Expression and Cellular Location of iRhoms
(A–C) RNA in situ hybridization of wild-type and iRhom mutant Drosophila embryos, third-instar eye discs, and larval brains with a probe against iRhom.
(A) iRhom is expressed in the embryonic CNS, including the ventral nerve cord and brain.
(B) Third-instar larval eye discs showed staining posterior to the morphogenetic furrow (MF).
(C) iRhom expression is detected at low levels throughout the larval brain and is elevated in the optic lobes (arrowheads).
(D) HA-tagged Drosophila iRhom, human iRhom1, and mouse iRhom2 were located in the endoplasmic reticulum (ER) in COS7 cells. Drosophila Rho-1 localized to the Golgi apparatus, and the mouse rhomboid RHBDL3 localized to the Golgi apparatus, the plasma membrane, and punctate endosome-like structures. HA antibodies were used to detect iRhom; antiprotein disulfide isomerase (PDI) was an ER marker; and anti-P230 was used as a Golgi marker.
See also Figure S1.

Drosophila iRhom Inhibits EGFR Signaling
(A) The rough eye caused by sevenless-driven Rhomboid-1 overexpression (sev-rho-1/+) is strongly enhanced by halving the dose of iRhom (iRhom^KO1/+; sev-rho-1/+).
(B) Loss of iRhom suppressed the rough eyes caused by reduction of EGFR signaling mediated by UAS-driven: dominant negative Egfr (U-DN-Egfr); argos (U-argos); and sprouty (U-sty). The driver in all cases was the eye-specific GMR-Gal4.
(C) Expression of UAS-iRhom under the control of GMR-Gal4 suppressed the rough eye caused by sev-rhomboid-1 expression. This was particularly obvious at the anterior edge of the eye indicated by white arrowheads.
(D) Genetic interactions in the wing. Examples of wing phenotypes and severity levels indicated by wild-type (WT), mild (+), and severe (++) are shown. Wings of flies that were heterozygous for argos (argos^lΔ7/+) or iRhom (iRhom^KO1/+) were wild-type; 6.4% (± 6.4%) of wings that were heterozygous for sprouty (sty^S73/+) showed mild extra vein phenotypes (black arrowheads), typical of slight EGFR hyperactivation. A transheterozygous combination of sty^S73 and argos^lΔ7 enhanced the extra vein phenotype (74.5% ± 6.4% with mild extra veins). Similarly, halving iRhom in combination with sty^S73 (iRhom^KO1/+; sty^S73/+) enhanced the phenotype (62.5% ± 11.6% mild; 3.65% ± 5.4% severe). iRhom^KO1/+; sty^S73/ argos^lΔ7 flies showed a further increase in penetrance and strength of the extra vein phenotype (87.5% ± 2.3% mild; 10.3% ± 3.8% severe). All flies for this experiment were grown at 29°C. Error bars represent standard deviations of three to four independent experiments (n = 50 per experiment).
(E) iRhom^KO1/KO1 flies (n = 44) showed increased daytime sleep compared to controls (elav-GAL4, n = 37). Inhibition of EGFR signaling in the nervous system by expression of RNAi constructs against spitz (elav-GAL4; iRhom^KO1/KO1; UAS-spi^RNAi, n = 27) and Egfr (elav-GAL4; iRhom^KO1/KO1; UAS-Egfr^RNAi, n = 36) significantly (p < 0.0001) suppressed the iRhom^KO1/KO1 sleep phenotype. Error bars represent mean ± SEM.
See also Table S1.

