A human retinal cDNA library was screened using a 647 base pair (bp) probe containing the WI-18706 STS (located at the RP26 locus, see Materials and methods). A total of 13 positive clones were isolated, subcloned in pBluescript II KS(+), and sequenced (Figure 1). Eight of the clones contained an apparently complete ORF, and the other five were truncated. The 5' and 3' ends of the messages were verified by rapid amplification of cDNA ends (RACE) using placental RNA as template. In the 5' experiment, two extended products were detected with the same 5' end but a differentially spliced 110 bp non-coding exon. The longer RACE product started 175 bp upstream of a putative initiation codon and this 5' untranslated region (5'-UTR) contained two in-frame stop codons. The shorter RACE product did not contain an in-frame stop codon. In the 3' experiment, a single extension product was detected which contained a polyadenylation signal (ATTAAA) situated 24 nucleotides 5' of the poly(A) tail. Some of the cDNA clones had an extended 3'-UTR which could be the result of the use of different polyadenylation signals further downstream. The full-length cDNA (1,092 bp) contained an ORF consisting of 462 bp, from nucleotides 176 to 637. The deduced protein chain consisted of 153 amino acids with an estimated molecular mass of 17.4 kDa.

When searching the nucleotide databases with the full-length human ORMDL1 cDNA, human homologous EST sequences were identified which belonged to two separate UniGene clusters (Hs.13144 and Hs.293711). Corresponding IMAGE cDNA clones were obtained and sequenced. The deduced ORFs (denoted ORMDL2 and ORMDL3) had the same size as ORMDL1. Comparison of the proteins showed between 80% and 84% positional identities (Table 1), and 116 out of 153 amino-acid residues were conserved between the three sequences. Moreover, in 26 of the 37 remaining positions the substitutions are conservative.

No homologous sequences were identified in Drosophila EST databases. Screening of an adult Drosophila melanogaster cDNA library using the human ORMDL1 cDNA as a heterologous probe was not successful either. We then designed a homologous probe based on the Drosophila genomic high-throughput sequence. Five positive clones were isolated and sequenced. Although none of the clones was full-length, one covered more than 80% of the ORF. The complete ORF could then be deduced by overlapping this sequence to the genomic data. The conceptual translated sequence was one amino acid longer at the amino terminus than the human forms and shared between 48% and 50% residue identities. Subsequently, in the completed Drosophila Genome Project, this sequence has appeared annotated as predicted gene CG14577 [3].

The ORMDL1 ORF was used for database sequence comparisons using the National Center for Biotechnology Information (NCBI) BLAST server [4,5] (tBLASTN against nucleotide databases, and BLASTP, PSI-BLAST and PHI-BLAST against the non-redundant protein database). Homologous protein sequences were found in yeast, microsporidia (Encephalitozoon cuniculi, an opportunistic pathogen in AIDS), plants (including Arabidopsis), invertebrates (Drosophila), urochordates (Ciona intestinalis) and vertebrates. The search for distant relatives using PSI-BLAST and PHI-BLAST programs and for short, nearly exact, matches (BLASTP) did not yield any additional homologs. In vertebrates, three different genes were distinguished, corresponding to ORMDL1, ORMDL2 and ORMDL3. S. cerevisiae and A. thaliana showed two copies each, while a single copy was found in C. intestinalis, D. melanogaster, Saccharomyces monacensis, Schizosaccharomyces pombe and E. cuniculi. Table 2 lists most of the genes, with their corresponding annotations in the public databases, if available, and summarizes other relevant features.

When all the protein sequences were aligned using CLUSTALW, several conserved domains became evident (Figure 2). In addition, all yeast proteins were longer, at both the amino- and carboxy-terminal ends. Pairwise comparisons revealed a very high level of identity (Table 1). Searches for characterized functional domains did not reveal any significant homology. However, four putative transmembrane domains were detected when considering the protein alignment of all known members of the family. Similar results were obtained when analyzing the ORMDL1 sequence on its own (Figure 2).

