Cancer cells must maintain lipid supplies for their proliferation and do so by upregulating lipogenic gene programs. The sterol regulatory element-binding proteins (SREBPs) act as modulators of lipid homeostasis by acting as transcriptional activators of genes required for fatty acid and cholesterol synthesis and uptake. SREBPs have been recognized as chemotherapeutic targets in multiple cancers, however it is not well understood which SREBP target genes are essential for tumorigenesis. Using parallel in vitro and in vivo CRISPR knockout screens, we identified terpenoid backbone biosynthesis genes as essential for pancreatic ductal adenocarcinoma (PDAC) tumor development. Specifically, we identified the non-sterol isoprenoid product of the mevalonate pathway, geranylgeranyl diphosphate (GGPP), as an essential lipid for tumor growth. Mechanistically, we observed that restricting mevalonate pathway activity using statins and SREBP inhibitors synergistically induced apoptosis and caused disruptions in small G protein prenylation that have pleiotropic effects on cellular signaling pathways. Finally, we demonstrated that geranylgeranyl diphosphate synthase 1 (GGPS1) knockdown significantly reduces tumor burden in an orthotopic xenograft mouse model. These findings indicate that PDAC tumors selectively require GGPP over other lipids such as cholesterol and fatty acids and that this is a targetable vulnerability of pancreatic cancer cells.

Sterol regulatory element-binding protein (SREBP) transcription factors are central regulators of lipid homeostasis and are essential for lipid metabolic reprogramming that supports tumor growth in multiple cancers. SREBP pathway inhibitors have been identified, but bioavailable compounds are lacking. To address this need, we designed a novel approach for screening a collection of 4,474 FDA-approved drugs. SREBPs are conditionally essential and required under low lipid conditions. Leveraging this property, we screened for drugs that inhibited pancreatic cancer cell growth in lipid-poor, but not lipid-rich, medium. The primary screen identified 83 drugs that inhibited cell growth in a lipid-dependent manner. Secondary assays examining SREBP target gene expression, SREBP proteolytic cleavage, and effects on human breast cancer cells identified 13 FDA-approved drugs that inhibit SREBP pathway activation. Taken together, we demonstrated that our screening approach can identify SREBP inhibitors from a small library of compounds. This high-throughput screening platform enables screening of large compound collections to discover novel small molecule SREBP inhibitors.

SCAP supports PDAC development and loss of SCAP prolongs survival in the KPC mouse model of pancreas cancer. A, Mice of the indicated genotype were aged until death. The figure shows the cumulative incidence curve of death due to PDAC for each genotype group. For the analysis, we considered death due to other causes (i.e., not due to PDAC) as a competing event. The Gray test was used for pairwise comparisons. B, Incidence of adverse event (death due to PDAC) in KPC cohorts from A either pooled (left figure) or stratified by sex (right figure). C, Incidence of PanINs and invasive PDAC identified histologically in KPC cohorts from A. D, Representative hematoxylin and eosin–stained sections of formalin-fixed pancreas tissue from KPC, KPCSfl/+ and KPCSfl/fl cohorts in A showing a magnification (2–4×) image (top row) and a higher magnification (20×) image (bottom row) of the changes in the pancreas. The C mouse is used as a control. E, Figure shows the survival curves by genotype group. Mice of the indicated genotypes were aged until death at which time histology was performed to determine the presence of PanINs and invasive PDAC. For this analysis, we treated time to invasive PDAC as interval-censored data. The permutation test was used for pairwise comparisons.

Table S1. Primer sequences used for quantitative PCR

SUPPLEMENTARY MATERIALS AND METHODS

FIGURE S2 – Mouse genotyping methods and survival study data

FIGURE S6 – Site-1 protease activation of the SREBP pathway is required for PDAC cell growth in low serum conditions.