iRhoms Induce Degradation of EGFR Ligands Independently of Active Rhomboids
(A and B) Secretion and intracellular levels of the indicated EGFR ligands were decreased by human iRhom1 (iR1-HA) and mouse iRhom2 (iR2-HA) but unaffected by mouse Unc93B.
(C) Secretion and intracellular levels of Flag-tagged Delta were unaffected by iR1, iR2, and Unc93B. EGFR ligands and Delta were detected with myc and Flag antibodies, respectively. The effects of mammalian iRhoms and Unc93B on intracellular levels of EGF, TGFα, and Delta were quantified from three independent experiments and plotted as relative percentages (error bars indicate standard deviations). Levels of iRhom1, iRhom2, and Unc93B were monitored by anti-HA antibodies in the cell extracts. In all panels, blots were probed with anti-actin to control for equal loading.
(D) HA-tagged human iRhom1 and mouse iRhom2, but not Unc93B, reduced secretion and intracellular levels of Flag-EGF-A1031F.
(E) iRhom1 and iRhom2 destabilize myc-EGF levels in cell extracts, but a KDEL-tagged (and therefore ER-localized) catalytically inactive mutant of mouse RHBDL2 (HA-R2SA-KDEL) has no effect.
See also Figure S3.

Generation of iRhom/rhomboid-5 Mutant Drosophila, Related to Figure 2
(A) At the top is the donor construct, as it would appear on the third chromosome (Chr III) when initially introduced by P-element transformation. FRT sites allow the FLP recombinase to free the DNA from its chromosomal integration site. This results in circular extra-chromosomal donor DNA, which is subsequently linearized by I-SceI. Homologous recombination can then occur between the donor construct comprising 5′ and 3′ homology arms either side of the mini-white marker gene (whs), which is flanked by loxP sites. Note that donor transformant lines carry the donor construct on the third chromosome and the iRhom/rhomboid-5 locus is on the second chromosome. Upon successful homologous recombination, the whs marker becomes genetically linked to the second chromosome, and the donor construct loses its FRT sites, which was exploited to enrich for correct targeting events as described in Experimental Procedures. The lower panels show the recombination event between the extra-chromosomal donor DNA (after FLP- mediated excision and I-SceI cutting) with the genomic locus containing iRhom/rhomboid-5. The black bar indicates the position of the DNA probe used to screen for correct site-specific integration by Southern blot. Note that the probe was generated against genomic sequence just outside the 5′ homology arm, to detect the wild-type 5kb genomic fragment after HinDIII digest upon site specific integration. Correct targeting replaces the iRhom genomic region with the whs marker, leading to the loss of a HinDIII site, which increases the predicted size of the genomic HinDIII DNA fragment to 11kb.
(B) Southern blot analyzing three independent mutant fly lines (KO1, KO2 and KO3) with the mini-white marker on chromosome II. All three homozygous mutant lines (KO1-3/KO1-3) showed the expected 11kb fragment predicted by the correct targeting, while a HinDIII digest of genomic DNA from heterozygous (+/KO1-3) lines resulted in a 5kb wild-type and 11kb mutant fragment.
(C) Genomic PCR strategy detecting a correct genomic targeting replacing the iRhom/rhomboid-5 locus with the mini-white gene. Note one primer was designed outside the 3′ homology arm, and the second within the mini-white gene; a PCR product can only be formed when correct targeting occurs. Correct integration was further confirmed by sequencing the PCR product covering the insertion region.

iRhoms Reduce Secretion and Intracellular Levels of Multiple EFGR Ligands, Related to Figure 6
(A–E) Secretion (sup, upper panel) and intracellular levels (extr, second panel from top) of the indicated EGFR ligands were decreased by human iRhom1 (iR1-HA) and mouse iRhom2 (iR2-HA) but unaffected by mouse Unc93B. Secretion and intracellular levels of Flag-tagged prolactin was unaffected by iR1, iR2 and Unc93B (E). EGFR ligands, and prolactin were detected with myc and Flag antibodies, respectively. Levels of iRhom1, iRhom2 and Unc93B were monitored by anti-HA antibodies in the cell extracts (extr, third panel from top). In all panels blots were probed with anti-actin to control for equal loading (extr, lower panel).
(F) In HeLa cells intracellular levels of myc-tagged EGF was decreased by human iRhom1 (iR1-HA) whereas levels of Flag-tagged Delta were unaffected by iR1-HA. EGF, and Delta were detected with myc and Flag antibodies, respectively. Levels of iR1 were monitored by anti-HA antibodies in the cell extracts. In all panels, blots were probed with anti-actin to control for equal loading.