A phylogenetic tree using all the available protein sequences was drawn with CLUSTALW and the bootstrap values were also calculated (Figure 3). According to this tree, each human ORMDL gene has an extremely conserved counterpart in all vertebrates analyzed. Vertebrate sequences cluster together in a separate branch. Similarly, yeast, microsporidian and plant ORMDLs group in well differentiated lineages.

The full-length human ORMDL1 cDNA clone was compared to non-redundant DNA databases and the 5' end was found to be homologous to the promoter region of the PMS1 gene [6]. The P1 genomic clone 1670, containing this region, was analyzed by restriction mapping and Southern blot with the ORMDL1 cDNA as a probe. Positive fragments were sub-cloned and sequenced. The nucleotide sequence contained 4.8 kilobases (kb) of the upstream region, all of the five exons, and the four introns (Figure 4). The third intron was not fully sequenced and the size was determined by restriction mapping after amplification by polymerase chain reaction (PCR). All introns were flanked by the consensus donor and acceptor splice sites. No putative TATA box was found on the upstream genomic sequence, although several Sp1-binding sites were detected using the TESS server [7]. Using the EMBL CpG Islands service [8] a presumptive 0.9 kb CpG island containing 0.4 kb of the upstream region, the first exon of the 5'-UTR, and parts of the first intron was detected.

Database searches against releases of the public Human Genome Project [4] confirmed the genomic ORMDL1 structure. Furthermore, the genomic structures of the coding region of the two human paralogs ORMDL2 and ORMDL3 and the D. melanogaster homolog were deduced after alignment of the characterized human and Drosophila cDNAs with genomic sequence databases. For the A. thaliana ORMDL homolog, tBLASTN searches were carried out with the human ORMDL proteins against the Arabidopsis genome. The data obtained were contrasted with gene predictions from the Arabidopsis Genome Initiative (actually, one of these predictions has already been confirmed; GenBank accession no. AF360237). In all these genes a genomic structure of three coding exons is totally conserved, including the positions of the exon-intron boundaries ( 4). In contrast, the yeast homologs were each encoded in one single continuous ORF. Interestingly, the introns of human ORMDL2 and ORMDL3 genes were found to be much smaller than those of ORMDL1. The human ORMDL homologs mapped to different chromosomes from ORMDL1 (ORMDL2 to 12q13.2 and ORMDL3 to 17q21.1). No further human homologs were detected in the public databanks [4] or in the Celera database [9], except for two putative processed pseudogenes. One of these showed 86% positional identity to ORMDL1 and mapped to chromosome 10p14, and the other showed 84% identity to ORMDL2 and mapped to 8q22.1. Both ORMDL1 and ORMDL2 lacked introns and appeared to have acquired several indels (insertions/deletions). In addition, ORMDL2 lacked the first coding exon of ORMDL2.

To study the expression pattern of human ORMDL genes, RT-PCR and northern analyses were performed. RT-PCR expression analysis on sixteen adult and fetal tissue samples showed a ubiquitous expression pattern for all three human genes (Figure 5a). Northern analysis using full-length as well as 3'-UTR ORMDL1 cDNA probes identified three transcripts of 1.4, 2.5 and 3.3 kb in most human adult tissues (Figure 5b). The expression of ORMDL1 was moderately high in pancreas, placenta and brain but low in skeletal muscle and lung.

