The anti-cancer transition-state inhibitor MTDIA inhibits human MTAP, inducing autophagy in humanized yeast

Abstract

Methylthioadenosine-DADMe immucillin-A (MTDIA) is a transition-state analog that potently inhibits the human protein 5'-methylthioadenosine phosphorylase (MTAP) at picomolar concentrations and elicits anti-tumor activity against lung, prostate, colon, cervical, head and neck, and triple-negative breast cancers in cell and animal models. The anti-cancer mechanisms of MTDIA involve elevated methylthioadenosine levels but are not fully understood. The yeast protein MEU1 is functionally equivalent to human MTAP. To gain further understanding, we performed chemical genetic analyses via gene deletion and GFP-tagged protein libraries in yeast that express a member of the human equilibrative nucleoside transporter (ENT) family to permit MTDIA uptake. Genomic and proteomic analyses identified genes and proteins critical to MTDIA bioactivity. Network analysis of these genes and proteins revealed an important link to ribosomal function, which was confirmed by observing reduced levels of ribosomal subunit proteins. Network analysis also implicated autophagy, which was confirmed by analyzing intracellular trafficking of GFP-Atg8 and Phloxine B viability. In yeast, a comparable effect occurred after deletion of MEU1, indicating a single target for MTDIA in yeast. Overall, our yeast model reveals specific components of the ribosome as well as induction of autophagy as integral mechanisms that mediate the bioactivity of MTDIA.

Keywords: Autophagy; Betweenness centrality; Chemical biology; Chemical genetics; Drug−drug synergy; Network analysis; Nucleoside/nucleotide metabolism; Synthetic lethality; Transition state analogs; Yeast genetics.

Fig. 1.

The current working model for the MTDIA anti-cancer mechanism of action is conserved from yeast to humans. MTDIA is an inhibitor of 5′-methylthioadenosine phosphorylase (MTAP in humans, MEU1 in yeast), resulting in 5′-methylthioadenosine (MTA) accumulation and disruptions to methionine metabolism and polyamine synthesis. MTAP catalyzes the phosphorolysis of MTA to form adenine and 5-methylthio-α-D-ribose 1-phosphate (MTR-1-P) that, ultimately, can be converted to S-adenosyl-L-methionine (SAM).

Fig. 2.

Heterologous expression of hENT1 permits nucleoside uptake. (A) Growth of ade2Δade3Δ in Y7092 (an adenine and methionine auxotroph) expressing empty vector (EV) or hENT1 was quantified in the presence of adenine, adenosine or 5′-methylthioadeonisine (MTA) after 48 h growth. (B,C) Growth of Y7092 (a methionine auxotroph) expressing EV or hENT1 was quantified in the presence of varying concentrations of toyocamycin (B) or cordycepin (C). Cells were first cultured overnight in SD medium lacking uracil but supplemented with raffinose as the carbon source (SD-U+R), then subcultured in SD medium lacking adenine and uracil but supplemented with raffinose and galactose as the carbon source (SD-AU+RG) (panel A) or in SD medium lacking uracil with raffinose and galactose as the carbon source (SD-U+RG) (panels B,C) to 5×10⁵ cells/ml, treated with nucleosides or nucleobases and incubated at 30°C. Cell growth was quantified hourly via OD₆₀₀ readings. Plotted is the growth of treated and untreated cells (in %) at time points when untreated cells were at mid-log. Results shown in each panel represent the mean±s.d. for three biological replicates.

Fig. 3.

MTDIA is a potent, cytotoxic inhibitor of yeast MTAP when methionine salvage is essential. (A,B) Growth of Y7092 cells expressing hENT1 (Y7092+hENT1) was quantified in MTA-supplemented medium lacking uracil (SD-U) for selection of hENT1 with or without methionine, adenine and cysteine (the MEU1 essential condition SD-MAUC) in varying concentrations of MTDIA (A) or DIA (B). Cells were cultured overnight without galactose in SD-U with raffinose (SD-U+R) medium, subcultured in SD-U with raffinose and galactose (SD-U+RG) or SD-MAUC with raffinose and galactose (SD-MAUC+RG) supplemented±250 µM MTA, and treated with varying concentrations of MTDIA (A) or DIA (B). All cells were then incubated at 30°C, growth was quantified hourly via OD₆₀₀ readings. Plotted is the mean growth (in %) of treated and untreated cells at time points when untreated cells were at mid-log. (C) Growth of MTDIA-treated Y7092+hENT1 cells on medium lacking MTDIA. Mid-log cells were treated at 30°C with varying concentrations of MTDIA for different lengths of time, plated on medium selecting for the Meu1 essential condition (left) or medium selecting only for the hENT1 plasmid (right), and incubated at 30°C for 48 h. Results shown in each panel are representative of three biological replicates.