iRhoms Are Inactive Members of the Rhomboid-like Family
(A) iRhoms are highly conserved and lack the catalytic serine and/or histidine (arrows) of active rhomboids. Distinctive iRhom elements are long N termini, a conserved iRhom homology domain (IRHD, blue), and an invariant proline residue (red) before the catalytic serine. Protein names and identifiers are indicated. Note that the gene for Drosophila iRhom (Swiss-Prot:Q76NQ1; Rhomboid-5) is ambiguously predicted (Lemberg and Freeman, 2007).
(B and C) Western blots assaying proteolytic activity in COS7 cells. Flag-tagged Gurken and Star (Grk+S) (B) or Flag-Spitz and Star (Spi+S) (C) were cotransfected with HA-tagged versions of Rhomboid-1 (R1), a catalytic serine-to-alanine mutant of Rhomboid-1 (SA), a mutant Rhomboid-1 with an iRhom-like proline (AP), or Drosophila iRhom. Cleavage of substrate inside the cells (extract) and secreted into the supernatant (sup) was detected with anti-Flag antibody. Cell extracts were also probed with anti-HA and anti-actin as controls for rhomboid expression and gel loading. Wild-type Rhomboid-1 cleaves and releases Spitz and Gurken; the SA mutant, AP mutant, and iRhom show no activity. Once cleaved, Spitz is rapidly secreted, explaining the absence of a cleavage product in the cell extract (Urban and Freeman, 2003).
(D) A similar experiment using mouse HA-tagged RHBDL2 (mR2) and myc-tagged mouse EGF. The alanine-to-proline mutation (AP) strongly reduced EGF secretion, and neither human iRhom1 nor mouse iRhom2 showed any catalytic activity. Mammalian iRhoms have been reported to be proteolytically processed (Nakagawa et al., 2005), explaining the multiple bands in cell extracts. This assay was done in the presence of 10 μM BB94 to inhibit metalloprotease shedding. Throughout the paper, “M” represents a mock transfection control and “SA” represents a serine-to-alanine catalytic mutant.
(E) The iRhom-like proline mutation in the active site of a bacterial rhomboid protease abolishes its enzymatic activity. AarA (WT), its catalytic serine-to-alanine mutant (S150A, SA), and iRhom-like mutant (A149P, AP) were overexpressed in E. coli and purified in the presence of detergent. The substrates TatA and LacY TM2 were in vitro translated and radiolabeled. Enzyme and substrates were incubated at 37° for 40 min. Wild-type AarA concentrations were 280 nM for TatA and 560 nM for LacY TM2. Mutants were equimolar to the wild-type or 5-fold higher when indicated (5×). Cleavage products (P) were separated from the substrates (S) by SDS PAGE and detected by autoradiography.

Loss of iRhom Results in Increased Sleep
(A and B) Activity levels over 2 days of male flies of indicated genotypes plotted as activity counts per hr (error bars represent mean ± SEM). Bars above the diagrams indicate light (white) and dark (black) periods of 12 hr each. Data represent measurements from two to four independent experiments.
(A) Activity patterns of heterozygous (iRhom^KO1/+; n = 18), homozygous iRhom mutant (iRhom^KO1/KO1; n = 16), and transheterozygotes (iRhom^KO1/Df (2L)J17; n = 17). iRhom homozygotes and transheterozygotes between iRhom^KO1 and Df(2L)J17 show highly suppressed activity levels.
(B) Neuronal expression of UAS-iRhom (n = 27), but not UAS-RFP (n = 24), by ELAV-Gal4 rescues the inactivity phenotype of iRhom^KO1/KO1 (n = 16) flies. Expression of iRhom in muscle under the control of how24B-Gal4 (n = 38) does not rescue the activity pattern.
(C) Waking activity, expressed as an average number of beam crossings in each minute in which activity was detected, was slightly increased in iRhom^KO1/KO1 (n = 17) flies compared to iRhom^KO1/+ (n = 16) and wild-type controls (n = 21). In this and subsequent panels, significance was determined with Student's t test (ns, not significant). Error bars represent mean ± SEM.
(D) The movement of wild-type (n = 5) and iRhom^KO1/KO1 (n = 5) flies was filmed. Fly movement was tracked during an active period, and the average walking speed (excluding short stops) was calculated in arbitrary units. Error bars represent mean ± SEM.
(E) Sleep, defined as periods of inactivity greater to or equal than 5 min, was significantly elevated in iRhom^KO1/KO1 flies (n = 39) compared to heterozygous (iRhom^KO1/+, n = 33) and wild-type (n = 44) controls. Error bars represent mean ± SEM.
See also Movie S1.