To determine the subcellular localization of the ORMDL proteins, the cDNAs corresponding to the human ORMDL genes were cloned into pEGFP-derived vectors and transiently expressed in transfected COS-7 cells. Twenty-four hours after transfection, fusion proteins of enhanced green fluorescent protein and ORMDL (EGFP-ORMDL) appeared predominantly in the perinuclear endoplasmic reticulum (ER) and, to a lesser extent, throughout the extended ER network (Figure 6a). Confocal scanning microscopy using an antibody against protein-disulfide isomerase (PDI), an ER-marker protein, showed an overlapping signal in the endoplasmic reticulum (Figure 6a,6b,6c). In contrast, double-label fluorescence with biotin-labeled concanavalin A (which shows affinity for terminal α-D-mannosyl and α-D-glucosyl residues, and thus binds to the plasma membrane in non-permeabilized cells), detected with Texas Red-conjugated streptavidin, showed no overlapping signal with EGFP-ORMDL at the plasmatic membrane (Figure 6d,6e,6f). The subcellular patterns observed for EGFP-ORMDL1, EGFP-ORMDL2 and EGFP-ORMDL3 fusion proteins were similar, in either fixed (Figure 6g,6h,6i, respectively) or in vivo cells (Figure 6j,6k), co-localizing with an in vivo ER marker. Similar results were obtained from COS-7 cells transfected with either pEGFP-N2- or pEGFP-C2-derived constructs. In contrast, EGFP alone was uniformly distributed throughout the cell, including the nucleus (Figure 6l), in all experiments. Overall, these data show that the ORMDL proteins locate at the ER membrane in agreement with their predicted transmembrane topology.

We analyzed the expression pattern of the Drosophila ORMDL homolog (DORMDL) during different developmental stages. To this end, a 391 bp antisense riboprobe containing most of the coding sequence was used for in situ hybridization on wild-type Canton S D. melanogaster 0 h to 24 h embryos (stages 1 to 17). DORMDL was ubiquitously and homogeneously expressed in the syncytial blastoderm and during the cellularization stage (Figure 7a,7b). In stages 11-12, the germ-band layer (which later gives rise to all somatic cell lines, including ectoderm, mesoderm and endoderm) gave a positive hybridization signal (Figure 7c,7d). In later stages (stage 14), the signal was mainly detected in the ectodermal tissues (Figure 7e). The DORMDL riboprobe was also used to analyze the expression pattern in imaginal discs in third-instar larvae. Uniform and homogeneous expression was observed in eye-antenna, leg, wing, (Figure 7g,7h,7i) and haltere (data not shown) imaginal discs, which was consistent with the former ectodermal expression detected in late-stage embryos.

The S. cerevisiae genome contained two members of this family: ORM1 (ORF YGR038w) at chromosome VII and ORM2 (ORF YLR350w) at chromosome XII. To provide functional clues for the human homologs, we generated and analyzed single and double knockouts. We used the W303-1A and W303-1B yeast wild-type strains (see genotypes in Materials and methods) and the kanMX4 cassette for gene disruption [10]. Single gene targetings were assessed by designing specific PCR reactions. The double knockout mutant was obtained by mating the two single knockout strains, subsequent sporulation, tetrad dissection and analysis of single spores. Verification of ORM1 and ORM2 gene integrity and/or deletion for the clones derived from each spore was carried out by independent PCRs with specific primers (Table 3; Figure 8). Overall, 42 independent clones were analyzed, from which 23.8% were wild-type for ORM1 and ORM2 (and thus did not grow on G418-supplemented medium), 28.57% single ORM1 deletants, 28.57% single ORM2 deletants, 14.28% double knockouts and 4.76% diploids. The two single and double knockouts were viable and no morphological differences were detected between them and the wild-type cells. However, growth rates on rich media (YPD) differed: the double mutant grew much slower than the wild-type and single-knockout strains (data not shown).

A battery of tests was carried out to characterize the yeast mutants on the basis of their sensitivity or resistance to toxic compounds. The YPAD medium was supplemented with one of the following: dithiothreitol (DTT), tunicamycin, HgCl₂, cycloheximide, CaCl₂, KCl, caffeine, EGTA or SDS. In all cases the double-knockout strain was more sensitive to the toxic agent. In contrast, the growth of the single knockouts did not differ from that of the wild-type strain. This result was particularly evident in the plates supplemented with DTT, HgCl₂ or tunicamycin (Figure 8b,8c). We also carried out complementation analyses with ORMDL3, the human counterpart with the greatest sequence similarity to yeast ORM1 and ORM2. To this end, double-knockout yeast cells were transformed with either multicopy or centromeric expression vectors bearing ORMDL3 under the control of a constitutive promoter, and dropout-plated in media supplemented with DTT, HgCl₂ or tunicamycin. In all cases, some degree of phenotypic rescue was clearly observed, further arguing in favor of functional conservation in the ORMDL family (Figure 8c).

