreated control cells and rapidly translocates to the nucleus of cells incubated with LPS.
(C) An anti-GFP Western blot of lysates harvested from GFP-Relish cells treated with LPS for different periods. GFP-Relish rapidly shifts from a full-length form to a shorter processed form after exposure to LPS, and full-length Relish gradually reaccumulates.
(D–E) Immunohistochemistry of GFP-Relish cells incubated with GFP-expressing E. coli (arrowheads) and treated with (E) or without (D) dsRNA against PGRP-LC. Imd pathway inactivation prevents bacterial-induced Relish nuclear translocation.
(F) Shows effects of treatment of GFP-Relish cells with DDRi dsRNAs for 4 d prior to LPS treatment. GFP-Relish was scored as cytoplasmic (uninduced), nuclear (induced), or reduced in amount (abnormal).
(G) Shows an epistatic analysis of the DDRi, sick. Suppression of sick interferes with Dipt-lacZ induction by Group III, IV, and V CDRi dsRNAs, but not those of Groups I and II, suggesting that Sick acts downstream of Imd and Dredd, but upstream of Relish in signal transduction.

(A) A comparison of the amino acid sequence of Dnr1 with XP_32149 from Anopheles gambiae and human MIR. Shaded regions indicate the N-terminal ezrin/radixin/moesin domain and C-terminal RING finger. Asterisks indicate conserved residues.
(B) Measurements of β-galactosidase activity in lysates prepared from Dipt-lacZ in control cells, LPS-treated cells, dnr1 dsRNA –treated cells, and sick dsRNA–treated cells exposed to LPS, respectively. Each experiment was performed in triplicate.
(C) Similarity between the RING finger in Dnr1 and other IAPs. Critical residues are shaded. Asterisks indicate conserved residues.
(D) Lysates from S2 cells transfected with equal amounts of N- and C-terminally HA-tagged wild-type Dnr1 (lanes 1 and 3, respectively), or N- and C-terminally HA-tagged C563Y Dnr1 (lanes 2 and 4, respectively), and analyzed by an anti-HA Western blot. Residue C563 is critical for RING finger function and is indicated with an arrowhead in (C).
(E) Higher resolution of lysates from C-terminally HA-tagged wild-type or C563Y Dnr1. Mutation of the RING finger prevents accumulation of a lower isoform of Dnr1.
(F and G) Subcellular localization of HA-Dnr1 transiently expressed in S2 cells treated without (F) or with (G) LPS, with HA in green, DNA in blue, and actin in red.

(A) Amounts of HA-Dnr1 transiently increase in S2 cells treated with LPS. Anti-HA Western blot of lysates from HA-Dnr1-transfected S2 cells that were incubated with LPS for indicated periods.
(B) Anti-HA Western blot of lysates from S2 cells transfected with HA-Dnr1. Coexpression of the caspase inhibitor p35 dramatically inhibits HA-Dnr1 accumulation in the absence (lanes 1 vs. 3) or presence (lanes 2 vs. 4) of LPS. Actin levels are shown as a loading control.
(C) Upper panel shows the percentage of cells with nuclear GFP-Relish after the indicated treatments. The lower panel is an anti-GFP Western blot of lysates from S2 cells treated in the identical manner. z-VAD-FMK prevents nuclear accumulation of GFP-Relish and GFP-Relish processing in response to LPS.
(D) Anti-HA Western blot of lysates from S2 cells transiently transfected with HA-Dnr1. While 2 h incubation with LPS normally leads to a 4-fold increase (quantified by titration) in HA-Dnr1 (lanes 1 vs. 2), incubation with z-VAD-FMK prevents the accumulation (lanes 3 vs. 4).
(E) Anti-HA Western blot of lysates from S2 cells transfected with HA-Dnr1. Cells had been previously incubated with (lanes 3 and 4) or without (lanes 1 and 2) Dredd dsRNA. Results are shown for two independent experiments. Actin levels are shown as a loading control.
(F) Anti-HA Western Blot of lysates from S2 cells transfected with HA-Dnr1 shows that prior RNAi against the caspases Dcp-1, Ice, Nc, and Decay does not substantially affect HA-Dnr1 levels (compare with control without LPS).
(G) The number of Dipt-lacZ-expressing cells after LPS treatment is greatly reduced after Dredd RNAi, while RNAi against Dcp-1, Ice, Nc, or Decay has no effect.

Pointed and blunt arrows indicate activation and inhibition, respectively. Both Sick and Dredd are required for translocation of Rel to the nucleus and for activation of Dipt expression and they are consequently positioned upstream of Rel as activators. In the absence of Sick or Dredd, Dnr1 function is not needed to maintain pathway quiescence. Thus, Dnr1 is ordinarily required to either inhibit Sick and Dredd functions or to negate their actions, and we have indicated these regulators as being downstream of Dnr1 (A). Although we have no epistatic data that separates the action of Sick and Dredd, Dredd appears to directly cleave Rel and is hence likely to be immediately upstream of Rel. Sick might function in conjunction with Dredd or as an activator of Dredd. Since dnr1 RNAi does not enhance the response to LPS, we suggest that its inhibitor activity is either repressed or bypassed upon exposure to LPS. Consequently, we have shown that treatment with LPS counteracts Dnr1-dependent Dredd inhibition (B). We do not mean to preclude other actions of LPS that might contribute to induction, but it is notable that inactivation of Dnr1 is sufficient to activate signaling. Finally, we have shown that Dnr1 levels are affected by Dredd, and we have indicated this with a positive feedback arrow.