Table S3. Detailed cause of death and observed lesions in KPC mouse cohorts

Gemcitabine (dFdC) and emtricitabine (FTC) are first -line drugs that are used for the treatment of pancreatic cancer and human immunodeficiency virus, respectively. The above drugs must undergo sequential phosphorylation to become pharmacologically active. Interindividual variability associated with the responses of the above drugs has been reported. The molecular mechanisms underlying the observed variability are yet to be elucidated. Although this could be multifactorial, nucleotidases may be involved in the dephosphorylation of drug metabolites due to their structural similarity to endogenous nucleosides. With these in mind, we performed in vitro assays using recombinant nucleotidases to assess their enzymatic activities toward the metabolites of dFdC and FTC. From the above in vitro experiments, we noticed the dephosphorylation of dFdC-monophosphate in the presence of two 5 ' -nucleotidases (5 ' -NTs), cytosolic 5 ' -nucleotidase IA (NT5C1A) and cytosolic 5 ' -nucleotidase III (NT5C3), individually. Interestingly, FTC monophosphate was dephosphorylated only in the presence of NT5C3 enzyme. Additionally, nucleoside triphosphate diphosphohydrolase 1 (NTPDase 1) exhibited enzymatic activity toward both triphosphate metabolites of dFdC and FTC. Enzyme kinetic analysis further revealed Michaelis-Menten kinetics for both NT5C3-mediated dephosphorylation of monophosphate metabolites, as well as NTPDase 1-mediated dephosphorylation of triphosphate metabolites. Immunoblotting results confirmed the presence of NT5C3 and NTPDase 1 in both pancreatic and colorectal tissue that are target sites for dFdC and FTC treatment, respectively. Furthermore, sex -specific expression patterns of NT5C3 and NTPDase 1 were determined using mass spectrometry- based proteomics approach. Based on the above results, NT5C3 and NTPDase 1 may function in the control of the levels of dFdC and FTC metabolites. SIGNIFICANCE STATEMENT Emtricitabine and gemcitabine are commonly used drugs for the treatment of human immunodeficiency virus and pancreatic cancer. To become pharmacologically active, both the above drugs must be phosphorylated. The variability in the responses of the above drugs can lead to poor clinical outcomes. Although the sources of drug metabolite concentration variability are multifactorial, it is vital to understand the role of nucleotidases in the tissue disposition of the above drug metabolites due to their structural similarities to endogenous nucleosides.

SREBP pathway is broadly required for cancer cell growth and is activated in human PDAC tumors. A–C, The essentiality of SCAP, SREBF1, and SREBF2 across cancer cell lines was examined using the Cancer Dependency Map project database (Public Chronos 23Q4). The mean Chronos dependency scores are shown for multiple organ systems, including the pancreas (red). Negative scores indicate gene essentiality. D, Expression of SREBP target genes in PDAC tumor tissue compared with normal tissue is shown from two Oncomine data sets Pei and Badea (33). Blue, lower gene expression; red, higher gene expression; black, P > 0.05. Genes are grouped and color-coded by bioinformatic process (https://www.genepattern.org/#gsc.tab=0). E, The association of SREBP target genes with survival from patients with PDAC was queried using the Kaplan–Meier Plotter tool (https://kmplot.com/analysis/) and RNA sequencing datasets (n = 177 patients for each curve).

SREBP pathway activation is required for PDAC cell growth in low-serum conditions. A, PDAC cells (Pa03c and Pa16c) were cultured in either 10% or 1% FBS with indicated concentrations of 25-hydroxycholesterol (25-HC) for 72 hours, and cell growth was determined using an MTS assay with data normalized to 10% FBS untreated cells. Data are representative of two biological replicates with three technical replicates for each biological replicate. Error bar denotes SD. B, Immunoblot analysis of nuclear extracts from human Pa03c and Pa16c cells cultured in either 10% FBS, 1% FBS, or 1% FBS-containing 25-hydroxycholesterol (2.5 μmol/L) for 16 hours. Blots were probed for either SREBP1-N or SREBP2-N (arrow), and lysine-specific histone demethylase 1A (KDM1A) served as a loading control. The result is representative of two biological replicates. C, Pa03c and Pa16c cells were cultured under the same conditions as in B, and quantitative RT-PCR was performed for target genes of SREBP1 (SCD, FASN, and INSIG1) and SREBP2 (HMGCR, HMGCS1, and LDLR). GAPDH served as a control. Data are representative of two biological replicates with three technical replicates for each biological replicate. Error bar denotes standard deviation. Statistical significance was determined using one-way ANOVA and Tukey HSD test. P values are indicated: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not significant. D, Immunoblot analysis of whole cell lysates from Pa03c and Pa16c cells cultured as in B for indicated SREBP target protein expression. The asterisk indicates a nonspecific band present in Pa03c and Pa16c cell lines. Actin served as a loading control. The result is representative of two biological replicates.

Site-1 protease activation of the SREBP pathway is required for PDAC cell growth in low-serum conditions. A, PDAC cells (Pa03c and Pa16c) were cultured in either 10% or 1% FBS with indicated concentrations of the Site-1 protease inhibitor PF-429242 for 72 hours, and cell growth was determined using an MTS assay with data normalized to 10% FBS untreated cells. Data are representative of two biological replicates with three technical replicates for each biological replicate. Error bar denotes SD. B, Immunoblot analysis of nuclear extracts from human Pa03c and Pa16c cells cultured for 16 hours in either 10% FBS, 1% FBS, or 1% FBS-containing PF-429242 (10 μmol/L). Blots were probed for either SREBP1-N or SREBP2-N (arrow), and lysine-specific histone demethylase 1A (KDM1A) served as a loading control. The result is representative of two biological replicates. C, Pa03c and Pa16c cells were cultured under the same conditions as in B, and quantitative real-time PCR was performed for target genes of SREBP1 (SCD, FASN, and INSIG1) and SREBP2 (HMGCR, HMGCS1, and LDLR). GAPDH served as a control. Data are representative of two biological replicates with three technical replicates for each biological replicate. Error bar denotes standard deviation. Statistical significance was determined using one-way ANOVA and Tukey honestly significant difference (HSD) test. P values are indicated: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not significant. D, Immunoblot analysis of whole cell lysates from Pa03c and Pa16c cells cultured as in B for indicated SREBP target protein expression. The asterisk indicates a nonspecific band present in Pa03c and Pa16c cell lines. Actin served as a loading control. The result is representative of two biological replicates.