A PCR-derived probe of 647 bp was obtained from an IMAGE clone (220362, Research Genetics) using a primer pair (5'-GGGAACAACTGGACTATG-3' and 5'-ACTTGTATAGAGAGGATTCC-3') designed after the retinal EST H86933. This EST contained the STS WI-18706, which mapped to the RP26 locus. The probe was labeled with [α-³²P]dCTP by random priming (Roche Molecular Biochemicals) and was used to screen 10⁶ recombinant phages from a commercial λ gt10 retinal cDNA library (Clontech). Hybridization was carried out using standard protocols under highly stringent conditions [23]. Positive λ phages were isolated through several rounds of rescreening. Purified phage DNA was EcoRI-digested and inserts were subcloned into pBluescript II KS(+) (Stratagene). Sequencing was carried out using a ThermoSequenase II sequencing kit (Amersham Biosciences) and an ABI377 automatic sequencer (Applied Biosystems).

The 5' and 3' transcript ends were characterized using the SMART RACE cDNA amplification kit (Clontech) according to the manufacturer's instructions. Total RNA from human placenta was used for first-strand cDNA synthesis using SuperScript II RNase H^- reverse transcriptase (Invitrogen Life Technologies). The primers used were: 5'-AGGAGCCTTGCTTTACCCTGGTCAG-3' for the 5'-RACE and 5'-CAATGGCTGGTCCTTCAAGTGCTGT-3' for the 3'-RACE. The RACE products were subcloned using the SureClone ligation kit (Amersham Biosciences) and sequenced.

BLAST searches [4,5] against the EST section of the GenBank/EMBL databases identified sequences homologous to ORMDL1, which belonged to the two UniGene clusters Hs.13144 and Hs.293711. The IMAGE clones 212500, 221470, 1407036, 1407667, and 2364281 were obtained from the UK HGMP Resource Centre and sequenced.

A search for homologous protein sequences was carried out with tBLASTN, BLASTP, PSI-BLAST and PHI-BLAST programs at the NCBI BLAST server [4,5]. Translated sequences were aligned using the program CLUSTALW [24], and visualized using the program Boxshade (created by Kay Hofmann and Michael D. Baron). Phylogenetic trees were constructed with the program CLUSTALW, excluding gap positions and correcting for multiple substitutions. Bootstrap analysis [25] was used for confidence evaluation. Potential transmembrane segments were predicted using the HMMTOP, TMAP and TMPRED programs [26,27,28]. Conserved protein domains and signal peptide searches were performed with reverse position-specific BLAST at the NCBI Conserved Domain Database, including Pfam and Smart databases [4,5,29,30,31]. Analysis of putative binding sites for transcription factors was carried out using the TESS server [7]. CpG islands were detected using the EMBL CpG Islands service [8]. The human ORMDL sequences were mapped using the BLAT program against the Human Genome Working Draft (December 22, 2001 assembly) at UCSC [32,33].

