UBC Faculty Research and Publications

A yeast phenomic model for the gene interaction network modulating CFTR-ΔF508 protein biogenesis Louie, Raymond J; Guo, Jingyu; Rodgers, John W; White, Rick; Shah, Najaf A; Pagant, Silvere; Kim, Peter; Livstone, Michael; Dolinski, Kara; McKinney, Brett A; Hong, Jeong; Sorscher, Eric J; Bryan, Jennifer; Miller, Elizabeth A; Hartman IV, John L Dec 27, 2012

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-13073_2012_Article_407.pdf [ 1.42MB ]
JSON: 52383-1.0220746.json
JSON-LD: 52383-1.0220746-ld.json
RDF/XML (Pretty): 52383-1.0220746-rdf.xml
RDF/JSON: 52383-1.0220746-rdf.json
Turtle: 52383-1.0220746-turtle.txt
N-Triples: 52383-1.0220746-rdf-ntriples.txt
Original Record: 52383-1.0220746-source.json
Full Text

Full Text

RESEARCH Open AccessA yeast phenomic model for the geneinteraction network modulating CFTR-ΔF508protein biogenesisRaymond J Louie3†, Jingyu Guo1,2†, John W Rodgers1, Rick White4, Najaf A Shah1, Silvere Pagant3, Peter Kim3,Michael Livstone5, Kara Dolinski5, Brett A McKinney6, Jeong Hong2, Eric J Sorscher2, Jennifer Bryan4,Elizabeth A Miller3* and John L Hartman IV1,2*AbstractBackground: The overall influence of gene interaction in human disease is unknown. In cystic fibrosis (CF) a singleallele of the cystic fibrosis transmembrane conductance regulator (CFTR-ΔF508) accounts for most of the disease. Incell models, CFTR-ΔF508 exhibits defective protein biogenesis and degradation rather than proper trafficking to theplasma membrane where CFTR normally functions. Numerous genes function in the biogenesis of CFTR andinfluence the fate of CFTR-ΔF508. However it is not known whether genetic variation in such genes contributes todisease severity in patients. Nor is there an easy way to study how numerous gene interactions involving CFTR-ΔFwould manifest phenotypically.Methods: To gain insight into the function and evolutionary conservation of a gene interaction network thatregulates biogenesis of a misfolded ABC transporter, we employed yeast genetics to develop a ‘phenomic’ model,in which the CFTR-ΔF508-equivalent residue of a yeast homolog is mutated (Yor1-ΔF670), and where the genomeis scanned quantitatively for interaction. We first confirmed that Yor1-ΔF undergoes protein misfolding and hasreduced half-life, analogous to CFTR-ΔF. Gene interaction was then assessed quantitatively by growth curves forapproximately 5,000 double mutants, based on alteration in the dose response to growth inhibition by oligomycin,a toxin extruded from the cell at the plasma membrane by Yor1.Results: From a comparative genomic perspective, yeast gene interactions influencing Yor1-ΔF biogenesis wererepresentative of human homologs previously found to modulate processing of CFTR-ΔF in mammalian cells.Additional evolutionarily conserved pathways were implicated by the study, and a ΔF-specific pro-biogenesisfunction of the recently discovered ER membrane complex (EMC) was evident from the yeast screen. This novelfunction was validated biochemically by siRNA of an EMC ortholog in a human cell line expressing CFTR-ΔF508.The precision and accuracy of quantitative high throughput cell array phenotyping (Q-HTCP), which captures tensof thousands of growth curves simultaneously, provided powerful resolution to measure gene interaction on aphenomic scale, based on discrete cell proliferation parameters.Conclusion: We propose phenomic analysis of Yor1-ΔF as a model for investigating gene interaction networks thatcan modulate cystic fibrosis disease severity. Although the clinical relevance of the Yor1-ΔF gene interactionnetwork for cystic fibrosis remains to be defined, the model appears to be informative with respect to human cellmodels of CFTR-ΔF. Moreover, the general strategy of yeast phenomics can be employed in a systematic manner* Correspondence: em2282@columbia.edu; jhartman@uab.edu† Contributed equally1Department of Genetics, University of Alabama at Birmingham, 730 HughKaul Human Genetics Building, 720 20th Street South, Birmingham, AL 35294USA3Department of Biology, Columbia University, 1212 Amsterdam Ave. MC2456,New York, NY 10027 USAFull list of author information is available at the end of the articleLouie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103© 2013 Mckinney et al.; licensee BioMed Central Ltd. This is an open access article distributed under the terms of the CreativeCommons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly cited.to model gene interaction for other diseases relating to pathologies that result from protein misfolding orpotentially any disease involving evolutionarily conserved genetic pathways.Keywords: Gene interaction, Genetic buffering, Genotype-phenotype complexity, Phenomics, Quantitative highthroughput cell array phenotyping (Q-HTCP), Cystic fibrosis transmembrane conductance regulator (CFTR), ER mem-brane complex (EMC), ATP binding cassette (ABC) transporter, Membrane protein biogenesis, Yeast model ofhuman disease, Comparative functional genomicsBackgroundSince release of the human genome sequence, genome-wide association studies (GWAS) and other advances ingenomic technology have challenged simplistic notions ofthe genetic basis of human disease. Even Mendelian dis-ease phenotypes are now thought to be driven by complexgenetic relationships [1]. For example, modifier genes caninfluence the severity of cystic fibrosis [2]. However, theinfluence on disease contributed by multi-locus, combina-tion-specific pairs of allelic variants remains largelyunmapped and uncharacterized biologically. Moreover,most disease traits are non-Mendelian (that is, ‘complex’traits), where expression of the phenotype involves multi-ple different gene activities, none of which is individuallyrequired or accounts for a large fraction of heritability[3,4]. Thus Mendelian and complex traits can be seen asdifferent ends of the same continuum in which multiplegenetic and environmental effects impact disease risk and/or severity in a combination-dependent manner. It is pre-sumed that in some genetic or environmental contextsparticular variant alleles are phenotypically expressed, andin other contexts they are buffered. However, whetherprinciples for disease variation can be deduced throughsystematic analysis of gene-gene interaction remainsunknown [5]. In this study we developed a yeast model ofgene interaction for a clinically relevant disease mutation,CFTR-ΔF508, to investigate whether it can potentiallyserve as a useful tool to better understand the geneticcomplexity underlying the human disease, cystic fibrosis[6]. Saccharomyces cerevisiae is a workhorse for funda-mental biology, but the extent to which experimentalmodels of gene-gene interaction employing an endogen-ous yeast cellular context could provide disease-relevantinsight via gene homology is unknown [5]. To investigatethis question, we applied the Q-HTCP method to systema-tically query the yeast genome for modifiers of a specificphenotype resulting from Yor1-ΔF670, and provide evi-dence validating this yeast phenomic (genome-wide analy-sis of phenotypic modification due to gene interaction)model for CFTR-ΔF508, the most prevalent human allelecausing cystic fibrosis [7].To model the evolutionarily conserved network of geneinteraction involving CFTR-ΔF508, we introduced thehomologous yeast ABC transporter, Yor1-ΔF670 [8,9],into the library of non-essential yeast gene deletionstrains [10-12], and used Q-HTCP [13,14] to measurethe influence of gene-gene interactions on cell prolifera-tion in the presence of oligomycin, a toxin extruded fromcells by Yor1. From a drug discovery perspective, proteinregulators of CFTR-ΔF biogenesis represent novel tar-gets, and cell culture experiments indicate such targetsare numerous [15,16]. Many of these regulators are evo-lutionarily conserved, thus a quantitative systems levelmodel of a gene interaction network model derived fromyeast could complement human and animal studies [17].From a systems biology perspective, the quantitativedescription of a gene network that modulates biogenesisof a misfolded ABC transporter could provide usefulinsight for understanding the phenotypic complexity ofcystic fibrosis in association with human genetic data,and might similarly aid study of other diseases related toprotein misfolding. If successful for cystic fibrosis, thesame general strategy of yeast phenomic modeling shouldbe applicable to derive understanding about disease com-plexity involving any conserved cellular pathway.MethodsYeast strainsDeletion mutants were from the MATa collection, cre-ated by the Saccharomyces Genome Deletion Project[18], and obtained from Open Biosystems. The querystrain background for double mutant construction was15578-1.2b [10]. The R1116T mutation (Figure 1) wasintroduced into pSM2056 (yor1-ΔF670-HA-GFP::URA3integrating vector) [19] by Quik Change mutagenesis(Stratagene) to create plasmid pRL026. This vector wasused as a template to amplify a PCR fragment corre-sponding to yor1-ΔF670/R1116T-HA-GFP-3’UTR whichwas combined with another PCR fragment encoding theNATMX cassette flanked by further YOR1 3’UTRsequence by splice overlap PCR. The full product (yor1-ΔF670/R1116T-HA-GFP-3’UTR-NATMX-3’UTR) wastransformed into yeast and selected for on media con-taining nourseothricin (’ClonNat’, Werner BioAgents);the presence of the genomic ΔF670/R1116T mutationwas confirmed by sequence analysis, creating strain RL4.The endogenous YOR1 promoter was replaced with aTet-OFF regulatable element by insertion of pJH023, aspreviously described [20], at the YOR1 locus to createstrain RL8. RL8 was mated to the MATa deletion strainLouie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103Page 2 of 20BCkDa18011582644937260 5 10 20Yor1 Yor1-)FYor1-)F/R1116T0 5 10 200 5 10 20Trypsin(Rg/mL)T + -F-Sec22F-HAT + - T + -Yor1 Yor1-)FYor1-)F/R1116TEpRS316YOR1yor1-)F670yor1-)F670/R1116TYPEG + 0.2Rg/ml oligomycinAF-HAYor1chase time (h): 0 1 2 3 0 1 2 3 0 1 2 3Yor1-)FYor1-)F/R1116TYor1Yor1-)FYor1-)F/R1116T0501000 1 2 3Chase time (h)Yor1 remaining(% total at t=0)DFkDa170-130-X-linker:NativeX-linkedAgg.Yor1 Yor1-)F Yor1-)F/R1116T0 10 20 30 400. efflux(% total at t=0)Time (min)Yor1-R1116TYor1Yor1-)FYor1-)F/R1116TvectorFigure 1 Characterization of Yor1-ΔF/R1116T. (A) The initial characterization of YOR1 alleles was performed using plasmid-based mutagenesis. Ayor1 null strain, yor1-Δ0, was transformed with a plasmid control (pRS316), or plasmids expressing YOR1, yor1-ΔF670, or yor1-ΔF670/R1116T as indicated,and the strains were serially diluted and spotted onto YPEG media with and without 0.2 μg/mL oligomycin. The yor1-ΔF670 mutation was associatedwith a trafficking and pump defect that rendered it phenotypically equivalent to yor1-Δ0. However, an additional intragenic mutation, Yor1-ΔF670-R1116T, exhibited an intermediate phenotype. (B) Capture of Yor1 into ER-derived transport vesicles was measured using an in vitro vesicle buddingassay that quantifies uptake of newly synthesized cargo proteins from radiolabeled permeabilized cells after addition of purified COPII proteins in thepresence (+) and absence (-) of GTP. Total membranes (T) were separated from the liberated vesicles by differential centrifugation. Packaging of Yor1into the vesicle fraction was monitored by immunoprecipitation; Sec22 is a control cargo protein that demonstrates efficient vesicle production evenin the absence of packaging of the mutant forms of Yor1. Neither Yor1-ΔF nor Yor1-ΔF/R1116T were captured into COPII vesicles whereas wild-typeYor1 was packaged normally. (C) Trypsin sensitivity of Yor1 was assessed by limited proteolysis of microsomal membranes expressing wild type andmutant forms of Yor1. Increasing concentrations of trypsin were added as indicated prior to processing of membranes for immunoblot analysis. Wild-type Yor1 is cleaved to several stable bands whereas both Yor1-ΔF and Yor1-ΔF/R1116T were significantly more susceptible to proteolytic attack. (D)Cross-linking between transmembrane domains of Yor1 was measured following introduction of paired cysteine substitutions into wild-type andmutant Yor1 as indicated. Addition of increasing concentrations of cross-linker resulted in the accumulation of wild-type Yor1 in a