Levels of gene expression do not necessarily correlate with gene function. Immunoblot detection of anti-βPS-integrin (Mys, top panels) and anti-Enabled (Ena, middle panels) after 3 days RNAi. Columns represent Kc₁₆₇ cells (left) and S2R⁺ cells (right) treated with different dsRNAs (gfp, ena, mys, R, talin). Both cell types expressed Mys and Ena in cells treated with a nonspecific dsRNA. The respective proteins were completely and specifically depleted by treatment with mys or ena dsRNAs. Anti-α-tubulin (bottom panels; Tub) shows a loading comparison.

A co-RNAi screen for modifiers of Pten-dsRNA phenotype. Microtubules are visualized by α-tubulin immunostaining. (a) Control Kc₁₆₇ cells exhibited normal, round morphology. (b) In response to Pten dsRNA at the same concentration as the original screening conditions, Kc₁₆₇ cells were bipolar and spindle-shaped with microtubule extensions (arrows). (c) In response to a relatively low concentration of Pten dsRNA, the conditions used for the modifier screen, Kc₁₆₇ cells exhibited a less pronounced phenotype with asymmetric microtubule accumulation (arrowheads). Specific dsRNA suppressors of the Pten-RNAi-induced cell shape restored the normal, round cell morphology and microtubule organization, identifying (d) Pi3K92E, (e) Akt1 and (f) LIMK1.

The distribution of phenotypic annotations. (a) Frequency of genes associated with a number of different RNAi phenotypes (0–6) per cell type. Phenotypes refer to those identified by seven major annotation categories. From 0 up to 6 phenotypes per gene were observed; '0' indicates those genes without detectable phenotypes in the one cell type (but were detected in the other). The set included all 160 genes identified by an RNAi phenotype in each of either S2R⁺ (gray) or Kc₁₆₇ (black) cell types. (b) The percentage of genes associated with a certain combined phenotypic annotation in both cell types screened. The percentage is the number of genes identified with 0 to 6 phenotypic annotations in Kc₁₆₇ cells (normalized to 100%) that were also associated with 0 to 6 phenotypic annotations in S2R+ cells (colored fractions of columns; see the key).

High-throughput RNAi screens by cell imaging. (a) Cellular phenotypes were visualized 3 days after the addition of dsRNA. In the example shown Kc₁₆₇ cells changed shape from round to polarized, with F-actin puncta (arrowhead) and extended microtubules (arrow), in response to Cdc42 dsRNA. (b) Kc₁₆₇ and (c) S2R⁺ cells at low (left) and high (far right) magnifications, fluorescently labeled for F-actin (red), α-tubulin (green) and DNA (blue). Cell-shape changes could be induced using drugs that affect the cytoskeleton or using extracellular signals, as seen upon treatment of Kc₁₆₇ cells with (d) latrunculin A or (e) 20-hydroxyecdysone (20-H-ecdysone). Scale bar, 30 μm.

RNAi screens identified a wide range of gene functions based on diverse morphological phenotypes. Cells were stained for F-actin (red), α-tubulin (green) and DNA (blue), imaged using automated microscopy and scored visually. (a) Control Kc₁₆₇ cells. (b-e) Kc₁₆₇ cells with RNAi phenotypes. (f) Control S2R⁺ cells. (g-j) S2R⁺ cells with RNAi phenotypes. (b) F-actin accumulation; CG13503 RNAi (encoding a predicted WH2-containing actin-binding protein). (c, h) Flatter, polarized cells with actin-rich lamellipodia (arrows); CG5169 RNAi (a predicted kinase). (d) Opposing protrusions rich in F-actin (arrow) or microtubules (arrowhead), Hsp83 RNAi (chaperone). (e) Flat cells; puckered RNAi (JNK phosphatase). (g) Widely-distributed F-actin puncta; capping protein beta RNAi (component of CapZ). (i) Radial protrusions (arrows) and reduced cortical actin (asterisk); CG31536 RNAi (predicted Rho-GEF with FERM domain). (j) Rounder cells, decreased in size; CG4629 RNAi (predicted kinase). Scale bar, 30 μm.