Accession numbers of sequences used are as follows. This work: translations from AF395704 (HsapORMDL1), AF395705 (HsapORMDL2), and AF395708 (HsapORMDL3). From data banks: AAH02146 (MmusORMDL2), BAB26397 (MmusORMDL3), AAF51758 (DmelORMDL), AAK25947 (AthaORMDLa), BAB08433 (AthaORMDLb), CAA97026 (ScerORM1), AAB67252 (ScerORM2), CAA17924 (SpomORM), and translations from BE625123 (MmusORMDL1), AC098491 (RnorORMDL3), BI182456 (SscrORMDL1), BE588447 (BtauORMDL2), BM486897 (GgalORMDL2), BM489594 (GgalORMDL3), BJ037340 (XlaeORMDL1), BJ067072 (XlaeORMDL2), AL633210 (StroORMDL2), fugu clone T008329 (FrubORMDL1), fugu clone T005201 (FrubORMDL2), fugu clone T008602 (FrubORMDL3), AL591593 (DrerORMDL1), AL669202+AV885287 (CintORMDL), BE421632 (HvulORMDL), AW747032 (SbicORMDLa), AW678224 (SbicORMDLb), BG522023 (SrebORMDL), AI772256 (LescORMDLa), AI899675 (LescORMDLb), BM093316 (GmaxORMDL), BE202511 (MtruORMDL), BG140432 (LpenORMDLb), BE238653 (ZmayORMDLb), Y08688 (SmonORM1), and AL590450 (EcunORMDL).

Sequence data for this article were deposited with the GenBank data library under accession nos. AF395701 (ORMDL1 exons 1, 2, and 3); AF395702 (ORMDL1 exons 4 and 5); AF395704 (ORMDL1 mRNA); AF395705, AF395706, AF395707 (ORMDL2 mRNA); AF395708 (ORMDL3 mRNA); AF395703 (DORMDL mRNA).

The human P1 genomic clone 1670, mapped to 2q31-2q33 [6], was obtained from Incyte Genomics. Plasmid DNA was obtained according to the supplier's instructions. Restriction fragments containing the ORMDL1 exons were detected by Southern blot analysis using a full-length ORMDL1 cDNA probe. Positive fragments were subcloned into pBluescript II KS(+), and sequenced. The size of intron 3 was determined by restriction mapping after PCR-amplification.

For RT-PCR expression analysis on 16 samples from adult and fetal tissues (Multiple Tissue cDNA Panels, Clontech), a pair of primers was designed for each human ORMDL gene to amplify the complete ORF, as follows: ORMDL1 forward: 5'-CATGAACGTTGGAGTTGCC-3', ORMDL1 reverse: 5'-GTAAAATTTTTTTTCAGTTTCAAA-3', ORMDL2 forward: GAAGAATGAATGTGGGGGTG-3', ORMDL2 reverse: 5'-CATGGAGCTGTCCCAAAAC-3', ORMDL3 forward: 5'-GCAGGATGAATGTGGGCACA-3', ORMDL3 reverse: 5'-GGCAGGGGAAGGGGCTGCA-3'.

PCR amplifications were carried out in 25 μl reactions containing 2.5 μl template cDNA, 0.2 mM dNTPs, 5 μM each primer, 20 mM Tris-HCl, 50 mM KCl, 3 mM MgCl₂, and 1 U Taq Platinum (Invitrogen Life Technologies). After an initial 3 min step at 94°C, reactions were subjected to 35 cycles of a denaturation step of 30 sec at 94°C, and an annealing-extension step of 45 sec at either 59°C (ORMDL1), 62°C (ORMDL2), or 65°C (ORMDL3). A final extension step was performed at 72°C for 5 min.

For northern blot analysis, radiolabeled full-length and 3'-UTR ORMDL1 cDNA probes were hybridized against a commercial human Multiple Tissue Northern blot (Clontech) using standard highly stringent conditions. A human actin probe was used as a control for RNA loading.