cross-linked specieswith distinct gel mobility. Cross-linking of Yor1-ΔF and Yor1-ΔF/R1116T resulted in the disappearance of the non-cross-linked species and theappearance of high molecular weight aggregates, suggesting abnormal assembly of transmembrane domains in these mutants. (E) Yor1 stability wasmonitored by pulse-chase analysis. Cells expressing wild-type or mutant forms of Yor1 were radiolabeled for 10 min, and then chased for 180 min withnon-radioactive amino acids. The amount of Yor1 present at each time point was determined by immunoprecipitation and autoradiography (toppanel). The percentage of Yor1 remaining was calculated relative to the starting material at t = 0 (bottom panel). (F) Yor1 function was probed using arhodamine-pumping assay. Yor1-Δ0/Pdr5-Δ0 cells carrying the indicated Yor1 alleles on a plasmid were loaded with the fluorescent dye, rhodamine,and the amount of fluorescence released over time into the culture supernatant was measured. The Yor1-ΔF670-R1116T mutation was associated withrhodamine extrusion intermediate between that of the wild-type Yor1 allele and either the Yor1-Δ0 or Yor1-ΔF670 mutant forms (which werefunctionally equivalent in this assay, as in the oligomycin resistance growth phenotype assay).Louie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103Page 3 of 20collection and double mutants selected by the syntheticgenetic array (SGA) method [11].Yeast mediaFor SGA [11], media was prepared with the followingmodifications. Mating was carried out in YPD liquid fol-lowed by diploid selection in YPD containing G418 andClonNat, and a second round of diploid selection sub-stituting Pre-Spo media 5 for YPD as described [21].Cultures were sporulated at room temperature for 1 week,before two rounds of transfer to haploid double mutantselection media [11]. For Q-HTCP, YPEG media (10 g/Lyeast extract, 20 g/L peptone, 3% ETOH, 3% glycerol, and1.5% agar) was used with 2 ng/mL doxycycline and con-centrations of oligomycin ranged from 0.05 to 0.25 ug/mLfor yor1-ΔF strains, and 0.05 to 0.35 ug/mL for YOR1strains. Doxycycline was used at 2 ng/mL to optimize theexpression level of Yor1-ΔF for phenotypic screening todetect enhancers and suppressors at the indicated concen-trations of oligomycin.Cell proliferation measurements and quantification ofgene interactionCells were inoculated from glycerol stocks in a 384 wellformat and grown for 36 to 48 hours in YPD with G418(200 ug/mL) and ClonNat (100 ug/mL), and withoutdoxycycline. Overnight-grown cell arrays were spottedto agar plates using a 384-pin tool (FP6 pins from V-PScientific) after first transferring to a ‘dilution plate’ toreduce the number of cells transferred, as described pre-viously [13]. Quantitative high throughput cell arrayphenotyping was used to obtain growth parameters bytime lapse imaging of cell arrays and fitting to a logisticgrowth equation (Figure 2B), as described previously[13,14]. The parameter L, which is equivalent to thetime at which half the final carrying capacity is reached,was used to quantify interactions (Figure 2C). Thegrowth curve parameters obtained from the fitted curvesare provided in Additional File 1. Interactions werequantified on the basis of a change in the response tooligomycin attributable to a gene deletion (Figure 2D),where interaction strength is a function of oligomycinresponse as determined by departure of the L value fora given double mutant strain vs. the Yor1-ΔF singlemutant across all oligomycin concentrations. To com-pute the interaction strength, the following algorithmwas used to determine the difference between each dou-ble mutant and the yor1-ΔF670 single mutant:Yi = Observed growth parameter for the knockout atdose i (Di)Ki = the effect of the knockout and its interaction withyor1-ΔF at a dose of oligomycinK0 = the effect of knockout when no oligomycin ispresent (D0)Li = the interaction effect of a knockout with yor1-ΔFat each dose of oligomycin1. Compute the average value of the 768 referencecultures at (Di): RDi,To simplify visualization of the interaction graphically,2. Remove the dose effect to oligomycin on the refer-ence: Ki = Yi - RDi3. Remove the effect of knockout (K0) when no oligo-mycin is present (D0): Li = Ki - K0Therefore L0 = 0 by definition.4. Fit a quadratic curve: Li = A + B*Di + C*Di25. Compute the interaction value at the max dose:Li-max= INT = A + B*Dmax + C*Dmax2Positive interaction values, termed ‘deletion enhan-cers’, denote increasing L and thus indicate exacerbationof the growth delay induced by oligomycin. For deletionstrains failing to grow at the higher concentrations ofoligomycin, interactions were ranked in tiers, with thestrains failing to grow at a greater number of concentra-tions grouped as stronger deletion enhancers (Addi-tional File 1). Conversely, strains that grew faster(shorter time to reach L) had negative interaction valuesand we refer to loss of the gene having a ‘deletion sup-pressor’ effect on the oligomycin sensitivity phenotype.Interaction plots for each gene deletion strain in boththe context of wild-type YOR1 and yor1-ΔF670/R1116Texpression are given in Additional Files 2 and 3. Thegraphs are ranked by the interaction strength of theyor1-ΔF670/R1116T allele. To help further partition thelist of genes influencing the yor1-ΔF/R1116T phenotype,gene-drug interaction data were incorporated with theprimary screen data for clustering (described below). Forgene-drug interactions, the number of concentrations ofeach drug tested was too few to fit a quadratic, thuseach perturbation was considered separately and interac-tions were quantified as the difference between the dele-tion and the wild-type reference strains and plotted afteradjusting for the dose effect of oligomycin and the effectof the deletion on growth in the control media. The inter-action data submitted to BioGRID [22] for inclusion in theBioGRID database and SGD [23] are indicated in Addi-tional File 5 in column L of the worksheet ‘REMc_dataand clustering’.Recursive expectation-maximization clustering (REMc)Interaction values selected for clustering represented theunion of genes from the yor1-ΔF670/R1116T screen withinteraction values >10 or <-16 and the screen with wild-type YOR1 in the same background with interaction values>10 or <-12. These thresholds were chosen to representthe tails of the distributions of interaction strength.Among deletion strains not growing at one or more con-centrations of oligomycin, higher interaction values wereassigned for cultures that failed to grow at lower concen-trations (see Additional File 5). Gene-drug interaction datawere incorporated to create profiles for genes selectedfrom the primary screen, as previously described [13].Louie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103Page 4 of 20ABChlj1-∆0sop4-∆0Time (h)0 20 40Doligomycin (μg/ml)deletion suppressordeletion enhancerE-40 20 40-60-40-2024060-2y=0.69x-0.118R2=0.423ORF deletionoverlapping genomic region60 80 100 120 140WT/Ref18.520.416.355.457.350.366.768.574.776.682.7 90.5 106.484.5 92.3 103.8 126.288.1 99.6 104.0 117.4 122.0 70 90 110 −20 −10 10 200hlj1-∆0sop4-∆0hlj1-∆0 sop4-∆0Relative LL (h)G(t)Time (h)50 100 1501001021010KrL0.25 μg/ml oligomycin0.15 μg/ml oligomycin0 μg/ml oligomycinG(t)Time (h)50 100 1501001020101100102101G(t)interaction scoreFoligomycin (μg/ml). 5 10 15 0-5 5 10 15 0-5 5 10 15 0-5 5 10 15L K R AUCRelative L Relative K Relative R Relative AUCFigure 2 A genome-wide screen for Yor1-ΔF gene interaction. (A) Time-lapse imaging was used to measure the growth phenotypes ofhaploid double mutants. Shown are example spot cultures (time indicated below each image) for strains with deletion suppressor and deletionenhancer effects on oligomycin sensitivity, along with a yor1-ΔF single mutant control, grown on media containing 0.2 μg/mL oligomycin. (B) Toquantify phenotypes, spot culture image series were analyzed for pixel density and fit to a logistic growth equation [14]. See Materials andMethods and Additional File 1 - Discussion B for further details. (C) Multiple concentrations of oligomycin were used to assess the interactionstrength for each gene deletion, using the growth parameter, L, corresponding to the time at which a culture reaches its half maximal density,K (r denotes the maximum specific rate). Gene deletion suppressor effects (interactions reducing L) are highlighted in green, whereas genedeletion enhancer effects (interactions increasing L) are indicated by blue. The three panels contain growth curves for the deletion strains shownin panel A at different oligomycin concentrations (0, 0.15, and 0.25 μg/mL). G(t) is the logistic fit for the data obtained for each culture timeseries; raw values for culture growth are indicated by black circles (WT/Ref), green squares (hlj1-Δ0 strain), and blue triangles (sop4-Δ0 strain).(D) Gene interaction is shown for hlj1-Δ0 (green squares) and sop4-Δ0 (blue triangles). Divergence of L for the double mutants is displayed as afunction of oligomycin concentration, compared to the phenotypic distribution of replicates of the yor1-ΔF670/R1116T single mutant (graydiamonds represent the distribution of central 95% of L values for 768 single mutant replicates). The data for the double mutants were shiftedby their difference with the single mutant (median response) at the zero oligomycin concentration (filled symbols), correcting for growthdifferences not attributable to oligomycin response. To quantify interactions, the data for each deletion mutant were first fit to a quadraticequation, and then the difference between the deletion mutant and the reference median was taken at an indicated concentration ofoligomycin. To highlight the interactions, the raw data (left panel) were transformed to remove the oligomycin dose effect (right panel). (E) Ascatter plot of interaction scores for pairs of gene deletion strains with overlapping open reading frames (obtained at oligomycin = 0.25 μg/mL).The open reading frames with a greater degree functional annotation in SGD were designated as the ‘ORFs’, and those with less functionalannotation designated ‘overlapping genomic regions’ [78]. (F) The affect of oligomycin dose on the growth curve parameters (left to right), L(time to half carrying capacity, K (carrying capacity), and r (maximum specific rate), and the area under the curve (’A’), for 384 replicates of theyor1-ΔF single mutant strain. Each diamond represents the central 95% of the standardized data for that oligomycin dose. The data for eachparameter at each dose was standardized (arbitrary units) by subtracting the mean and dividing by the standard deviation of the group nottreated with oligomycin. The oligomycin = 0 group is centered at 0. Using standard units for the data allows the dose trend between the panelsto be directly compared. The oligomycin dose effect is greatest for L, followed by AUC, and with minor effects on K and r.Louie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103Page 5 of 20REMc was used to identify groups of genes having similarinteraction profiles [24]. To obtain a dendrogram andfiner grain view of each REMc cluster, hierarchical cluster-ing using Euclidian distance and complete linkage wasperformed using Matlab. For all heat maps, the order ofthe perturbations is the same and labels indicate the inter-action values from: (A) the yor1-ΔF670/R1116T/gene dele-tion double mutants; (B) the screen of single-mutant(wild-type YOR1 background) gene deletion strains; (C)the growth defect of the deletion strain in Cold SpringHarbor SC media [25]; gene-drug interactions on the fol-lowing media (D) SC media lacking threonine (usingmedia in (C) as the reference); (E) SC media lacking threo-nine and with 80 ug/mL beta-chloro-alanine (using mediain (D) as the reference); SC media supplemented with (F)0.7 nM rapamycin; (G) 1.4 nM rapamycin; (H) 1 nM FK-506; (I) 0.7 nM rapamycin and 1 nM FK-506; (J) 50 mMhydroxyurea; (K) 125 mM hydroxyurea; (L) 75 ng/mLcycloheximide; (M) 125 ng/mL cycloheximide; (N)150 nM miconazole; or (O) 225 nM miconazole (see Addi-tional File 5).Gene homology mappingThe Princeton Protein Orthology Database [26] was usedto identify yeast-human homologs for relating the resultsof our yeast screen to