The MATa haploid yeast deletion library with BY4741 background (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0) was purchased from Invitrogen, United States. The library was transformed with pYES2-CdtB, a plasmid carrying AaCdtB under galactose-inducible promoter [described in Matangkasombut et al. (2010)], in 96-well plates following a protocol previously described (Voynov et al., 2006). Briefly, yeast cultures were incubated with PEG/LiOAc/TE/ssDNA mixture for 3 h at 30°C, and heat shocked for 45 min at 42°C. Transformants were selected first in liquid synthetic complete media (SC) lacking uracil (SC-Ura) (ingredients from Sigma-Aldrich, United States) for 3 days and on solid SC-Ura for 2 days. Any mutants that failed to transform in batch were individually transformed using standard protocol (Gietz and Woods, 2002).

The transformed yeast library was grown overnight in SC-Ura supplemented with 2% glucose. The cultures were serially diluted and inoculated with a 96-pin floating pin replicator (V&P Scientific, Inc., United States) on SC-Ura media containing 2% galactose (inducing media) and on that containing 2% glucose (repressing media) as a control. Mutants showing significant growth on galactose plates, in the dilutions where the wild-type yeast strain BY4741 did not grow, were selected for further confirmatory screening.

Subsequently, the mutants selected in the primary screen were compiled in 96-well plates and rescreened by spotting dilutions of cultures onto glucose and galactose plates. Mutants that grew well on galactose plates were further analyzed by spot tests as described (Matangkasombut et al., 2010). Briefly, log phase cultures were serially diluted and spotted on inducing and repressing media. After 40–48 h of incubation at 30°C, yeast growth was observed. Mutants that showed better growth than wild-type on the galactose plate were selected for rescreening. After at least four repeats, mutants that consistently show resistance to CdtB were classified as CdtB resistant mutants.

CdtB expression in CdtB resistant mutants was examined using an immunoblot assay. CdtB resistant mutants were grown to log phase and galactose was added to 2% to induce CdtB expression. Whole cell lysates were extracted using glass beads and trichloroacetic acid (Keogh et al., 2006). The samples were resolved using SDS-PAGE and transferred onto nitrocellulose membranes. Immunoblot was performed using Rabbit anti-AaCdtB serum (Sugai et al., 1998) and goat anti-rabbit IgG antibody, conjugated with HRP to detect CdtB protein in the whole cell extract.

The list of genes whose deletion conferred CdtB resistance from our genome-wide screen was compiled and analyzed using web-based tools available on the Saccharomyces genome database (SGD). GO Slim Mapper¹ was used to map the gene list to broad GO terms (Yeast-GO slim) related to biological processes, molecular functions, and cellular components. In addition, GO Term Finder² was used to search for GO terms that are shared among the genes in the list where p < 0.05 is considered significant.

The plasmid carrying CdtB-EGFP fusion was constructed by inserting a Klenow-treated NotI-XbaI cdtB-containing fragment from pYES-CdtB into BamHI-XbaI sites of a pYES2-EGFP plasmid (a gift from Prof. C. Boonchird, Mahidol University, Thailand) by blunt-end ligation. A 11-amino acid linker was inserted between the CdtB and EGFP open reading frames. After transformation into BY4741, CdtB-EGFP protein expression was confirmed by immunoblotting and using the CdtB susceptibility spot assay (Supplementary Figure 1).

For the protein localization study, yeast cultures were transformed with pYES2-CdtB-EGFP and grown to mid-log phase. After CdtB induction by growing in media containing 2% galactose for 8 h, cells were fixed in 3.7% formaldehyde in PBS buffer (pH7.4) for 15 min and permeabilized with 1% triton X-100 at 42°C for 30 min. Samples were incubated with 0.1 μg/ml DAPI for 5 min for yeast nuclei staining. Cells were washed twice with PBS and dropped on 1% agarose patch on glass slides. Fluorescent images were captured with confocal fluorescent microscope (Olympus, FV10i-ASW model) at 60× optical magnification.

Fluorescent images were analyzed by Mander’s coefficient function of Just Another Co-localization Plug-in (JACoP) in the ImageJ program (Bolte and Cordelieres, 2006). The mean fluorescent intensity and standard deviation (SD) for each cell were calculated for DAPI and EGFP channels. The threshold level for the positive signal of DAPI was set at ≥ mean+1SD, while that of EGFP was set at ≥ mean+2SD. For each cell, we first defined the DAPI-positive area as the nucleus, and counted the pixel number with EGFP signal above the threshold. Then, we calculated the relative percentage of nuclear localization of CdtB by comparing the percentage of EGFP-positive pixel numbers in nucleus over the total in yeast cells expressing CdtB-EGFP vs. EGFP control. The data was analyzed using Mann-Whitney U test with p < 0.05 considered statistically significant.