RNAi profiles identify known and novel genes with related morphogenetic functions. Table headings are as defined as in Figure 2. (a, b) Profile I: binucleate cells that identified genes required for cytokinesis, as detected either in (a) both cell types or (b) a single cell type. (c, d) Profile II: F-actin phenotypes observed in both cell types identified genes with potentially conserved roles in F-actin dynamics. (c) Increased or polarized (uneven) accumulation of F-actin identified genes with potential roles in F-actin capping, severing or depolymerization. (d) Reduced F-actin and altered cell shape identified genes with potential roles in F-actin polymerization. (e, f) Profile III: a common RNAi phenotype observed in Kc₁₆₇ cells was a change from round to spindle-shaped, with the formation of F-actin puncta and microtubule extensions. (e) Cases with phenotypes also observed in S2R⁺ cells identified genes involved in F-actin and microtubule regulation. (f) Cases with phenotypes observed only in Kc₁₆₇ cells identified components of receptor signaling pathways. (g-i) Profile IV: RNAi phenotypes resulting in round, detached S2R⁺ cells. (g) Phenotypes detected in both S2R⁺ and Kc₁₆₇ cells identified genes with probable indirect effects on cell adhesion and spreading, including roles in the cell cycle and cell viability; (h) RNAi phenotypes specific for S2R⁺ cells identified genes that may distinguish the flat S2R⁺ cell morphology, including genes encoding cell-matrix adhesion components. (i) Genes identified by a related RNAi phenotype, resulting in retracted (unspread but flat) S2R⁺ cells.

Similar phenotypic profiles identified genes in pathways and protein complexes. Cells were stained for F-actin (red), α-tubulin (green) and DNA (blue). Distinct phenotypes were observed with dsRNAs targeting different members of the same functional family (for example, GTPases, in the left panels). (a, b) Phenotypes observed in both cell types. (a) RNAi-induced binucleate cell phenotypes identified genes required for cytokinesis, including Rho1 (encoding a GTPase), pebble (a Rho-GEF) and CG10522 (a predicted citron kinase). Kc₁₆₇ cells are shown. (b) RNAi resulting in loss of actin filaments from the cell cortex identified regulators of actin-filament formation, including Cdc42 (GTPase), enabled (actin-binding protein) and SCAR (actin-binding, Arp2/3 regulator). Kc₁₆₇ cells (shown) also formed microtubule extensions and a polarized cell shape. (c, d) Some phenotypes were unique to one cell type. (c) RNAi resulting in round, non-adherent S2R⁺ cells identified genes required for cell-matrix adhesion, including Roughened (a Rap1 GTPase), Tenascin-major (an adhesion protein with a laminin domain) and myospheroid (β integrin). (d) An RNAi-induced amorphous S2R⁺ cell phenotype identified genes in the mitogen-activated protein (MAP) kinase pathway, including Ras85D (a GTPase), Downstream of raf1 (a MAP kinase kinase, or MEK) and kinase suppressor of Ras (a MAP kinase scaffold protein).

A test of RNAi screen efficacy: identifying genes involved in cell-cycle progression. (a) Gene identity and phenotypic annotation for RNAi phenotypes identifying predicted cell-cycle regulators. The 'Profile' column provides a summary of the phenotypic profiles distinguishing sets of genes involved in specific stages of the cell cycle. The 'Classification' column gives a single predicted functional category assigned to each targeted gene on the basis of primary sequence and/or known functional data. The 'FlyBase ID' and 'Gene name' columns are information as annotated at FlyBase [12]. The 'Predicted function' column provides detail on the putative molecular function of each specific gene. 'Cell type' refers to whether the phenotype was observed in Kc₁₆₇ (Kc) and/or S2R⁺ (S2R) cells. Profile I: RNAi phenotypes resulting in an increase in cell size, uniform or disorganized microtubules, irregular cell shapes and decreased cell numbers identified genes involved in cell-cycle progression through G1 to S and G2 to M stages. Phenotypes were further distinguished on the basis of levels of F-actin accumulation and DNA morphology. Profile II: RNAi phenotypes resulting in aberrant morphology or increased frequency of microtubule-based mitotic spindles identified genes involved in mitosis. Profile III: RNAi phenotypes observed in S2R⁺ cells identified additional genes with putative roles in cell cycle/mitosis progression. (b-g) Kc₁₆₇ cells stained for F-actin (red), α-tubulin (green), DNA (blue), imaged using automated microscopy and scored visually. (b) Control. (c, d) Profile I: Dp and string RNAi resulting in big cells. (e, f) Profile II: fizzy and polo RNAi resulting in increased frequency of cells with mitotic spindles. (g) Cdk5 RNAi resulting in smaller cells and disorganized microtubules (and increased spindles in S2R⁺ cells; not shown). Scale bar, 30 μm.