the larger literature of CFTR-ΔF508protein biogenesis factors [27]. In cases where homologywas not one-to-one, the best functional matches were dis-cussed [28]. For example, human isoforms of HSP90(HSP90A and HSP90B) have opposite effects on CFTR-ΔF508 biogenesis when knocked down by siRNA [16],thus deletion of yeast HSP82, an HSP90 family member inyeast that acts as a deletion suppressor, mimics only theeffect of siRNA knockdown of HSP90A. As another exam-ple, yeast HLJ1 and three different homologous humanproteins (CSP, DNAJB12, and DNJB2) exert comparableeffects on Yor1-ΔF and CFTR-ΔF biogenesis, respectively(see Additional File 1 - Discussion C).Biochemical analysis of Yor1-ΔF670 andYor1-ΔF670/R1116TIn-vitro uptake of Yor1 into COPII vesicles was per-formed from radiolabeled semi-intact cells, and limitedproteolysis, chemical cross-linking, and in-vivo pulse-chase experiments were all performed as described [29].Rhodamine efflux assayyor1-Δ0/pdr5-Δ0 double mutant strains bearing plasmidsexpressing YOR1 variant alleles (as indicated in Figure 1)were grown to mid-log phase (OD600 of approximately0.5) in SD-ura medium (0.67% yeast nitrogen base, 20%glucose, -ura dropout mix). Cells equivalent to fiftyOD600 units were harvested, washed with 50 mM HEPESpH 7.0, and loaded with rhodamine B (Sigma-Aldrich) byincubating cells in 5 mL of 50 mM HEPES, pH 7.0,5 mM 2-deoxyglucose, and 100 μg/mL rhodamine B for2 h at 30ºC. Cells were washed and resuspended in 5 mLof 50 mM HEPES, pH 7.0, supplemented with 10 mM D-glucose (Sigma-Aldrich). Every 2 min, 500 μL aliquots ofcell suspension were removed, cells collected by centrifu-gation, and the rhodamine-containing supernatant wasremoved and quantified by measuring absorbance atOD555.siRNA experimentsFor TTC35 mRNA knockdown experiments, HeLa cells(CCL2, ATCC) were transfected with pcDNA-CFTR-ΔF508 plasmids using TransIT-HeLaMONSTER® trans-fection reagent (Mirus Bio, Madison, WI, USA) perinstruction manual. Cells were split into a 12-well plateand the next day transfected with TTC35 specific siRNA(sc-77588, Santa Cruz Biotechnology, Santa Cruz, CA,USA) at 10 or 25 nM, using RNAiMAX (Invitrogen). As anegative control siRNA, Stealth RNAi™ siRNA negativecontrol lo GC (Invitrogen, 12935-200) was used at 25 nMfinal concentration. The next day, cells were moved to27°C and incubated for an additional 72 h before harvest.For western blot analysis, cells were lysed in RIPA contain-ing Halt protease inhibitor cocktail (Thermo-Pierce), andthen analyzed on 4% to 20% gradient SDS-PAGE (Invitro-gen). After blotting onto a PVDF membrane, the blot wascut laterally into three pieces at 75kD and 35kD markers.The top piece (>75kD) was developed for CFTR protein(150 to 180 kD) using 3G11 rat monoclonal antibody [30], the middle piece (between 75kD and 35kD) was probedfor a-tubulin (approximately 55kD) as an internal control(DM1A antibody, GeneTex), and the bottom piece(<35kD) was probed with TTC35 antibody (sc-166011,Santa Cruz Biotechnology). Blots were developed usingSuperSignal West Pico Chemiluminescent substrate(Thermo-Pierce), and exposed to Kodak BioMax MR film.Densitometry was performed using ChemiDoc XRS andImage Lab software (BioRad).ResultsYor1-ΔF and CFTR-ΔF are membrane proteins withshared biogenesis defectsYor1 is a close homolog of CFTR in the ATP-bindingcassette family of membrane transporters that includespleiotropic drug transporters [31], and it is the primarydeterminant of oligomycin resistance due its plasmamembrane-localized function in extruding oligomycinfrom the cell [32]. Analogous to CFTR-ΔF508, mutationof the highly conserved phenylalanine residue in thefirst nucleotide binding domain, Yor1-ΔF670, results inER-retention and degradation by proteolysis, yielding anoligomycin-sensitive phenotype [13]. However, unlikeCFTR-ΔF508, Yor1-ΔF670 appears not to retain residualmembrane transport function [8]. Therefore, we per-formed an intragenic suppressor screen and identified asecond site mutation (R1116T) that restored partial pumpfunction (Figure 1 and Additional File 1 - Discussion A).Louie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103Page 6 of 20The oligomycin growth phenotype associated with Yor1-ΔF670-R1116T was intermediate between that of Yor1-ΔF670 (which was indistinguishable from the yor1-Δ0deletion mutant) and wild-type Yor1 (Figure 1A). Theintracellular fate of the partially functional R1116T mutantwas identical to that of the original Yor1-ΔF mutant: theprotein was less efficiently packaged into transport vesiclesreconstituted in vitro (Figure 1B), Yor1-ΔF670-R1116Twas misfolded, as detected by limited proteolysis (Figure1C) and intramolecular cross-linking (Figure 1D), andturnover was indistinguishable from Yor1-ΔF670 by pulse-chase analysis (Figure 1E). We assessed the effect of theR1116T mutation on pump function using a rhodamineexclusion assay, which revealed partial rescue of Yor1-ΔF670-R1116T relative to Yor1-ΔF670 (Figure 1F).Although we do not know the precise mechanism bywhich the R1116T mutation impacts the activity of Yor1-ΔF, the aggregate of our evidence suggests that it is adominant gain-of-function mutation that confers addi-tional drug-pumping activity (see Additional File 1 - Dis-cussion A and Additional file 1, Figure S1 for furtherdescription and characterization of the R1116T mutation).The molecular characteristics and intermediate oligomycinresistance conferred by Yor1-ΔF670-R1116T (referred tohere forward as ‘Yor1-ΔF’) resemble the defects of CFTR-ΔF508, and thus provided a model to screen the yeast gen-ome for canonical protein regulators of ‘ΔF-associated’biogenesis by introducing yor1-ΔF into the yeast genedeletion strain collection [10,11].Measurement of gene interaction strength from growthcurvesFor quantitative phenotypic analysis of the genomic col-lection of deletion strains, we used growth curve analysisat multiple concentrations of oligomycin, and examinedthe entire library alternatively in the context of expres-sion of Yor1-ΔF or Yor1 wild-type protein. The phe-nomic method of time series analysis of cell arrayimages (Figure 2A) provides growth curves on a geno-mic scale for measuring strength of gene interaction[13]. The kinetic analysis is based on density of eachspot culture over time [13,33], in contrast to qualitativemethods or quantitative strategies that employ singletime points of culture area [34,35]. Q-HTCP, by virtueof imaging cultures arrayed on agar rather than measur-ing optical density of liquid cultures in multi-well plates,provides orders of magnitude greater throughput, withspot density time series for each strain (Figure 2A) thatfit to a logistic growth equation (Figure 2B) [14]. Weused a parameter from the curve fitting to quantify eachgene interaction by comparing growth inhibitionbetween the Yor1-ΔF single mutant and each respectivedouble mutant across multiple oligomycin concentra-tions (Figure 2C).In this study, we focused on a specific parameter oflogistic growth, termed L, which represents the time ittakes a culture to reach half its final density, K [14] (Addi-tional File 1 - Discussion B). Thus, the L parameter isinversely proportional to fitness, such that double mutantstrains exhibiting a shorter L relative to the yor1-ΔF singlemutant (that is, deletion suppressors of the oligomycinsensitivity phenotype) indicate genes that (when present)function to prohibit biogenesis of misfolded Yor1-ΔF.Conversely, gene interactions resulting in a longer L (thatis, deletion enhancers) correspond to candidates that nor-mally promote Yor1-ΔF biogenesis (Figure 2C). The nullhypothesis for gene interaction [36] was defined by a neu-trality function consisting of the median L value fromreplicate cultures of the Yor1-ΔF single mutant acrossincreasing oligomycin concentration, to account for thedrug effect. In addition, to account for the gene deletioneffect on growth (independent of oligomycin treatment)the L value of each double mutant culture was adjusted(for every oligomycin dose) by the constant differencebetween it and the Yor1-ΔF reference mutant median atthe zero-oligomycin concentration (Figure 2D, left panel).Next, a quadratic equation was fit to the L-value differ-ences for each double mutant over all oligomycin concen-trations. The difference between this quadratic fit and thereference median at the highest concentration of oligomy-cin having measurable growth was defined as the interac-tion score (enhancing interactions were further rankedaccording to the number of oligomycin concentrationswhere growth was completely inhibited). To more clearlyvisualize only the interactions, the data were transformedto remove the dose effect of oligomycin on the yor1-ΔFsingle mutant cultures (Figure 2D, right panel).Our screen, by virtue of incorporating multiple con-centrations of oligomycin and examining the trend ofresponse, contains an intrinsic form of replication. Theconsistent trends of phenotypic response observedserves as evidence of technical reproducibility in thephenotypic analysis. We also repeated the entire screenat all concentrations, which again indicated high repro-ducibility (Additional file 1, Figure S2).Reproducibility of the gene interaction measurements wasfurther evidenced by positive correlation between valuesobtained for deletion strains that shared chromosomalstrand overlap in their open reading frames (Figure 2E). Toassess this type of correlation, each overlapping ORF pairmember was assigned to one of two groups according to itbeing the ‘better’ or ‘less well’ annotated gene/orf. Less well-annotated orfs would, for example, include computationallydetermined chromosomal regions that were systematicallyknocked out by the Yeast Gene Deletion Consortium, butdo not necessarily encode expressed genes [37]. Strongerinteractions tended to correlate with the extent of geneannotation, perhaps due to residual functional activity inLouie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103Page 7 of 20the non-overlapping regions of the better annotatedgenes that were not deleted by removal of overlappingORFs (Figure 2E). The phenotypic parameter, L, which weused in this study to quantify interactions was more sensi-tive to detect the growth inhibitory effect of oligomycin(Figure 2F). This is the first study we are aware of demon-strating the utility of genome-scale growth curve acquisitionand use of the L parameter for quantitative assessment ofgene interaction in phenomic analysis.Detection of molecular mechanisms associated with weakgene interactionWe found yor1-ΔF gene interaction to occur abundantly,across the genome and with wide-ranging strengths ofeffect. To help clarify the many interactions, we performeda similar analysis of oligomycin growth inhibition in thegene deletion strain collection endogenously expressingwild-type YOR1 (Figure 3). The comparison of Yor1 andYor1-ΔF candidate regulators was focused on four generalclasses: those that impact (positively or negatively) onlyYor1-ΔF, and those that impact (positively or negatively)both wild-type and the misfolded form of Yor1. Each classof mutant holds potential for uncovering novel mechanis-tic insight into biogenesis of topologically complex mem-brane proteins. Yor1-ΔF-specific interaction suggestspathways that recognize the misfolded protein, whereasinteraction with both the misfolded and wild-type formsof the protein could represent either proteins that gener-ally influence ABC transporter biogenesis or genes thataffect oligomycin resistance independent of Yor1 function,such as pleiotropic drug resistance (PDR) genes or mito-chondrial components (Figure 3).Given the high sensitivity of the cell array method formeasuring gene interaction, we sought perspective as towhether weak interactions reflected effects on Yor1 bio-genesis that could be detected with molecular assays.Cue1 is an ER membrane protein that serves to recruit theubiquitin-conjugating enzyme, Ubc7, to the ER, where it isrequired for ubiquitination of misfolded proteins prior totheir disposal by proteasome-mediated ER-associateddegradation. Rpn4 is a transcription factor that activatesexpression of proteasome genes; the depletion of protea-some subunits in an rpn4-Δ0 null strain would