COS-7 (African green monkey kidney) cells were routinely grown in DMEM containing 10% fetal calf serum, 2 mM L-glutamine, 100 U/ml of penicillin and 100 μg/ml of streptomycin (Invitrogen Life Technologies). ORMDL1, ORMDL2 and ORMDL3 cDNAs were cloned in-frame either upstream or downstream of the gene for GFP using the pEGFP-N2 and pEGFP-C2 vectors (Clontech), respectively. All constructs were verified by sequencing. Transient transfections on COS-7 cells grown on coverslips were performed using the FuGENE reagent (Roche Molecular Biochemicals) following the manufacturer's protocol (the ratio was 1.5 μg DNA to 7.5 μl FuGENE). Empty pEGFP-N2 and pEGFP-C2 vectors were used as controls. Twenty-four hours after transfection, cells were rinsed with 100 mM PBS and fixed in 3% paraformaldehyde and 2% sucrose in 0.1 M phosphate buffer at 4°C for 45 min, then washed and permeabilized (if required) with 0.1% Triton X-100/20 mM glycine/10 mM PBS for 10 min. Following permeabilization, cells were rinsed and blocked in 0.5% BSA/20 mM glycine/10 mM PBS. Cells were incubated either with a specific anti-PDI primary antibody (permeabilized cells) or with biotin-labeled concanavalin A (Sigma) (non-permeabilized cells) at 37°C for 1 h. Upon washing, cells were incubated either with Rhodamine Red™-conjugated anti-rabbit secondary antibody (Jackson ImmunoResearch Laboratories) (permeabilized cells) or with Texas Red-streptavidin (Amersham Biosciences) (non-permeabilized cells) at 37°C for 45 min. In vivo co-localization with the ER marker (ER-Tracker Blue-White DPX, from Molecular Probes) was performed 24 h after transfection by rinsing coverslips twice with PBS followed by 30 min incubation at 37°C in a 4 μM ER-Tracker solution. All preparations were embedded in Vectashield mounting medium (Vector Laboratories). Fluorescence was analyzed with a Zeiss Axiophot epifluorescence microscope. Confocal laser scanning microscopy was performed with an Olympus IX-70 inverted laser scanning microscope.

A 529-bp probe was amplified by PCR from genomic D. melanogaster DNA. The primers used (5'-TCAGTGCC-CTTCGTTAGC-3' and 5'-AGTGTTCCGTGTTGTTTC-3') were designed after the high-throughput genomic sequences (GenBank) that were identified in a BLAST search using the human ORMDL1 cDNA sequence. The radiolabeled probe was used to screen 8 × 10⁵ recombinant phages from an adult D. melanogaster cDNA library (Stratagene) constructed in Lambda ZAP II. Hybridization, washes, and autoradiography were carried out in standard conditions. After several rescreening rounds, pBluescript SK(+) plasmids containing the cDNA insert were excised from the isolated phages and sequenced.

Digoxigenin (DIG)-labeled antisense and sense riboprobes covering 84% of the Drosophila ORMDL ORF were synthesized using the DIG RNA labeling kit (Roche Molecular Biochemicals) according to the manufacturer's instructions. Drosophila embryos at different developmental stages were collected from a 24-h lay, dechorionated, fixed in 2% formaldehyde and 0.5 M EGTA in PBS, prehybridized for 1 h, and hybridized overnight at 55°C. After removing the hybridization solution and washing, embryos were incubated with 1/2000 anti-DIG antibody conjugated to alkaline phosphatase (Roche Molecular Biochemicals), preabsorbed against fixed embryos. The signal was detected with NBT/BCIP (Roche Molecular Biochemicals). Embryos were mounted on Permount, and examined and photographed under a Zeiss Axiophot microscope.

Wild-type D. melanogaster Canton S third-instar larvae were dissected in PBS and fixed in 4% paraformaldehyde in PBS for 20 min. After rinsing in PBS, a second round of fixation was performed in 4% paraformaldehyde in PBS with addition of 0.1% Triton X-100 and 0.1% deoxycholate. After further rinsing with PBS, the imaginal discs were dehydrated through a graded ethanol series. Further protocol steps before the hybridization were carried out as described [34] except that proteinase K digestion was omitted. After 1 h prehybridization at 60°C, the riboprobe was added hybridized overnight at 65°C. Thereafter, the discs were washed for 20 min each in a series of decreasing concentrations of hybridization solution in PBT at 65°C. Two final washes were done with PBT at room temperature. After 1 h of blocking with 2% BSA in PBT, the discs were incubated with preabsorbed anti-DIG, washed and detected as described above. Dissected discs were then mounted in glycerol, examined and photographed under a Zeiss Axiophot microscope.