Approximately 5,000 non-essential open reading frames (ORFs) from the yeast deletion library were successfully transformed with pYES-CdtB and were screened for CdtB susceptibility. From the primary screening, we identified 513 deletion mutants that showed growth on 2% galactose (CdtB inducing media) in the dilutions where most mutants did not grow (Figure 1A). These were compiled in 96-well plates and dilutions were spotted on solid media with 2% galactose and with 2% glucose (repressing media) (Figure 1B). Mutants that showed growth on a galactose plate, at the dilution where the wild-type did not grow, were selected for further confirmation. The CdtB resistant phenotype was confirmed by spotting serial dilutions of cultures for CdtB susceptibility tests (Figure 1C). When comparing the growth on galactose with that on glucose, CdtB expression leads to reduction of both the number of colonies and colony size in wild-type yeast (Figure 1C). We previously showed, using survival plating assays to quantify the effect of CdtB, that CdtB expression reduces the number of surviving colonies by approximately 50% in wild-type yeast (Matangkasombut et al., 2010). Different degrees of CdtB expression led to dose-dependent effects on yeast growth. These could be observed by both survival plating and plate sensitivity (spot) assay (Supplementary Figures 2A–C). However, the plate sensitivity assay allows us to better observe the overall effect of CdtB, including the effect on colony size. Since CdtB-induced cell cycle arrest can lead to a small colony size, formation of larger colonies on galactose media also reflects resistance to the effect of CdtB. The effect of CdtB could be more clearly observed at an early time after incubation (approximately 40 h) (Supplementary Figure 2D). In addition, since various gene deletion mutants may grow at different rates, we examined the effect of CdtB by comparing the degrees of growth reduction on media with galactose vs. with glucose of the mutants to that of the wild-type (Figure 1C).

Altogether, we identified 281 mutants that consistently showed better growth than wild-type at various levels, in terms of number of colonies and colony size, in at least four repeats (Supplementary Figure 3). To rule out the possibility of resistance due to defect in CdtB expression, whole cell extracts of these 281 mutants were examined by anti-CdtB immunoblot assay (data not shown). Mutants that showed no or weak expressions of CdtB were omitted from further analysis. Finally, 243 mutants were classified as CdtB resistant (Supplementary Table 1).

The list of genes whose deletion confers CdtB resistance (243 genes) was analyzed by GO Slim Mapper tool on SGD. These genes have been annotated to localize to a variety of cellular components (Figure 2A). The top 10 broad GO terms identified according to the 3 categories: cellular components, biological processes, and molecular functions, are summarized in Figure 2B–D, respectively.

Further analysis using GO Term Finder identified “organic anion transport (OAT)” (GO:0015711) to be significantly enriched in the gene list, as compared to the genome frequency in the biological process category (p-value = 0.03846). There are 16 genes in the list of 243 CdtB resistant genes (6.6%) with this GO term annotation, while the genome frequency was 135 in 7165 genes (1.9%). These 16 genes are listed in Table 1. We searched for GO terms associated with these 16 genes in relation to cellular components and found that most genes are localized in the membrane part (15 genes), vacuole (8 genes) and vacuolar membrane (6 genes). When we searched for GO terms related to molecular functions, most genes show transporter activity (14 genes) and terms related to transmembrane transporter activity (9 genes). These results suggest that these 16 genes, that localize to membrane/vacuole and function as transporters, may play an important role in CdtB cytotoxicity.

Since a significant gap in our knowledge on intracellular translocation of CdtB is with regards to how CdtB moves from the ER to the nucleus, we identified genes in the CdtB resistant list that localize to the ER. From our search, 25 genes showed ER localization (Table 2). Among these, 12 genes localize to the ER membrane. Interestingly, HRD3, which encodes an ER membrane protein that plays a central role in ER-associated protein degradation (ERAD), was also included in this list.

As we found an enrichment in genes related to organic anion transport in the CdtB resistant list, we hypothesized that the mutations may disrupt CdtB translocation into the nucleus. We used CdtB-EGFP fusion protein as a tool to examine CdtB localization in these 16 mutants. After inducing CdtB-EGFP expression, fluorescent signals were observed using a confocal fluorescent microscope. The results are shown in Figure 3A. CdtB nuclear localization was evaluated by Mander’s coefficient in comparison to the EGFP vector control. The results showed that CdtB nuclear localization in all mutants are lower than that in the wild-type, as shown in Figure 3B.