beexpected to impair ER-associated degradation of misfoldedproteins, potentially increasing their biogenesis. Yor1-ΔF670 turnover has been previously reported as dimin-ished by mutation of UBC7/QRI8 [8,9]. Therefore, weexamined Yor1-ΔF670 stability in the functionally relatedcue1-Δ0 and rpn4-Δ0 mutants, which showed weak inter-action (Figure 4A, B). The half-life of Yor1-ΔF670 wasindeed prolonged in both the cue1-Δ0 and rpn4-Δ0 strainsrelative to wild type (Figure 4C, D). Thus, the screen wassensitive to genes affecting proteasome-mediated turnoverof Yor1-ΔF670, validating the yeast model with respect tothis aspect of CFTR-ΔF biology [38,39] and confirmingthe molecular basis of phenotypic effects revealed by thescreen.Genes interacting with Yor1-ΔF map to homologousregulators of CFTR-ΔF508An open question is the extent to which gene interac-tion is evolutionarily conserved, and thus the extent towhich simple genetic systems like yeast can reveal prin-ciples about gene interaction relevant to human disease[5]. A study comparing worms and yeast concludedgene interaction lacks conservation [40], whereas studiescomparing evolutionarily divergent yeast have foundthat substantial conservation exists [41,42]. However,previous studies were not designed to model a specificdisease-related mutation. Our data represented anopportunity to probe conservation of gene interactionwithin a clearly defined molecular and cellular context,namely biogenesis of homologous ABC proteins carryingmutation of a conserved disease-causing residue[5,13,14].To assess relevance of our dataset to CFTR-ΔF bio-genesis, we surveyed the literature for evidence of evolu-tionarily conserved cellular responses to the ‘ΔF-like’folding defect. The (P-POD) [26] was used to identifyhomologous genes [27], yielding many examples of func-tional concordance between biogenesis factors for Yor1-ΔF and those known for CFTR-ΔF (Figure 5A). MostCFTR-ΔF protein regulators have been characterizedusing RNAi methods aimed at identifying targets forincreasing CFTR-ΔF processing by small molecule inhi-bitors [15,16]. Accordingly, the majority of homologousyeast gene deletions found to modulate Yor1-ΔF biogen-esis also functioned to enhance biogenesis. Broadlydefined functional categories highlighted the sharedfates of Yor1-ΔF and CFTR-ΔF, falling into at leastthree classes including (Figure 5A and Additional File 1- Discussion C): Syntaxins, which mediate vesicle fusionwithin the secretory pathway and may also regulateCFTR channel activity more directly [43-46]; Rab pro-teins, which regulate vesicular trafficking of CFTR-ΔFand other plasma membrane proteins [47,48]; and ERquality control machineries, a class of regulators of thatencompasses chaperones and other machineries that caninfluence folding and ER-associated degradation (ERAD)to govern the fate of misfolded proteins in the ER[16,49-52]. Each of these regulator classes exhibitedhomologous genes that encode regulators of CFTR-ΔFbiogenesis (see Additional File 1 - Discussion C forfurther explanation of homologies). Moreover, becausethe homologous regulators were not the strongest ineffect from the overall screen, additional conserved reg-ulators were likely identified (Figure 5B). Together,these results indicate evolutionary conservation of geneLouie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103Page 8 of 20interaction and suggest novel interactors from the Yor1-ΔF screen may represent as yet uncharacterized modi-fiers CFTR-ΔF biogenesis.Identification of functional gene modules by clusteringanalysisWe used REMc to search for functional gene modules[24]. Gene profiles selected for clustering had Yor1-ΔFinteraction scores >10 or <-16, or in the context ofwild-type Yor1 had gene-drug interaction >10 or <-12.We created interaction profiles for each gene by includ-ing additional gene-drug interaction data, and thenassessed modularity (similar influence on the phenotypeacross different perturbations) by clustering [13,53,48].The REMc algorithm objectively specified the numberof clusters and provided an indication of cluster quality[24]. We used the GOid_z method to quantify overallenrichment of gene ontology (GO) functional informa-tion within clusters [24]. GOTermFinder, was used toidentify specific terms associated with each cluster aswell as the representative genes [54]. All clusters wereenriched for functional information and many wereassociated with specific GO terms (Additional File 1 -Table S1 and Additional File 5). We also note that somefunctionally related genes appeared in different clusters,even though they exerted similar effects on Yor1-ΔF010-50 -40 -30 -2 20 30 40 50-50-40-302020304050Interaction score(YOR1)Interaction score(yor1-∆F)deletion enhancersdeletion suppressorsEMC1EMC2EMC3EMC4EMC5EMC6SOP4POR1CYC2PDR1PDR16CUE1RPN4ERV14Figure 3 Gene-gene interaction with Yor1 and Yor1-ΔF serves as a resource for identification of novel regulators of membraneprotein biogenesis. Interaction scores for individual yeast mutants in the context of either wild type Yor1 (x-axis) or Yor1-ΔF (y-axis) wereplotted to illustrate classes of protein biogenesis regulators suggested by the screen. Deletion enhancers (positive value indicates prolonged Land slower growth) and suppressors of oligomycin sensitivity represent factors predicted to promote or prohibit protein biogenesis, respectively.Interactions along the y-axis may indicate proteins that act more specifically on misfolded proteins, while those in the upper right and lower leftquadrants are considered to have more general effects on Yor1 protein biogenesis, or to affect the phenotype independently of Yor1/Yor1-ΔF.Examples of genes with functions suspected to directly influence oligomycin resistance, without necessarily acting through Yor1 biogenesis aredenoted by ‘x’. EMC members are colored in burgundy. CUE1, RPN4, and ERV14 are other genes that were further validated by molecular studies(see Figures 4, 6, and 7).Louie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103Page 9 of 20biogenesis (for example, the EMC genes describedfurther below). This suggests that though they cooperateto determine the fate of Yor1-ΔF, they can function differ-entially in other cellular contexts. Other explanations forthe appearance in different clusters of genes known to befunctionally related include over-estimation of the numberof clusters, measurement error, and the gene-specific func-tional relevance of particular gene interaction profilesselected for clustering.Validation of Erv14 as a cargo-specific sorting factorfor Yor1Cluster 2-0.1-1 contained genes previously shown tofunction cooperatively in protein transport through thesecretory pathway (Figure 6A). Namely, Sys1, Mak3, andMak10 cooperate in the recruitment of the ARF-likeGTPase, Arl3, to the Golgi to regulate vesicular trans-port [55]. Given the clustering of these gene interactionprofiles with Erv14, which can function as a cargo adap-tor for enrichment of newly synthesized proteins intoER-derived transport vesicles, we suspected they mayfunction in a common pathway, with Erv14 acting in anupstream compartment distinct from the others in thecluster (Figure 6B). Moreover, we observed oligomycinsensitivity to be more strongly dependent on ERV14 inthe context of wild-type Yor1 than Yor1-ΔF (Figure 6C).This difference raised the possibility that Erv14 pro-motes capture of wild-type Yor1 into ER-derived trans-port vesicles [56] more efficiently than it does for themisfolded Yor1-ΔF substrate. According to this hypoth-esis, reduced recognition of the misfolded Yor1-ΔF byErv14 would lessen the phenotypic impact of the erv14-Δ0 null allele on oligomycin sensitivity in an allele-specific manner (Figure 6C). We tested Erv14 functionby in vitro reconstitution of COPII vesicle formation[57], comparing capture of Yor1 in the presence orabsence of Erv14. Indeed, ERV14 deletion specificallyreduced capture of Yor1, leaving a control cargo, Sec22,unaffected (Figure 6D). Yor1-ΔF capture was weakregardless of ERV14 status. Thus the gene-drug interac-tion between erv14 and oligomycin can be explained bya physical interaction between Erv14 and Yor1 that pro-motes ER export. Accordingly, our vesicle budding assayrevealed no defects associated with capture into ER-derivedvesicles in sys1-Δ0, arl3-Δ0, mak3-Δ0, and mak10-Δ0strains (Figure 6E). Thus, taken together, the phenotypic-60-40-2002040600501000 1 2 3WTcue1-∆0rpn4-∆0Chase time (h)α-HAα-Gas1WT cue1-∆0 rpn4-∆0C DBnormalized Loligomycin (μg/ml)−60 −20 0 20 40 −50 0 500. −14.9/-16.5 int −5.2/-6.8 CUE1/RPN4YOR1 yor1-∆FYor1-∆F remaining(% total at t=0)AInteraction StrengthGenechase time (h): 0 1 2 3 0 1 2 3 0 1 2 3CUE1RPN4UBC7/QRI8deletion enhancersdeletion suppressorsFigure 4 Molecular validation of weak gene interactions. (A) Ranking of interaction strength shows the distribution of phenotypic influenceof deletion mutants. Some physiologically relevant hits (for example, UBC7/QRI8) fell below our clustering thresholds of >10 or <-16 (blue andgreen shading, respectively); the related components, CUE1 and RPN4, were on the cusp of our threshold but were still functionally relevant.(B) The cue1-Δ0 and rpn4-Δ0 strains showed deletion suppressor phenotypes specific for yor1-ΔF (at right), since the single mutant (that is, in thecontext of wild-type Yor1) did not affect oligomycin resistance. (C) By pulse-chase analysis, Yor1-ΔF turnover was reduced in the cue1-Δ0 andrpn4-Δ0 backgrounds relative to the wild-type background. Maturation of a control protein, Gas1, was unaffected. (D) Quantification of threereplicates of the experiment shown in Figure 2B; error bars represent standard deviation.Louie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103Page 10 of 20and molecular data suggest this cluster of genes functionsas a linear pathway to regulate progress of Yor1 out of theER and through the Golgi for delivery to the plasmamembrane.An ER membrane complex discovered in yeast promotesbiogenesis of CFTR-ΔFIn light of homologous genes exerting analogous influ-ences on Yor1-ΔF in yeast and CFTR-ΔF processing inSyntaxin Module:Yeast gene    Human homologSSO2      STX1aVAM7      STX8SNC1      VAMP8TLG2      STX16Rab Module:Yeast gene    Human homologVPS21      RAB5YPT7      RAB7MYO4      MYOSIN-5ER Quality Control Module:Yeast gene    Human homologSSA2      HSPA8HSP82      HSP90AHLJ1      DNAJC5/CSPHLJ1      DNAJB12TOM1      HACE1UBC13      UBC13VAM7CFTRCl-HSP82 HLJ1TLG2SNC1MYO4SSO2YPT7SSA2VPS21AoligomycinYor1pTOM1 UBC13BInteraction StrengthChromosome number1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16-40-200204060EMC1 EMC4 ERV14 EMC5 SOP4 EMC2 EMC3 EMC6 YDJ1SNC1 MYO4 UBC13 TOM1 VAM7 SSA2 YPT7 SSO2 HLJ1 TLG2 VPS21 HSP82deletion enhancersdeletion suppressorsFigure 5 Modulation of CFTR-ΔF and Yor1-ΔF biogenesis occurs via homologous gene interaction. (A) At left, gene interactions areillustrated by plotting the parameter L vs. oligomycin concentration (increasing upward: 0, 0.05, 0.1, 0.15, 0.2, and 0.25 μg/mL) for the doublemutant and the full distribution of 768 Yor1-ΔF single mutant cultures (gray diamonds). Leftward departure indicates faster growth andimproved biogenesis of Yor1-ΔF. At right, human genes found in the literature to modulate CFTR-ΔF biogenesis are paired with homologousyeast genes. Gene pairs are grouped into protein classes associated with discrete cellular functions (see text and Additional File 1 - Discussion C).