iRhom Inhibits Secretion of EGF Family Ligands
(A–C) COS7 cell supernatants (sup) were analyzed for Grk secretion, and cell extracts were blotted for levels of HA-tagged iRhom and Unc93B (Brinkmann et al., 2007). The ratios of iRhom to Drosophila Rhomboid-1 (R1) indicate relative amounts of transfected DNA.
(A) Rhomboid-1 (R1)-induced secretion of Flag-Grk was inhibited by increasing amounts of iRhom.
(B) Secretion of the FLAG-tagged Delta was unaffected by increasing amounts of iRhom.
(C) Overexpression of the ER resident polytopic membrane protein Unc93B did not interfere with Grk secretion.
(D) Mouse iRhom2 (iR2) inhibited mouse RHBDL2 (R2)-induced secretion of EGF. Secreted mouse EGF was detected by anti-EGF. The metalloprotease inhibitor BB94 suppressed nonspecific shedding of EGF. Increasing amounts of HA-tagged mouse iRhom2 (bottom) inhibited RHBDL2-mediated EGF secretion (top).
(E and F) Drosophila iRhom destabilizes intracellular Gurken. Rhomboid-1-mediated Grk processing was inhibited by iRhom (E), but not Unc93B (F). Histograms show relative substrate (S) to product (P) conversion from the western blots above. The iRhom-to-Rho1 (iRhom:R1) and Unc93B:R1 ratios indicate relative amounts of transfected DNA. Equal loading was confirmed by probing cell extracts for actin levels. Error bars represent mean ± SD.
(G) The iRhom effect was rescued by expression of ER-localized, but not Golgi-localized, active rhomboid. KDEL-tagged Rhomboid-1 (R1-KDEL), but not the inactive serine-to-alanine mutant Rhomboid-1 (Rho1 SA-KDEL), induced secretion of Grk in supernatants. A constant amount of iRhom inhibited Grk secretion (8× more iRhom DNA transfected than R1-KDEL). Increasing amounts of HA-tagged R1-KDEL rescued Grk secretion. Using the same approach, untagged Rhomboid-1, which is Golgi localized (see Figure 2D), did not rescue Grk secretion (right). HA-tagged iRhom, R1-KDEL, and R1 were detected in cell extracts. Flag-tagged Grk was detected in the supernatants (top) and extracts (middle). The Grk substrate band was absent when R1-KDEL is used because of efficient processing when substrate and enzyme are both in the ER.
See also Figure S2.