S. cerevisiae W303-1A (MATa; ura3-1, ade2-1, leu2-3,112; his3-11,15; trp1-; cam-100) and W303-1B (MATα; ura3-1, ade2-1, leu2-3,112; his3-11,15; trp1-2; can1-100) strains were grown on yeast extract peptone dextrose (YPD) media. G418-resistant strains were grown on YPD plates containing 200 mg/l of G418 sulphate (geneticin, Invitrogen Life Technologies).

To disrupt YGR038w and YLR350W ORFs, kanMX4 substitution cassettes were PCR-amplified on plasmid pFA6a-kanMX4 [10] using two sets of primers with 18-19 nucleotide homology to the pFA6a MCS plus 45 nucleotide homology to the flanking regions of either YGR038w (ORM1) or YLR350W (ORM2) ORFs (Table 3). One microgram of each gel-purified PCR product was used to transform yeast W303-1A and W303-1B strains by the lithium acetate method [35]. Transformed yeast cells were recovered by G418 selection. Homologous integration of the kanMX4 cassette was verified by colony PCR. Two sets of primers were used for each gene to test for gene disruption and gene integrity, respectively (Table 3). Primer design and PCR conditions essentially followed the EUROFAN guidelines [36].

The W303-1A Δorm1 (MATa) and W303-1B Δorm2 (MATα) deletant strains were crossed to obtain the double heterozygous diploid strain. The reciprocal cross was also performed. The double heterozygous diploids were plated on sporulation medium (1% potassium acetate, 0.1% yeast extract, 0.05% dextrose, 2% agar) to induce meiosis [37]. Tetrad dissection and analysis were carried out as described [37]. Clones from spores were re-streaked on YPD plates containing 200 mg/l G418, and G418-resistant transformants were verified by colony PCR to assess single and double knockouts.

To test growth differences and sensitivity/resistance of the mutant strains against toxic compounds, single and double knockouts as well as wild-type clones were grown on liquid yeast extract peptone adenine dextrose (YPAD) medium for 14 h at 30°C, diluted to OD₆₀₀ 0.1 and grown to OD₆₀₀ 0.6-0.8. They were then sequentially diluted (5 dilutions, 1:5 each) with YPAD into the wells of microtiter plates and dropout-plated onto the corresponding test media (0.25 mM and 0.4 mM HgCl₂; 10 mM and 15 mM DTT; 0.5 μg/ml and 1 μg/ml tunicamycin; 0.7 M CaCl₂; 0.1% caffeine; 0.18 mg/ml and decreasing concentrations of cycloheximide; 3 mM EGTA; 1.5 M KCl; 0.01% SDS, all in YPAD). To test heat sensitivity, cultures were incubated in a water bath at 52°C for 35 min, sequentially diluted and plated on YPAD. We analyzed five independent clones for each mutant strain and three for each wild type, respectively.

A yeast centromeric plasmid, pSC16 [38] (kindly provided by S. Cervantes), and a non-centromeric plasmid (pVT-U) containing the ADH1 promoter and termination expression cassettes were used to clone human ORMDL3 ORF. Each construct (100 ng) was used to transform yeast W303 Δorm1 Δorm2 strain by the lithium acetate method. Transformed yeast cells were recovered by selection on minimal synthetic dropout (SD) medium without leucine (centromeric plasmid) or without uracil (non-centromeric plasmid). Transformants were grown on liquid YPAD, sequentially diluted and dropout-plated onto the corresponding test media (0.25 mM and 0.4 mM HgCl₂; 10 mM and 15 mM DTT; 0.5 μg/ml and 1 μg/ml tunicamycin, all in YPAD).