(B) The strength of gene interaction is depicted with respect to chromosomal position and highlighted gene interactions indicate thosediscussed in the manuscript (interactions are calculated at an oligomycin concentration of 0.2 μg/mL). Vertical lines demarcate chromosomes.Louie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103Page 11 of 20human cells, respectively, we anticipated other yeastgene interactions identified by our screen would simi-larly represent homologs that function as conserved,uncharacterized CFTR-ΔF modulators. Our attentionwas drawn to a cluster (2-0.3-1) that contained EMC1,EMC3, and EMC5, three components of a recentlydescribed ER membrane complex [58]. Three additionalmembers of this complex, EMC2, EMC4, and SOP4grouped together in cluster 2-0.2-0 (Figure 7A). Allseven of the EMC members were deletion enhancerswith interaction specificity for the Yor1-ΔF mutant pro-tein (Figure 7B), and all had comparable strengths ofeffect, suggesting removal of any one of the genes dis-rupts a function common to all [13,35].The molecular function(s) of the EMC are only begin-ning to be characterized. Deletion of EMC3 (but not otherEMC members) activated an unfolded protein responseelement (UPRE)-GFP reporter in a genome-wide screen,which led to identification of the complex. However, theEMC effect on Yor1-ΔF biogenesis appeared to be inde-pendent of any association with induction of the UPR,because deletion of HAC1 or IRE1 (which blocks the UPR)exerted no effect on oligomycin resistance, and there wasvery weak association between the strength of UPR activa-tion and the influence of Yor1-ΔF biogenesis given thesame gene deletion [58,59] (Additional file 1, Figure S3).Alternatively, EMC components might directly promotefolding of Yor1-ΔF, such that loss of EMC function resultsDT + -α-Sec22α-HAT + -WT erv14-∆0024681012(percent of total)WT erv14-∆0C0. 16.5 int 27 ERV14YOR1 yor1-∆F−60 −20 0 20 40 −50 0 50normalized Loligomycin (μg/ml)T + - T + - T + - T + - T + -WT erv14−∆0 mak10−∆0 sys1−∆0 arl3−∆09 4 2 1 8 3 8 3.5 9 420 4 19.5 4 23 5 22 6 23 5α-Sec22pα-HAEBER GolgiErv14Sys1, Arl3, Mak3, Mak10ARIM21SUR2SYS1ARL3ERV14PKR1CYB5MAK3MAK10a b c ed f g h i j k l m n oFigure 6 ERV14 promotes capture of Yor1 into COPII vesicles. (A) The genetic interaction profile of the erv14-Δ0 strain clustered with thosefor the sys1-Δ0, arl3-Δ0, mak3-Δ0, and mak10-Δ0 strains. Columns represent diverse gene-drug interactions as described in methods. (B) Based onknown data [55], and supported by our phenotypic and molecular findings, Erv14, Sys1, Arl3, Mak3, and Mak10 appear to function in a pathwaywith Erv14 acting in the ER and Sys1, Arl3, Mak3, and Mak10 functioning in the Golgi. (C) Gene-oligomycin interactions for erv14-Δ0 strains, in thecontext of either wild-type Yor1 or Yor1-ΔF, suggested Erv14 promotes the biogenesis of both Yor1 and Yor1-ΔF. (D) Packaging of wild-typeYor1 into COPII vesicles was quantified using an in-vitro budding assay that measures capture of newly synthesized cargo proteins from radio-labeled permeabilized cells after addition of purified COPII proteins in the presence (’+’) or absence (’-’) of GTP [9], followed byimmunoprecipitation of the cargo protein of interest. ‘T’ indicates the total membrane pool of labeled Yor1-HA. Erv14-containing membranesshowed approximately four-fold more efficient capture into vesicles of HA-tagged Yor1 than erv14-Δ0 membranes. The defect showed specificity,since ERV14 deletion did not affect packaging of another cargo protein, Sec22. Quantification of three independent experiments is shown atright; error bars represent standard deviation. (E) Similar vesicle budding assays from mak10-Δ0, sys1-Δ0, and arl3-Δ0 membranes showed nodefects in Yor1 capture into COPII vesicles in these mutants, suggesting they function downstream of Erv14.Louie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103Page 12 of 20B0 20 40 60Chase time (min)WT WTYor1-∆F Yor1WTGas1020406080100Percent remainingWTsop4-∆ cpmat t=0Eα-CFTRα-TTC35α-tubulin25nM10nMno siRNAcontrolsiRNATTC35 siRNA0. TTC35 siRNAp=0.02CFTR-ΔFTTC35Relative amountnormalized Loligomycin (μg/ml)oligomycin (μg/ml)sop4-∆0 sop4-∆0 sop4-∆ 24.3/31.9/30.1 int 6.6/2.2/11.8 EMC1/EMC3/EMC5YOR1 yor1-∆F0.−60 −20 0 20 400.−50 0 50int 31.7/31.3/21.7 int 7/1/-2.6 EMC2/EMC4/SOP4YOR1 yor1-∆Fneed labels for conditionsC Dcluster 2-0.2-0WHI3EMC5PSY2SDS23PHO86YPR044CGIM3RPL21ARPN10EMC1 o/lPDR1EMC3YNR025CHAL5BFR1GCS1EMC1GEF1THP2AEMC4SOP4RPL43ASNF4CDA2YEL067CEMC2YJL200CPHC4YMR103CUBA3cluster 2-0.3-1a b c ed f g h i j k l m n oFigure 7 A conserved ER membrane complex discovered in yeast promotes CFTR-ΔF biogenesis. (A) Six of seven members of therecently described ER membrane complex (EMC) fell into two clusters (EMC genes are labeled in red; ‘EMC1 o/l’ indicates YCL046w, whichoverlaps EMC1/YCL045c). Columns represent different gene-drug interactions as described in methods. (B) The interaction plots for EMC genes(corresponding to genes in panel A) suggested the complex has a pro-biogenesis effect specific for Yor1-ΔF. The similarity in profiles isconsistent with the hypothesis that each EMC gene is required for the function of the complex in promoting Yor1-ΔF biogenesis. (C) By pulsechase analysis, deletion of SOP4 did not affect the half-life of Yor1-ΔF670. Quantification of three independent experiments is shown; error barsrepresent standard deviation. (D) Yor1-ΔF670 synthesis was reduced in the sop4-Δ0 strain based on the total amount of 35S-Met/Cys incorporatedduring the initial 10-min pulse. Three independent experiments were quantified; error bars represent standard deviation. Wild-type Yor1 and theGPI-anchored protein, Gas1, were unaffected by SOP4 deletion. (E) TTC35, the human homolog of EMC2, is required for normal biogenesis ofCFTR-ΔF. HeLa cells, transiently expressing CFTR-ΔF at 27°C, were co-transfected with control siRNA, 10 nM TTC35 siRNA, or 25 nM TTC35 siRNAas indicated. Protein levels of TTC35, CFTR-ΔF, and a-tubulin were monitored from whole cell lysates by western blot, and (at right) relativeabundance from three independent experiments was quantified by densitometry; error bars represent standard deviation, significance wasassessed by a paired two-tailed t-test.Louie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103Page 13 of 20in ER retention of Yor1-ΔF specifically, followed byERAD-mediated degradation, with reduced delivery and/or stability at the plasma membrane. However, pulse-chase analysis revealed that the Yor1-ΔF half-life was notaltered in EMC mutants (Figure 7C and data not shown).Instead, we observed that less Yor1-ΔF was synthesized inthe initial 10-min pulse-labeling period when SOP4, amember of the EMC, was deleted (Figure 7D). Reducedlabeling without increased degradation suggested a rolefor the complex in early stages of Yor1-ΔF biogenesis,such as during synthesis and translocation through theSec61 translocation pore. Interestingly, this pro-biogenesiseffect seemed specific to the misfolded protein, since theoligomycin phenotype associated with wild-type Yor1 wasunaffected by deletion of EMC genes (Figure 7B). Further-more, wild-type Yor1 and an unrelated plasma membraneprotein, Gas1, were synthesized normally in the sop4-Δ0mutant (Figure 7D).Potential relevance of the EMC components to CFTR-ΔFprocessing was suggested by CFTR protein-protein interac-tion data indicating the homolog of EMC2, TTC35, physi-cally associates with CFTR-ΔF but not wild-type CFTR(see supplemental data of reference [60]). However, at thetime of that study, the EMC complex had not been charac-terized and only one subunit of the complex was identifiedby the interactome study. In contrast, we determined thatall of the subunits give the same quantitative strength ofinteraction and cluster together in their phenotypic geneinteraction profiles across several chemical perturbations.Thus our screen data provided a potential link betweentwo high impact studies involving the CFTR interactomeand the identification of the novel EMC complex [58,60].To test for functional homology, CFTR-ΔF was monitoredby immunoblot in the context of a TTC35 knockdown bysiRNA. HeLa cells were transiently transfected with a plas-mid expressing CFTR-ΔF, co-transfected with TTC35siRNA or control siRNA, and shifted to 27°C. The shiftfrom 37°C to 27°C was to allow adequate rescue of CFTR-ΔF protein so that we could see the detrimental impact oflosing function of a presumed pro-biogenesis factor. Addi-tionally, keeping the cells at 37°C during the knockdown ofTTC35 provided elimination of CFTR-ΔF protein poolsprior to TTC35 knockdown and shift to conditions whereCFTR-ΔF biogenesis can occur. Under the experimentalconditions performed, knockdown of TTC35 reducedCFTR-ΔF expression by 30% to 50% (Figure 7E and Addi-tional File 1 - Additional file 1, Figure S4). Thus CFTR-ΔFprocessing is dependent upon expression of TTC35, vali-dating the prediction from the yeast data for EMC involve-ment in biogenesis of ΔF-misfolded ABC transporters.DiscussionAlthough it is well known that genes, proteins, andpathways are conserved across evolution, conservationof interactions between genetic pathways having thepotential to differentially regulate expression of pheno-types is only just beginning to be characterized in modelsystems [61,62]. Therefore, the clinical relevance of suchnetworks remains to be elucidated [5,63,64]. In thisregard, our data suggest the intriguing possibility thatquantitative phenotypic analysis of Yor1-ΔF gene inter-action reports on a complex trait in yeast of relevanceto biogenesis of CFTR-ΔF508. Thus, evolutionary con-servation is sufficient to usefully model human geneticdisease in yeast - at least in the case of CF. This opens adoor for efforts to dissect gene interaction underlyingphenotypic complexity through integration of yeast phe-nomic data with human genetic data. A few clinicallyrelevant genetic modifiers of cystic fibrosis disease wererecently identified, however these variants are not sus-pected to function in CFTR protein biogenesis pathways[2]. The genetic interaction model we have developedcould be useful to mine CFTR-ΔF508 GWAS data forvariant alleles that that modulate disease through effectson protein biogenesis. The Yor1-ΔF model suggests thepotential existence of a large number of such modifiers.Thus, the yeast phenomic model may inform humangenetic studies, where systematic, comprehensive, andquantitative analysis of gene interaction is of interest.Furthermore, given the large number of interactions, itwill likely be important in the future to analyze higherorder epistasis networks (for example, comprehensivethree-way gene-gene-gene interaction experiments),which is unforeseen employing human genetic dataalone.The outbred genetic structure of human populations,due to its combinatorial complexity, severely limits thepower to analyze phenotypes with respect to gene interac-tion [65]. Thus, tractable yeast phenomic models couldprovide a powerful and complementary tool for dissectingdisease complexity if the principle of evolutionary conser-vation of gene interaction applies [5]. Our work providesevidence in support of this concept, as we demonstratethat gene interactions discovered from the yeast Yor1-ΔFmodel resemble by homology gene interactions similarlycharacterized for CFTR-ΔF biogenesis in human cell mod-els. The findings support the notion that even when thephenotypic manifestations of homologous gene interactionappear unrelated (for example, oligomycin resistance inyeast vs. maintenance of peri-ciliary fluid depth in lungs),the principle network modulating the associated pheno-types can nevertheless be similar [5,66].We examined whether homologous modifiers ofCFTR-ΔF were among the stronger Yor1-ΔF interac-tions (Figure 5B). Conserved interactions were notnecessarily the strongest overall, raising points for con-sideration in future studies: (1) although strong hitsfrom genetic screens receive the most attention, weakLouie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103Page 14 of 20and intermediate strength interactions are also impor-tant for understanding the evolution of phenotypic var-iation; (2) the throughput and precision of Q-HTCP,which provides over 50,000 growth curves per experi-ment, is an enabling technology to map disease-relevantgene interaction networks, particularly regarding highquantitative accuracy to detect weak and intermediatestrength interaction with high confidence; (3) high con-fidence measures of gene interaction across the entiregenome will advance the opportunity to assess conserva-tion of between homologs at a systems level to deducefunctional modules that are most rapidly evolving withingene networks [42,67]; and (4) the elucidation of con-served aspects of a ‘ΔF biogenesis network’ provides astarting point to predict novel human homologs ofYor1-ΔF regulators, and ultimately define higher-orderinteractions from a gene network perspective [65,68].Thus, the Yor1-ΔF phenomic model can