FIGURE S5 – SCAP is required for PDAC cell growth and survival in low serum conditions.

Inhibition of both SREBP1 and SREBP2 is required to prevent PDAC cell and tumor growth. A, Immunoblot of Pa03c wild-type, SREBF1 KO, SREBF2 KO, SREBF1/2 double KO cells for indicated antibodies. Membrane-enriched extracts (20 μg) were harvested and probed for SREBP1 and SREBP2. Calnexin served as a loading control. B, Wild-type (WT) Pa03c, SREBF1 KO, SREBF2 KO, SREBF1/2 double KO, and MBTPS1 KO cells were cultured in either 10% FBS or 1% FBS for 7 days. Media were changed every 3 days. Plates were stained with crystal violet and quantification is shown (n = 3 per group). For each cell line, growth was normalized to the 10% FBS condition. Statistical significance was determined using two-way ANOVA and Tukey HSD test. P values are indicated: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001, ns, not significant. Error bar denotes SD. C, Nude mice were subcutaneously injected with 1 × 106 Pa03c cells of the indicted genotype in both flanks (two tumors per mouse). Once visible, tumors were measured, and volume calculated. Each group contained five mice. Error bar denotes SD. D, Individual tumor volumes at day 25, n = 10 tumors per group. Statistical significance was determined using one-way ANOVA and Dunnett test. Error bar denotes SD.

FIGURE S4 – SCAP is required for human PDAC tumor growth in mouse orthotopic xenograft models

Solid tumors undergo metabolic reprogramming when growth outstrips local nutrient supply. Lipids such as cholesterol and fatty acids are required for continued tumor cell proliferation, and oncogenic mutations stimulate de novo lipogenesis to support tumor growth. Sterol regulatory element-binding protein (SREBP) transcription factors control lipid homeostasis by activating genes required for lipid synthesis and uptake. SREBPs have been implicated in the progression of brain, breast, colon, liver, and prostate cancers. However, the role of the SREBP pathway and its central regulator SREBP cleavage activating protein (SCAP) in pancreatic ductal adenocarcinoma (PDAC) has not been studied in detail. Here, we demonstrated that pancreas-specific knockout of Scap has no effect on mouse pancreas development or function, allowing for examination of the role of Scap in the murine KPC model of PDAC. Notably, heterozygous loss of Scap prolonged survival in KPC mice, and homozygous loss of Scap impaired PDAC tumor progression. Using xenograft models, we showed that SCAP is required for human PDAC tumor growth. Mechanistically, chemical or genetic inhibition of the SREBP pathway prevented PDAC cell growth under low-serum conditions because of a lack of lipid supply. Highlighting its clinical importance, the SREBP pathway is broadly required across cancer cell lines, target genes are upregulated in human PDAC tumors, and increased expression of SREBP targets is associated with poor survival in patients with PDAC. Collectively, these results demonstrate that SCAP and SREBP pathway activity are required for PDAC cell and tumor growth, identifying SCAP as a potential therapeutic target for PDAC. SIGNIFICANCE:Our findings demonstrate that SREBP pathway activation is a critical part of the metabolic reprogramming that occurs in PDAC development and progression. Therefore, targeting the SREBP pathway has significant therapeutic potential.

Table S4. Pa03c cell microarray gene expression data

SCAP is required for human PDAC tumor growth in mouse subcutaneous xenograft models. A, Nude mice were subcutaneously injected with 5 × 105 Pa03c cells in both flanks (two tumors per mouse). Once visible, tumors were measured, and volume calculated. Each group contained six mice. Error bar denotes SD. Statistical significance was determined using the Student t test. P values are indicated: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not significant. B, Individual tumor volumes at day 22, n = 12 tumors per group. Statistical significance was determined using the Student t test. C, Representative hematoxylin and eosin–stained sections of formalin-fixed tumor tissue from the mice in A showing a low magnification (4×) image (top row) and a higher magnification (20×) image (bottom row) of tumor sections. D, Nude mice were subcutaneously injected with 1 × 106 Pa16c cells in both flanks (two tumors per mouse) as in A–C. Each group contained six mice. Error bar denotes SD. Statistical significance was determined using the Student t test. P values are indicated: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not significant. E, Individual tumor volumes at day 61, n = 12 tumors per group. Statistical significance was determined using the Student t test. F, Representative hematoxylin and eosin–stained sections of formalin-fixed tumor tissue from the mice in D showing a low magnification (2–4×) image (top row) and a higher magnification (20–40×) image (bottom row) of tumor sections. WT, wild-type.

FIGURE S1 – Scap is not required for development or function of the mouse pancreas.