Fig. 4.

Genome-wide analyses identify gene deletions and biological functions that are epistatic to the MTDIA target protein Meu1. (A) Synthetic genetic array (SGA) analysis generated 4286 meu1ΔxxxΔ strains. Growth of these double mutant strains was compared with growth of the meu1Δ query strain and the xxxΔ strain in the deletion library. Three independent SGAs were performed by crossing the meu1Δ query strain with the strains of the haploid single-gene deletion library (Tong et al., 2001) and growth was quantified using Gitter (Wagih and Parts, 2014) and ScreenMill (Dittmar et al., 2010). By using liquid growth assay, genetic interactions were validated for 64 negative and six positive genetic interactions. Cells were cultured overnight, subcultured to a density of 5×10⁵ cells/ml and incubated in SD-MAUC+MTA+RG medium at 30°C. Percent growth (in %, y-axis) in each meu1ΔxxxΔ compared to the parental meu1Δ and xxxΔ was determined via OD₆₀₀ readings when xxxΔ cells were at mid-log. The criterium for all strains shown was a statistically significant growth difference (either 25% growth inhibition or 30% growth improvement) of each meu1ΔxxxΔ compared to the parental meu1Δ and xxxΔ strains (P<0.05). (B) Chemical genetics analysis measured growth of 4310 xxxΔ+hENT1 strains on 3 nM MTDIA relative to vehicle control from biological triplicates of technical quadruplicate experiments on agar. Three independent chemical genetic screens were performed on SD-MAUC+MTA+RG agar and growth was quantified using Gitter (Wagih and Parts, 2014) and ScreenMill (Dittmar et al., 2010). Genetic interactions were validated for 61 negative and 30 positive genetic interactions in liquid growth assays. Cells were cultured overnight, subcultured to 5×10⁵ cells/ml and incubated at 30°C in SD-MAUC+MTA+RG. Growth was quantified hourly via OD₆₀₀ readings and percent growth (in %, y-axis) in MTDIA compared to vehicle control was determined at time points when control cells were at mid-log. Shown A and B are the mean growth differential values for three biological replicates; mean±s.d. values are available in Tables S1 and S5. The criterium for all strains shown was a statistically significant growth difference (either 25% growth inhibition or 30% growth improvement) in each MTDIA-treated xxxΔ compared to the untreated xxxΔ, calculated using two-tailed Student's t-test (P<0.05).

Fig. 5.

High-throughput microscopy identifies changes in protein localization and abundance in response to MTDIA and its genetic mimic of MEU1-deficiency. Representative fluorescence images of proteins involved in specific processes (as indicated) in untreated Y7092 cells, Y7092 cells treated with 100 nM MTDIA and untreated meu1Δ cells. Libraries of ∼4900 strains, each expressing a GFP-tagged protein and dual RFP-tagged nuclei/cytoplasm were cultured overnight on SD-U+R agar and inoculated into black-walled, clear-bottom 384-well plates to an OD₆₆₀ of 0.3 in 50 µl volumes of SD-MAU+MTA+RG liquid medium with or without MTDIA. Plates were incubated at 30°C for 6 h and the fluorescent signal was detected at 488 nm (GFP) and 561 nm (RFP). Numbers at top right of middle- and right-column images indicate the percent change in GFP abundance after treatment with MTDIA or in meu1Δ relative to untreated Y7092 cells. Changes in protein location were confirmed by visual inspection and validated in independent, reproducible experiments. Cells shown here are representative of three biological replicates each monitoring abundance and localization of 200 cells (n=600 cells).

Fig. 6.

MTDIA or MEU1-deficiency reduces cellular levels of ergosterol. (A) Representative fluorescence images of proteins (as indicated) integral to ergosterol biosynthesis untreated Y7092 cells, Y7092 cells treated with 100 nM MTDIA and untreated meu1Δ cells. Cells were grown, fluorescence intensity of GFP-tagged proteins and dual RFP-tagged nuclei/cytoplasm was quantified, and localization was assigned as previously described. Numbers at top right of middle- and right-column images indicate the percent change in GFP abundance after treatment with MTDIA or in meu1Δ relative to untreated Y7092 cells. Cells shown here are representative of three biological replicates each monitoring abundance and localization of 200 cells (n=600 cells). (B) Quantification of sterol intermediates in MTDIA-treated Y7092 or untreated meu1Δ cells by using gas chromatography-mass spectrometry (GC-MS). Lipids extracted from 5 OD units of cells grown in SD-MAU+MTA+RG were dried, derivatized, diluted in hexane and quantified using GC-MS. The identity of each sterol was confirmed using authentic standards. The percent (mean±s.d.) of each sterol relative to the total sterols, i.e. ergosterol/(squalene+lanosterol+ergosterol)×100 from six biological replicates is presented. *P<0.05, two-tailed Student's t-test comparison with untreated Y7092 cells.