serve in severalways as a tool to discover and prioritize targets for ther-apeutic development as well as potential modifiers of CFdisease severity.We chose the CFTR-ΔF508 allele causing cystic fibrosisas proof of principle for modeling a human disease-rele-vant gene interaction network in yeast, because CFTR-ΔF508 is arguably the best-characterized human geneticdisease mutation. However, we anticipate that otherCFTR mutations in addition to CFTR-ΔF508 as well asother diseases entirely can be analogously modeled inyeast to generate useful insight and new hypotheses as tohow networks of interacting genes might modulate dis-ease expression. For diseases not having a single locusthat accounts for a high fraction of the phenotypic varia-tion, the power of experimentally tractable yeast epistasismodels may be even more beneficial [65]. Furthermore,yeast gene interactions also have been useful for uncover-ing genetic modifiers of foreign proteins; in one example,yeast gene interactions modulating alpha-synuclein toxi-city uncovered homologs that functioned similarly inanimal models of Parkinson’s disease, even though alpha-synuclein is not encoded by yeast genomes [69]. In asecond example, an informatics approach discovered‘phenologs’, defined as overlapping sets of homologousgenes associated with diverse phenotypic outcomesacross various species, thus discovering novel geneticrelationships between diverse phenotypes. Multiplepredictions were validated experimentally, includinghomologs of genes functioning in yeast cellular resistanceto HMG-CoA reductase inhibition influence angiogenesisin Xenopus embryos [66]. In a third example, a genome-wide screen revealed unexpectedly that threonine meta-bolism is required to buffer a deficiency of dNTPbiosynthesis, through augmenting provision of metabolicintermediates to overcome inhibition of a key enzyme,ribonucleotide reductase [20]. Although threoninebiosynthesis does not occur in multicellular eukaryotes, itwas nevertheless shown that threonine catabolism isrequired in a developmentally-regulated way for DNAsynthesis in mouse embryonic stem cells [70], and alsofor maintenance of stem cell chromatin state throughS-adenosyl-methionine metabolism and histone methyla-tion [71]. Our study, together with these and other mod-els indicate the power and utility of yeast genetic screensfor generating useful new hypotheses about the role ofgene interaction in phenotypic diversity, includinghuman disease [5,72].A novel aspect of the phenomic approach describedhere is the acquisition and analysis of time series datafrom proliferating cell arrays. These data fit well to alogistic growth equation so that growth curve parametersof individual cultures can be employed to precisely andaccurately quantify gene interaction (Figure 2). Couplingthis method with a gradation in perturbation states (forexample, multiple oligomycin concentrations) brings anew level of resolution to the powerful S. cerevisiaemethods for analyzing gene interaction. Previous large-scale gene interaction studies have used endpoint mea-surements of phenotypes (for example, colony outgrowthat one time point) and binary perturbation states, whichhave less sensitivity for detecting gene interaction due tolower precision and accuracy of quantifying growth phe-notypes [36]. The enhancement in quantitative resolutionprovided by Q-HTCP was significant, because many con-served interactions were intermediate in strength, andthus were more likely to have been missed by less quanti-tative methods (Figures 3 and 4) [13]. The validity ofweak to intermediate strength interaction was furtherclarified biochemically in several cases (Figures 4 to 7).The finding that gene interactions with Yor1-ΔF recapi-tulate homologous gene products interacting with humanCFTR-ΔF in mammalian cell-based studies provides evi-dence that gene interaction networks can be conservedover great evolutionary distances (Figure 5). Thus, despitedifferential selective pressure that these distantly relatedABC transporters have been subjected to, the cellular con-text in terms of interacting proteins that govern the bio-genesis of Yor1 and CFTR is conserved and renders yeasta useful and powerful model for cystic fibrosis. Although itremains to be tested, we speculate that GWAS-basedefforts to identify genetic modifiers of human diseasecould be aided by comprehensive and quantitative epista-sis data from yeast models [2]. An integrative/comparativeapproach could help prioritize findings diluted by multiplecomparisons from human genetic analysis. The yeast phe-nomic model provides a biological framework for identify-ing, within quantitative trait loci, candidate genes withputative functions worthy of further study [73].As another speculative example, it is plausible thatdeficiency of a cargo adapter protein, such as fromLouie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103Page 15 of 20Erv14 deletion, could give rise to a CF-like phenotypewithout mutations in CFTR itself (Figure 6). That Yor1required an ER export adaptor was in fact somewhatsurprising, because we had previously correlated ERexport of Yor1 with interaction between a well-charac-terized basic binding pocket on the surface of the vesiclecargo adaptor, Sec24, and a di-acidic export motif onYor1 [9]. Thus a potential explanation for the presentstudy findings is that Erv14 facilitates the Yor1/Sec24interaction. CFTR also employs a di-acidic motif, albeitin a distinct domain from that of Yor1, and Erv14 iswell conserved in metazoans [74], and therefore a simi-lar mechanism of ERV14 facilitating interaction duringcapture into transport vesicles is plausible for selectionof CFTR into ER-derived vesicles, and remains to betested.A potentially clinically relevant outcome of our studywas the discovery of a novel function for the recentlydescribed ER membrane complex. The EMC was discov-ered in a screen to find ER folding factors in yeast [58].We now show that deletion of any one of the members ofthe evolutionarily conserved protein complex yields aquantitatively similar deletion enhancer phenotype withrespect to Yor1-ΔF biogenesis (Figure 7). Interestingly,this interaction effect appears specific for the misfoldedprotein only, as deletion of members of the EMC did notaffect oligomycin sensitivity in the context of wild-typeYor1 expression. Further studies are needed to clarifythese findings, however we postulate a role for the EMC inthe early secretory pathway, and suspect it acts in a pro-biogenesis manner as part of the co-translational mechan-ism - perhaps for proteins prone to misfolding. We didnot see a role for the EMC proteins in protein turnover,since the half life or Yor1-ΔF was identical either in thepresence or absence of their expression. Instead, weobserved in the sop4-Δ0 mutant a reduced rate of produc-tion of Yor1-ΔF (Figure 7).Consistent with the above hypothesis, it was previouslynoted that deletion of the EMC proteins yields a geneticinteraction profile similar to over-expression of the sec61-2 mutation; thus, deletion of the EMC mimics genetic per-turbation of the Sec61 translocon. Furthermore, deletionof UBC7 or CUE1 (genes functioning in ERAD) wasaggravating in combination with deletion of either theEMC genes or sec61-2 overexpression [58,75]. Our inter-pretation of these data is that EMC and Sec61 act in afunctionally distinct pathway from ERAD, pathways thatcan buffer loss of one another [5,76]. Other evidence sug-gesting a role for the EMC in the early secretory pathwaycomes from a high content microscopy screen, which dis-covered loss of the EMC causes increased ER retention ofthe Mrh1-GFP fusion protein [77]. Importantly, we notethat the role of the EMC and other secretory protein bio-genesis network factors appears cargo-specific, since otherfactors that were found in the Mrh1-GFP screen exertedqualitatively different effects in our Yor1-ΔF screen [77].From a detailed comparison of our screen with the list ofgenes described by Bircham et al. to be required for for-ward transport of Mrh1-GFP, we noted that the EMCgenes and SOP4 were ΔF-specific deletion enhancers;GYP1, RAV2, VAC14, and MON2 were ΔF-specific dele-tion suppressors; PKR1 was a non-specific deletion enhan-cer; and most other genes (GOS1, PEP4, SPF1, VPS51,VPS53, VPS60, VTA1, YPT6, and OPI3) showed no effect.Thus, while several genes were found in both studies, onlyloss of function alleles of the EMC complex appeared tohave a consistent effect on prohibiting biogenesis of mem-brane proteins. Furthermore, for Yor1, prohibited biogen-esis was specific to the misfolded Yor1-ΔF.To test whether the EMC functions in a conservedmanner as a pro-biogenesis factor for CFTR-ΔF, weknocked down TTC35/EMC2 in transfected HeLa cellsexpressing CFTR-ΔF under temperature rescue condi-tions. Since we did not observe an effect of disrupting theEMC on Yor1-ΔF turnover, but rather a defect in Yor1-ΔF production, we tested for a pro-biogenesis function ofEMC2 on temperature-rescued CFTR-ΔF. We found thatloss of EMC2 reduced the steady state level of CFTR-ΔF,consistent with our Yor1-ΔF findings. These results pro-vide a strong rationale to utilize both yeast and humancells to clarify the pro-biogenesis mechanism for theEMC on ΔF-misfolded proteins (Figure 7).In summary, the datasets provided here will serve as aresource for further identification and prioritizationamong candidate genes and pathways contributing to cel-lular processing of misfolded proteins. Novel cellularpathways in addition to the ones discussed, were sug-gested by this study to be of importance for biogenesis ofmisfolded ABC transporter proteins and include mRNAprocessing (for example, SKI complex) and ribosome-associated functions, both of which were strong Yor1-ΔF-specific deletion suppressors. The similarity in theirimpact on Yor1-ΔF biogenesis could occur by influencingrates of translation and/or altering protein-foldingdynamics, although other mechanistic explanations areplausible. Future studies will be required to clarify therole of these and other genes in Yor1-ΔF biogenesis andtheir potential relevance to CFTR-ΔF. Additionally, giventhe abundance of pair-wise interactions revealed in ourstudy, three-way interaction analysis will become increas-ingly important to understand functional hierarchies inhigher-order epistasis networks that modulate Yor1-ΔFand, by extension, CFTR-ΔF protein biogenesis.ConclusionThe Yor1-ΔF670 gene interaction network was found tobe representative of CFTR-ΔF protein regulators identi-fied from human cell models. In addition, multiple newLouie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103Page 16 of 20functional categories of proteins were found to modu-late the activity of Yor1-ΔF, suggesting potential impor-tance of their homologs for CFTR-ΔF biogenesis.Validation of Yor1-ΔF interactors using biochemicalassays provided confidence in the functional significanceof the screening results, and led to the discovery that anevolutionarily conserved ER membrane complex simi-larly impacts biogenesis of Yor1-ΔF and CFTR-ΔF. Theoverall result suggests quantitative phenotyping of dou-ble mutant yeast expressing Yor1-ΔF is useful for mod-eling an evolutionarily conserved gene interactionnetwork functioning to modulate CFTR-ΔF biogenesis.The clinical relevance of the Yor1-ΔF gene interactionnetwork to cystic fibrosis remains to be established inpatients. Yet in principle, Q-HTCP affords a general plat-form to leverage the power of yeast genetics for exploringthe influence of gene interaction using other yeast phe-nomic models of disease. The approach could beextended, for example, to other cystic fibrosis-relevantmutations in Yor1, other molecular models of proteinmisfolding related disease, and homologous mutations inproteins covering a wide range of molecular functionswhere the cellular basis of disease involves evolutionarilyconserved processes.Supplementary informationThere are five additional files. Additional File 1 containsone table and four figures, as well as three supplementaldiscussion sections. All of the interaction data are avail-able in Additional Files 2, 3, and 4. REMc clusteringresults are provided in Additional File 5, and high confi-dence Yor1-ΔF