iRhoms Destabilize EGFR Ligands by ERAD
(A) Secretion and intracellular protein levels of myc-EGF were increased upon MG132 treatment at 10 μM for 12 hr (compare DMSO-treated lane 1 with MG132-treated lane 2 in supernatant and extracts). The inhibition of myc-EGF secretion and intracellular myc-EGF levels caused by human iRhom1-HA coexpression (compare lane 1 with lane 3) was rescued by MG132. The histogram quantifies the data: intracellular stabilization was expressed as the ratio of band intensity of MG132 treatment divided by DMSO control. MG132 stabilized steady-state levels of myc-EGF 3.4- ± 1.7-fold in the absence of coexpressed iRhom1 and 28.5- ± 15.2-fold in the presence of iRhom1. Error bars represent mean ± SD.
(B) EGF coimmunoprecipitates with both iRhom1 and iRhom2. The proteins were coexpressed in HEK293 cells as indicated. Two control proteins, TGN36 and prolactin (PRL), showed no interaction with either iRhom. The levels of each protein in the cell lysates are shown in the middle and bottom panels.
(C) A 10 min pulse of ³⁵S-methionine/cysteine was used to label Flag-tagged EGF, and its kinetics were followed for 3 hr ± MG132 and ± human iRhom1 (iR1). An autoradiogram of immunoprecipitated Flag-EGF (arrowhead) showed that iRhom1 reduced EGF levels at all chase time points (compare lanes 1/2, 5/6, and 9/10). MG132 increased EGF levels (compare lanes 1/3, 5/7, and 9/11) and rescued iRhom1-induced EGF destabilization (compare lanes 2/4, 6/8, and 10/12). The star indicates an unspecific degradation product. Quantification of the autoradiogram showed that, when the proteasome is inhibited, iRhom1 reduced the rate of secretion of EGF (compare slopes of red and purple lines).
(D) Expression of EGFR inhibitors Sprouty (GMR/+; U-sty/U-GFP) or Argos (GMR/+; U-argos/U-GFP) caused a rough eye. Reduction of ERAD by RNAi knockdown of Hrd1 (GMR/+; U-hrd1^RNAi) suppressed this EGFR inhibition.
(E) Model of iRhom function. EGF (in green) secretion is homeostatically regulated by ERAD; iRhoms (in red) bind to EGF, holding it in the ER and thereby enhancing ERAD and inhibiting secretion.
See also Figure S4.

iRhom Did Not Have an Impact on WNT3a Secretion Intracellular WNT3a or Delta Levels, Related to Figure 5
(A) Secretion of the HA-tagged WNT3a was detected by anti-HA antibody in the supernatants (upper panel). The iRhom to Drosophila Rhomboid-1 (R1) ratios (iRhom:R1) indicate amounts of transfected DNA. Increasing intracellular levels of HA-tagged iRhom was confirmed by anti-HA probing of the extracts (lower panel).
(B and C) Intracellular levels of HA-tagged mouse WNT3a and Drosophila Flag-tagged Delta were unaffected by elevated iRhom levels (upper panel). Equal loading was confirmed by probing the cell extracts with anti-actin antibody (lower panel).

iRhoms Destabilize EGFR Ligands by ERAD, Related to Figure 7
(A) Secretion of myc-EGF was increased upon lactacystin treatment at 10 μM for 12 hr (compare DMSO treated lane 1 with lactacystin treated lane 2 in supernatant and extracts). The inhibition of myc-EGF secretion and intracellular myc-EGF levels caused by human iRhom1-HA coexpression (compare lane 1 with lane 3) was rescued by lactacystin. Probing cell extracts with anti-HA antibody controls for iRhom1 and actin levels.
(B and C) EGF mRNA levels in the presence or absence of human iRhom1 (iR1) ± lactacystin or MG132 were measured by real time q-PCR. Inputs were normalized to endogenously expressed GAPDH and results graphed as relative fold changes of EGF mRNA levels. The data represents four independent experiments. In both cases the observed changes were not statistically significant (ns), as determined by Student's t test. Error bars represent mean ± SD.
(D) Expression of EGFR inhibitors Sprouty (GMR/+; U-sty/U-GFP) or Argos (GMR/+; U-argos/U-GFP) caused a rough eye. ERAD inhibition in the eye on (GMR/+; U-EDEM2^RNAi) effectively suppressed EGFR inhibition mediated by Sprouty (GMR/+; U-sty/U-EDEM2^RNAi) and Argos (GMR/+; U-argos/U- EDEM2^RNAi).