Fig. 7.

Network analysis distinguishes autophagy and longevity as enriched pathways in the cellular response to MTDIA. (A) First-order minimum network analysis of genes/proteins sensitive to MTDIA/meu1Δ in growth and microscopy assays. The network was generated in NetworkAnalyst (https://www.networkanalyst.ca/) (Zhou et al., 2019) with the STRING interactome database, where nodes and edges represent genes/proteins and interactions, respectively. Nodes are colored by community analysis (i.e. each community is distinguished in a different color), representing tightly clustered subnetworks with more internal connections than randomly expected in the whole network (P<0.05). Nodes not assigned to a community are shown in gray. The most central nodes based on betweenness centrality (TOR1, RPL40A, ACT1) are labeled with their gene name. (B) Pathway enrichment analysis of each community was carried out by comparing the genes/proteins in each community to those in specific pathways in the KEGG database. The size of the bubble reflects the combined score of fold-enrichment and adjusted P-value. The color of each bubble is consistent with the communities identified in panel A.

Fig. 8.

MTDIA treatment and meu1Δ exacerbates the growth defect induced by rapamycin and decreases chronological lifespan. (A) Growth of Y7092 cells treated with rapamycin (RPM) and MTDIA as indicated relative to MTDIA treatment alone. Cells were grown at 30°C and growth was quantified at 48 h via OD₆₀₀ readings. Percent growth was determined via growth in the liquid from the growth in the MTDIA+RPM co-treatment relative to MTDIA or RPM alone [(RPM+MTA/MTDIA)/(RPM/DMSO)×100] from six biological replicates. *P<0.05, two-tailed Student's t-test comparison of treated and untreated cells. (B) Sensitivity of meu1Δ cells to RPM is enhanced compared to that of Y7092 cells. Cells were grown at 30°C and growth was quantified at 48 h via OD₆₀₀ readings. *P<0.05, two-tailed Student's t-test comparison of Y7092 and meu1Δ cells at each RPM concentration for six biological replicates. (C) Representative fluorescence images of MTDIA-treated Y7092 and untreated meu1Δ cells in a chronological lifespan assay. Cells were cultured in 50 ml of SD-MAUC+MTA+RG medium at 30°C for 10 days. At 2-day intervals, 1 ml of cells was stained with 2 µg/ml Phloxine B to distinguish dead cells and visualized at 60× magnification under DIC and GFP filter with a fluorescent microscope (Olympus BX63) and a digital camera (Olympus Dp70) (n=1000 cells). (D) Quantification of viability based on Phloxine B fluorescence at 2-day intervals. Percent viability [(unlabeled cells/GFP labelled cells+unlabelled cells)×100] and standard deviation were determined from 1000 cells across five visual fields for each condition. *P<0.05, two-tailed Student's t-test comparing MTDIA-treated Y7092 or untreated meu1Δ cells with untreated Y7092 cells at each time point. Error bars indicate the mean+s.d.

Fig. 9.

MTDIA treatment or meu1Δ induces autophagy independent of starvation. (A) Fluorescence images showing abundance and localization of GFP/RFP-tagged autophagy proteins in untreated Y7092 cells, Y7092 cells treated with 100 nM MTDIA and untreated meu1Δ cells. Cells were grown in SD-MAUC+MTA+RG medium for 6 h at 30°C, fluorescence was quantified, and localization was assigned as previously described. Numbers at top right of middle- and right-column images indicate the percent change in GFP abundance in Y7092 cells after treatment with MTDIA or in meu1Δ cells relative to untreated Y7092 cells. (B) Fluorescence images showing abundance and localization of Vps34-GFP/RFP in Y7092 or meu1Δ cells. Cells were grown in synthetic complete medium supplemented with 5′-methylthioadenosine (SC+MTA) for 6 h at 30°C, GFP fluorescence was quantified and protein localization assigned as described in Materials and Methods. Number inset in the right image indicates the percent changes in GFP abundance in meu1Δ compared to Y7092. (C,D) Representative images (C) and quantification of fluorescence intensity (D) of GFP-Atg8 in the cytosol or vacuole of MTA-treated Y7092, MTA/MTDIA-treated Y7092 or MTA-treated meu1Δ cells. NLS indicates staining of nuclei. Error bars indicate the mean±s.d. Cells shown in panels A-C are representative of three biological replicates and 600 cells.