interactions submitted to BioGRID areindicated in column L of the ‘REMc_data and clustering’worksheet. The criteria for selecting genes as high confi-dence are described in the ‘readme’ page of AdditionalFile 5. Only high-confidence, manually reviewed interac-tions (instead of all interactions beyond a certain quanti-tative threshold) were submitted to BioGRID (http://thebiogrid.org), for inclusion in the BioGRID databaseand SGD (http://yeastgenome.org). Interactions that wereconsidered lower confidence were excluded based on cri-teria such as a large effect of the gene deletion on growthin the absence of oligomycin or if gene-drug interactionoccurred in the presence of wild-type Yor1 expression, orif the dose response of interaction across all oligomycinconcentrations was not well fit to the quadratic equation.Additional materialAdditional File 1: This file contains three supplemental discussionsections, one table and four figures.Additional File 2: This file contains tables of screen-related data.Two screens were performed in the genomic collection of non-essentialgene knockouts. One screen was with the unmodified collection(wild-type YOR1 allele), and the other was with the yor1-ΔF allele thatwas introduced by the SGA method. Results from each screen are givenin two sheets. The first sheets (named ‘_screen’) have columns indicating:(A) the oligomycin concentration; (B) name of the ORF; (C) name of thegene deletion; (D) area under the growth curve; (E) the carrying capacity(’K’) from the logistic growth equation fit; (F) the rate (’r’) from thelogistic growth equation fit; (G) ‘L’ from the logistic growth equation fit;(H) R-squared value indicating residual after logistic growth fitting; and (I-N) upper and lower bounds (95% confidence intervals) on the logisticgrowth curve parameters. The second sheets (named ‘_interactions’)contain the following fields: (A, B) ORF and gene name for deletionstrain; (C) ‘ORF effect’- indicating the difference in L between the doublemutant and the median of 768 replicate yor1-ΔF single mutant culturesgrown in the absence of oligomycin; (D) the interactions quantified byquadratic fitting of the difference in L between double mutants and themedian of 768 replicate yor1-ΔF single mutant cultures, across increasingconcentrations of oligomycin, and taking the difference at the highestoligomycin concentration observed for the double mutant; and (E) thenumber of high oligomycin concentrations where no growth wasobserved for the deletion mutant (that is, deletion enhancer effects).Additional File 3: This file contains graphs of oligomycin responsefor deletion strains in background of wild-type YOR1 or yor1-ΔF,searchable by gene and ORF names. The graphs are ordered bydescending interaction score for yor1-ΔF double mutants, grouped alsoby the number of high oligomycin concentrations at which no growthwas observed (that is, from strongest deletion enhancer to strongestdeletion suppressor). ‘miss’ indicates the number of high oligomycinconcentrations at which no growth occurred; ‘ORF’ indicates thedifference in L between the deletion mutant and its wild-type referencestrain; ‘int’ is the interaction quantity at the highest oligomycinconcentration observed; ‘DR’ indicates the difference between the ranksof the deletion mutant in each of the two screens to indicate differentialinteractions in the context of wild-type Yor1 vs. Yor1-ΔF. The filledsymbols indicate the raw ‘L’ value at the zero-oligomycin concentrationfor the deletion mutant, by which all L values for the deletion mutantswere adjusted to quantify the interaction. The gray diamonds representthe 95% central distribution of 768 replicates of the reference strain, fromwhich the dose effect at each oligomycin concentration has beenremoved.Additional File 4: Like Supplemental Data File 2, this file containsgraphs of oligomycin response for deletion strains in backgroundof wild-type YOR1 or yor1-ΔF. Herein, less-interactive and non-interactive genes have been included for completeness.Additional File 5: This file contains input data used for REMc, REMcresults, and results of GO Term Finder analysis of REMc clusters.The sheet ‘REMc_data and clustering’ contains the following columns ofdata: (A) an arbitrary cluster # indicating discrete clusters for each gene;(B) the ORF and (C) gene name for the deletion strain; (D) the root and(E) round 1, (F) round 2, and (G) round 3 cluster names; (H-K and M-V)the interaction values for each gene-perturbation combination, for whichthe column labels are described in order in methods and materials;column L contains gene interactors submitted to BioGRID. The sheet‘cluster summary’ contains the following columns of data: (A) REMccluster ID; (B) the number of genes in the cluster; (C) the log-likelihoodof the cluster; (D) the GOid_z (Gene Ontology information divergencescore), a measure of functional enrichment (across all GO terms) amonggenes in each cluster; (E) the number of enriched GO Terms for each firstround cluster; (G-K) REMc data for Round 2 clusters; (M-Q) REMc data forRound 3 clusters. The sheet ‘GTF’ contains the following columns of data:(A) REMc cluster ID; (B) ‘selective’ GO Terms, meaning terms empiricallychosen to highlight relatively specific cellular processes to which a smallnumber of gene functions is ascribed. These are highlighted ashypothesis-generating terms worthy of further validation due to thepresence of multiple genes from the same biological module exertingsimilar patterns of gene-gene and gene-drug interaction. Manual reviewindicated in the ‘REMc_data and clustering’ sheet (column M) refers tointeractions that were filtered out before submission to BioGRID basedon inspection of the interaction graphs in Supplemental Data File 2 withattention to the following criteria: (1) the oligomycin dose response wasuncharacteristic (for example, reaching a plateau); (2) the dose responseLouie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103Page 17 of 20was not well fit to the quadratic; (3) the Yor1-ΔF double mutant strainwas very slow growing in the absence of oligomycin; (4) the deletionmutant in the context of Yor1 wild type suggested that oligomycin druginteraction was independent of Yor1-ΔF expression; (5) weak effects.List of abbreviationsABC: ATP-binding cassette; CF: Cystic fibrosis; CFTR: Cystic fibrosistransmembrane conductance regulator; ER: endoplasmic reticulum; ERAD:ER-associated degradation; EMC: ER membrane complex; GO: Gene ontology;GWAS: Genome-wide association studies;P-POD: Princeton Protein OrthologyDatabase; Q-HTCP: Quantitative high throughput cell array phenotyping;REMc: Recursive expectation-maximization clustering; SGA: Synthetic geneticarray; UPR: Unfolded protein response; UPRE: Unfolded protein responseelementAuthors’ contributionsEAM and JLH conceptualized the project and wrote the manuscript. JG andRJL performed the screen. JWR and JLH carried out the Q-HTCP analysis. RW,JLH, and JB devised the interaction model. JG, JLH, and BAM performed theREMc analysis. ML, JLH, and KD performed the homology mapping. RJL, SP,PK, and EAM carried out the molecular validation experiments in yeast, andJH and EJS performed the RNAi experiments in HeLa cells. All authors readand approved the final manuscript.Competing interestsJLH has a financial interest in Spectrum PhenomX, LLC, which holds alicense to commercialize Q-HTCP technology. All other authors declare nocompeting interests.AcknowledgementsThe authors thank Kevin Kirk, Lee Hartwell, and Marcus Lee for critique ofthe manuscript. The authors thank Sven Heinicke for help with yeast-humangene homology tables and Maria Costanzo for overlapping ORFs. Theauthors thank the funding agencies, which enabled the work, includingHoward Hughes Medical Institute (Physician Scientist Early Career Award P/SECA 57005927 to JLH), Cystic Fibrosis Foundation (Research Grant to EAMand JLH; student fellowship to JG; RDP grant to EJS for UAB CF Center), andNIH (GM085089 to EAM and GM071508 to David Botstein, which supportsP-POD).Author details1Department of Genetics, University of Alabama at Birmingham, 730 HughKaul Human Genetics Building, 720 20th Street South, Birmingham, AL 35294USA. 2Gregory James Cystic Fibrosis Center, University of Alabama atBirmingham, 790 McCallum Basic Health Sciences Building, 1918 UniversityBoulevard, Birmingham, AL 35294 USA. 3Department of Biology, ColumbiaUniversity, 1212 Amsterdam Ave. MC2456, New York, NY 10027 USA.4Department of Statistics and Michael Smith Laboratories, University ofBritish Columbia, 3182 Earth Sciences Building, 2207 Main Mall, Vancouver,BC V6T 1Z4 Canada. 5Lewis-Sigler Institute for Integrative Genomics,Princeton University, Washington Road, Princeton, NJ 08544 USA.6Department of Computer Science, University of Tulsa, 2145 J. NewtonRayzor Hall, E 5th Pl, Tulsa, OK, 74104 USA.Received: 8 August 2012 Revised: 27 November 2012Accepted: 27 December 2012 Published: 27 December 2012References1. Badano JL, Katsanis N: Beyond Mendel: an evolving view of humangenetic disease transmission. Nat Rev Genet 2002, 3:779-789.2. Wright FA, Strug LJ, Doshi VK, Commander CW, Blackman SM, Sun L,Berthiaume Y, Cutler D, Cojocaru A, Collaco JM, Corey M, Dorfman R,Goddard K, Green D, Kent JW Jr, Lange EM, Lee S, Li W, Luo J, Mayhew GM,Naughton KM, Pace RG, Pare P, Rommens JM, Sandford A, Stonebraker JR,Sun W, Taylor C, Vanscoy LL, Zou F, et al: Genome-wide association andlinkage identify modifier loci of lung disease severity in cystic fibrosis at11p13 and 20q13.2. Nat Genet 2011, 43:539-546.3. Consortium TWTCC: Genome-wide association study of 14,000 cases ofseven common diseases and 3,000 shared controls. Nature 2007,447:661-678.4. Park JH, Wacholder S, Gail MH, Peters U, Jacobs KB, Chanock SJ,Chatterjee N: Estimation of effect size distribution from genome-wideassociation studies and implications for future discoveries. Nat Genet2010, 42:570-575.5. Hartman IV JL, Garvik B, Hartwell L: Principles for the buffering of geneticvariation. Science 2001, 291:1001-1004.6. Schadt EE, Bjorkegren JL: NEW: network-enabled wisdom in biology,medicine, and health care. Sci Transl Med 2012, 4:115rv111.7. Rowe SM, Miller S, Sorscher EJ: Cystic fibrosis. N Engl J Med 2005,352:1992-2001.8. Katzmann DJ, Epping EA, Moye-Rowley WS: Mutational disruption ofplasma membrane trafficking of Saccharomyces cerevisiae Yor1p, ahomologue of mammalian multidrug resistance protein. Mol Cell Biol1999, 19:2998-3009.9. Pagant S, Kung L, Dorrington M, Lee MC, Miller EA: Inhibiting endoplasmicreticulum (ER)-associated degradation of misfolded Yor1p does notpermit ER export despite the presence of a diacidic sorting signal. MolBiol Cell 2007, 18:3398-3413.10. Singh I, Pass R, Togay SO, Rodgers JW, Hartman IV JL: Stringent mating-type-regulated auxotrophy increases the accuracy of systematic geneticinteraction screens with Saccharomyces cerevisiae mutant arrays.Genetics 2009, 181:289-300.11. Tong AH, Boone C: Synthetic genetic array analysis in Saccharomycescerevisiae. Methods Mol Biol 2006, 313:171-192.12. Winzeler EA, Shoemaker DD, Astromoff A, Liang H, Anderson K, Andre B,Bangham R, Benito R, Boeke JD, Bussey H, Chu AM, Connelly C, Davis K,Dietrich F, Dow SW, El Bakkoury M, Foury F, Friend SH, Gentalen E,Giaever G, Hegemann JH, Jones T, Laub M, Liao H, Davis RW, et al:Functional characterization of the S. cerevisiae genome by genedeletion and parallel analysis. Science 1999, 285:901-906.13. Hartman IV JL, Tippery NP: Systematic quantification of gene interactionsby phenotypic array analysis. Genome Biol 2004, 5:R49.14. Shah NA, Laws RJ, Wardman B, Zhao LP, Hartman IV JL: Accurate, precisemodeling of cell proliferation kinetics from time-lapse imaging andautomated image analysis of agar yeast culture arrays. BMC Syst Biol2007, 1:3.15. Collawn JF, Fu LW, Bebok Z: Targets for cystic fibrosis therapy: proteomicanalysis and correction of mutant cystic fibrosis transmembraneconductance regulator. Expert Review of Proteomics 2010, 7:495-506.16. Okiyoneda T, Barriere H, Bagdany M, Rabeh WM, Du K, Hohfeld J, Young JC,Lukacs GL: Peripheral protein quality control removes unfolded CFTRfrom the plasma membrane. Science 2010, 329:805-810.17. Keith CT, Borisy AA, Stockwell BR: Multicomponent therapeutics fornetworked systems. Nat Rev Drug Discov 2005, 4:71-78.18. Saccharomyces Genome Deletion Project.. [http://yeastdeletion.stanford.edu].19. Louie RJ, Pagant S, Youn JY, Halliday JJ, Huyer G, Michaelis S, Miller EA:Functional rescue of a misfolded eukaryotic ATP-binding cassettetransporter by domain replacement. J Biol Chem 2010, 285:36225-36234.20. Hartman IV JL: Buffering of deoxyribonucleotide pool homeostasis bythreonine metabolism. Proc Natl Acad Sci USA 2007, 104:11700-11705.21. Codon AC, Gasent-Ramirez JM, Benitez T: Factors which affect thefrequency of sporulation and tetrad formation in Saccharomycescerevisiae baker’s yeasts. Appl Environ Microbiol 1995, 61:630-638.22. BioGrid.. [http://thebiogrid.org].23. Saccharomyces Genome Database. [http://yeastgenome.org].24. Guo J, Tian D, McKinney BA, Hartman JL: Recursive expectation-maximization clustering: a method for identifying buffering mechanismscomposed of phenomic modules. Chaos 2010, 20:026103.25. Burke D, Dawson D, Stearns T: Methods in Yeast Genetics. Long Island, NY:CSHL Press; 2000.26. Princeton Protein Orthology Database (P-POD). [http://ppod.princeton.edu/].27. Livstone MS, Oughtred R, Heinicke S, Vernot B, Huttenhower C, Durand D,Dolinski K: Inferring protein function from homology using the PrincetonProtein Orthology Database (P-POD). Curr Protoc Bioinformatics 2011,Chapter 6, Unit 6 11.Louie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103Page 18 of 2028. Dolinski K, Botstein D: Orthology and functional conservation ineukaryotes. Annu Rev Genet 2007, 41:465-507.29. Pagant S, Halliday JJ, Kougentakis C, Miller EA: Intragenic suppressingmutations correct the folding and intracellular traffic of misfoldedmutants of Yor1p, a eukaryotic drug transporter. J Biol Chem 2010,285:36304-36314.30. CFTR Folding Consortium. [http://www.cftrfolding.org/index.htm].31. Decottignies A, Goffeau A: Complete inventory of the yeast ABC proteins.Nat Genet 1997, 15:137-145.32. Katzmann DJ, Hallstrom TC, Voet M, Wysock W, Golin J, Volckaert G, Moye-Rowley WS: Expression of an ATP-binding cassette transporter-encodinggene (YOR1) is required for oligomycin resistance in Saccharomycescerevisiae. Mol Cell Biol 1995, 15:6875-6883.33. St Onge RP, Mani R, Oh J, Proctor M, Fung E, Davis RW, Nislow C, Roth FP,Giaever G: Systematic pathway analysis using high-resolution fitnessprofiling of combinatorial gene deletions. Nat Genet 2007, 39:199-206.34. Baryshnikova A, Costanzo M, Kim Y, Ding H, Koh J, Toufighi K, Youn JY,Ou J, San Luis BJ, Bandyopadhyay S, Hibbs M, Hess D, Gingras AC,Bader GD, Troyanskaya OG, Brown GW, Andrews B, Boone C, Myers CL:Quantitative analysis of fitness and genetic interactions in yeast on agenome scale. Nat Methods 2010, 7:1017-1024.35. Collins SR, Schuldiner M, Krogan NJ, Weissman JS: A strategy for extractingand analyzing large-scale quantitative epistatic interaction data. GenomeBiol 2006, 7:R63.36. Mani R, St Onge RP, Hartman IV JL, Giaever G, Roth FP: Defining geneticinteraction. Proc Natl Acad Sci USA 2008, 105:3461-3466.37. Kellis M, Patterson N, Endrizzi M, Birren B, Lander ES: Sequencing andcomparison of yeast species to identify genes and regulatory elements.Nature 2003, 423:241-254.38. Ward CL, Omura S, Kopito RR: Degradation of CFTR by the ubiquitin-proteasome pathway. Cell 1995, 83:121-127.39. Bebok Z, Mazzochi C, King SA, Hong JS, Sorscher EJ: The mechanismunderlying cystic fibrosis transmembrane conductance regulatortransport from the endoplasmic reticulum to the proteasome includesSec61beta and a cytosolic, deglycosylated intermediary. J Biol Chem1998, 273:29873-29878.40. Tischler J, Lehner B, Fraser AG: Evolutionary plasticity of geneticinteraction networks. Nat Genet 2008, 40:390-391.41. Dixon SJ, Fedyshyn Y, Koh JL, Prasad TS, Chahwan C, Chua G, Toufighi K,Baryshnikova A, Hayles J, Hoe KL, Kim DU, Park HO, Myers CL, Pandey A,Durocher D, Andrews BJ, Boone C: Significant conservation of syntheticlethal genetic interaction networks between distantly relatedeukaryotes. Proc Natl Acad Sci USA 2008, 105:16653-16658.42. Roguev A, Bandyopadhyay S, Zofall M, Zhang K, Fischer T, Collins SR, Qu H,Shales M, Park HO, Hayles J, Hoe KL, Kim DU, Ideker T, Grewal SI,Weissman JS, Krogan NJ: Conservation and rewiring of functional modulesrevealed by an epistasis map in fission yeast. Science 2008, 322:405-410.43. Bilan F, Nacfer M, Fresquet F, Norez C, Melin P, Martin-Berge A, Costa deBeauregard MA, Becq F, Kitzis A, Thoreau V: Endosomal SNARE proteinsregulate CFTR activity and trafficking in epithelial cells. Exp Cell Res 2008,314:2199-2211.44. Naren AP, Quick MW, Collawn JF, Nelson DJ, Kirk KL: Syntaxin 1A inhibitsCFTR chloride channels by means of domain-specific protein-proteininteractions. Proc Natl Acad Sci USA 1998, 95:10972-10977.45. Peters KW, Qi J, Watkins SC, Frizzell RA: Syntaxin 1A inhibits regulatedCFTR trafficking in xenopus oocytes. Am J Physiol 1999, 277:C174-180.46. Tang BL, Gee HY, Lee MG: The cystic fibrosis transmembrane conductanceregulator’s expanding SNARE interactome. Traffic 2011, 12:364-371.47. Gentzsch M, Chang XB, Cui L, Wu Y, Ozols VV, Choudhury A, Pagano RE,Riordan JR: Endocytic trafficking routes of wild type and DeltaF508 cysticfibrosis transmembrane conductance regulator. Mol Biol Cell 2004,15:2684-2696.48. Swiatecka-Urban A, Talebian L, Kanno E, Moreau-Marquis S, Coutermarsh B,Hansen K, Karlson KH, Barnaby R, Cheney RE, Langford GM, Fukuda M,Stanton BA: Myosin Vb is required for trafficking of the cystic fibrosistransmembrane conductance regulator in Rab11a-specific apicalrecycling endosomes in polarized human airway epithelial cells. J BiolChem 2007, 282:23725-23736.49. Goeckeler JL, Brodsky JL: Molecular chaperones and substrateubiquitination control the efficiency of endoplasmic reticulum-associated degradation. Diabetes Obes Metab 2010, Suppl 2: 32-38.50. Grove DE, Fan CY, Ren HY, Cyr DM: The endoplasmic reticulum-associatedHsp40 DNAJB12 and Hsc70 cooperate to facilitate RMA1 E3-dependentdegradation of nascent CFTRDeltaF508. Mol Biol Cell 2011, 22:301-314.51. Schmidt BZ, Watts RJ, Aridor M, Frizzell RA: Cysteine string proteinpromotes proteasomal degradation of the cystic fibrosis transmembraneconductance regulator (CFTR) by increasing its interaction with theC terminus of Hsp70-interacting protein and promoting CFTRubiquitylation. J Biol Chem 2009, 284:4168-4178.52. Yamamoto YH, Kimura T, Momohara S, Takeuchi M, Tani T, Kimata Y,Kadokura H, Kohno K: A novel ER J-protein DNAJB12 accelerates ER-associated degradation of membrane proteins including CFTR. Cell StructFunct 2010, 35:107-116.53. Hartwell LH, Hopfield JJ, Leibler S, Murray AW: From molecular to modularcell biology. Nature 1999, 402:C47-52.54. Boyle EI, Weng S, Gollub J, Jin H, Botstein D, Cherry JM, Sherlock G:GO::TermFinder–open source software for accessing Gene Ontologyinformation and finding significantly enriched Gene Ontology termsassociated with a list of genes. Bioinformatics 2004, 20:3710-3715.55. Setty SR, Strochlic TI, Tong AH, Boone C, Burd CG: Golgi targeting of ARF-like GTPase Arl3p requires its Nalpha-acetylation and the integralmembrane protein Sys1p. Nat Cell Biol 2004, 6:414-419.56. Powers J, Barlowe C: Erv14p directs a transmembrane secretory proteininto COPII-coated transport vesicles. Mol Biol Cell 2002, 13:880-891.57. Miller E, Antonny B, Hamamoto S, Schekman R: Cargo selection into COPIIvesicles is driven by the Sec24p subunit. Embo J 2002, 21:6105-6113.58. Jonikas MC, Collins SR, Denic V, Oh E, Quan EM, Schmid V, Weibezahn J,Schwappach B, Walter P, Weissman JS, Schuldiner M: Comprehensivecharacterization of genes required for protein folding in theendoplasmic reticulum. Science 2009, 323:1693-1697.59. Travers KJ, Patil CK, Wodicka L, Lockhart DJ, Weissman JS, Walter P:Functional and genomic analyses reveal an essential coordinationbetween the unfolded protein response and ER-associated degradation.Cell 2000, 101:249-258.60. Wang X, Venable J, LaPointe P, Hutt DM, Koulov AV, Coppinger J, Gurkan C,Kellner W, Matteson J, Plutner H, Riordan JR, Kelly JW, Yates JR, Balch WE:Hsp90 cochaperone Aha1 downregulation rescues misfolding of CFTR incystic fibrosis. Cell 2006, 127:803-815.61. Frost A, Elgort MG, Brandman O, Ives C, Collins SR, Miller-Vedam L,Weibezahn J, Hein MY, Poser I, Mann M, Hyman AA, Weissman JS:Functional repurposing revealed by comparing S. pombe and S.cerevisiae genetic interactions. Cell 2012, 149:1339-1352.62. Koch EN, Costanzo M, Bellay J, Deshpande R, Chatfield-Reed K, Chua G,D’Urso G, Andrews BJ, Boone C, Myers CL: Conserved rules govern geneticinteraction degree across species. Genome Biol 2012, 13:R57.63. Barabasi AL, Gulbahce N, Loscalzo J: Network medicine: a network-basedapproach to human disease. Nat Rev Genet 2011, 12:56-68.64. Schadt EE: Molecular networks as sensors and drivers of commonhuman diseases. Nature 2009, 461:218-223.65. Phillips PC: Epistasis–the essential role of gene interactions in thestructure and evolution of genetic systems. Nat Rev Genet 2008,9:855-867.66. McGary KL, Park TJ, Woods JO, Cha HJ, Wallingford JB, Marcotte EM:Systematic discovery of nonobvious human disease models throughorthologous phenotypes. Proc Natl Acad Sci USA 2010, 107:6544-6549.67. Kirschner M, Gerhart J: Evolvability. Proc Natl Acad Sci USA 1998,95:8420-8427.68. Dowell RD, Ryan O, Jansen A, Cheung D, Agarwala S, Danford T,Bernstein DA, Rolfe PA, Heisler LE, Chin B, Nislow C, Giaever G, Phillips PC,Fink GR, Gifford DK, Boone C: Genotype to phenotype: a complexproblem. Science 2010, 328:469.69. Auluck PK, Caraveo G, Lindquist S: alpha-Synuclein: membraneinteractions and toxicity in Parkinson’s disease. Annu Rev Cell Dev Biol2010, 26:211-233.70. Wang J, Alexander P, Wu L, Hammer R, Cleaver O, McKnight SL:Dependence of mouse embryonic stem cells on threonine catabolism.Science 2009, 325:435-439.71. Shyh-Chang N, Locasale JW, Lyssiotis CA, Zheng Y, Teo RY,Ratanasirintrawoot S, Zhang J, Onder T, Unternaehrer JJ, Zhu H, Asara JM,Daley GQ, Cantley LC: Influence of threonine metabolism on S-adenosylmethionine and histone methylation. Science 2013,339:222-226.Louie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103Page 19 of 2072. Hartman JL IV: Genetic and Molecular Buffering of Phenotypes. InNutritional Genomics: Discovering the Path to Personalized Nutrition. Volume1.. Edited by: Rodriguez R, Kaput J. 1st edn. Hoboken. NJ: John Wiley2006:496.73. Ritchie MD: Using biological knowledge to uncover the mystery in thesearch for epistasis in genome-wide association studies. Ann Hum Genet2011, 75:172-182.74. Bokel C, Dass S, Wilsch-Brauninger M, Roth S: Drosophila Cornichon acts ascargo receptor for ER export of the TGFalpha-like growth factor Gurken.Development 2006, 133:459-470.75. Sommer T, Jentsch S: A protein translocation defect linked to ubiquitinconjugation at the endoplasmic reticulum. Nature 1993, 365:176-179.76. Ihmels J, Collins SR, Schuldiner M, Krogan NJ, Weissman JS: Backup withoutredundancy: genetic interactions reveal the cost of duplicate gene loss.Mol Syst Biol 2007, 3:86.77. Bircham PW, Maass DR, Roberts CA, Kiew PY, Low YS, Yegambaram M,Matthews J, Jack CA, Atkinson PH: Secretory pathway genes assessed byhigh-throughput microscopy and synthetic genetic array analysis. MolBiosyst 2011, 7:2589-2598.78. Fisk DG, Ball CA, Dolinski K, Engel SR, Hong EL, Issel-Tarver L, Schwartz K,Sethuraman A, Botstein D, Cherry JM: Saccharomyces cerevisiae S288Cgenome annotation: a working hypothesis. Yeast 2006, 23:857-865.doi:Cite this article as: Louie et al.: A yeast phenomic model for the geneinteraction network modulating CFTR-ΔF508 protein biogenesis. GenomeMedicine 2012 4:103.Submit your next manuscript to BioMed Centraland take full advantage of: • Convenient online submission• Thorough peer review• No space constraints or color figure charges• Immediate publication on acceptance• Inclusion in PubMed, CAS, Scopus and Google Scholar• Research which is freely available for redistributionSubmit your manuscript at www.biomedcentral.com/submitLouie et al. Genome Medicine 2012, 4:103http://genomemedicine.com/content/4/12/103Page 20 of 20


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items