A Compound That Inhibits Glycolysis in Prostate Cancer Controls Growth of Advanced Prostate Cancer

Abstract

Metastatic castration-resistant prostate cancer remains incurable regardless of recent therapeutic advances. Prostate cancer tumors display highly glycolytic phenotypes as the cancer progresses. Nonspecific inhibitors of glycolysis have not been utilized successfully for chemotherapy, because of their penchant to cause systemic toxicity. This study reports the preclinical activity, safety, and pharmacokinetics of a novel small-molecule preclinical candidate, BKIDC-1553, with antiglycolytic activity. We tested a large battery of prostate cancer cell lines for inhibition of cell proliferation, in vitro. Cell-cycle, metabolic, and enzymatic assays were used to demonstrate their mechanism of action. A human patient-derived xenograft model implanted in mice and a human organoid were studied for sensitivity to our BKIDC preclinical candidate. A battery of pharmacokinetic experiments, absorption, distribution, metabolism, and excretion experiments, and in vitro and in vivo toxicology experiments were carried out to assess readiness for clinical trials. We demonstrate a new class of small-molecule inhibitors where antiglycolytic activity in prostate cancer cell lines is mediated through inhibition of hexokinase 2. These compounds display selective growth inhibition across multiple prostate cancer models. We describe a lead BKIDC-1553 that demonstrates promising activity in a preclinical xenograft model of advanced prostate cancer, equivalent to that of enzalutamide. BKIDC-1553 demonstrates safety and pharmacologic properties consistent with a compound that can be taken into human studies with expectations of a good safety margin and predicted dosing for efficacy. This work supports testing BKIDC-1553 and its derivatives in clinical trials for patients with advanced prostate cancer.

Figure 1.. BKIDC effects on Prostate Cancer Cell lines and other Cell Lines.

(A) Cell proliferation was evaluated by MTS assay at 72 h after BKIDC-1553 was added at the range between 0.63–20 µM, except for VCaP at 120 h (DMSO as a vehicle control). The data shown are representative of at least two independent experiments, normalized to the values obtained from cells grown in medium containing 0.1% DMSO vehicle and presented as mean ± standard deviation (SD) (n=3–5). One-way ANOVA followed by Dunnett’s test for multiple comparison: **p <0.01. (B) Structures of representative BKIDCs. Lead BKIDCs have two main scaffolds: Pyrrolopyrimidines (PrP) and pyrazolopyrimidines (PP) are two main scaffolds of BKIDCs. Toxoplasma gondii calcium dependent protein kinase 1 (TgCDPK1) has the small gatekeeper residue glycine and can hold BKIDC-1553 in the ATP binding pocket. BKIDC-1553-N-Me, 1820, and GP228 have distinct R2 groups while commonly lacking the protein kinase binding and thus inhibition activity. (C) Graphical presentation of screening results of over 123 BKIDC compounds (Table S3) tested at 10 μM in LNCaP95 and PC3 (Table S4). Each dot represents one compound and the data are shown as mean ± SD. Compounds were categorized into three groups depending on the degree of inhibition of proliferation in LNCaP95 and PC3.

Figure 2.. Cytostatic effects of anti-proliferative BKIDCs.

(A) The percentage of cells at G1, S, and G2 phase of cell cycle at 30 h after treatment with the indicated BKIDCs. Cell cycle analysis showing G1 cell cycle arrest in LNCaP and LNCaP95 cells compared to PC3 when treated with BKIDC-1553 and 1553-N-Me. These results were confirmed in 2 subsequent independent experiments (Figure S5) (B) Heat map of antigen levels detected by RPPA. Cellular lysates of LNCaP95 treated with the indicated BKIDC at 20 µM for 24 h were subjected to RPPA with 70 distinct antibodies (Antibodies and the corresponding RPPA signal intensities relative to β-actin signals are listed in Table S5). Note BKIDC-1553 and 1553-N-Me commonly affected only abundance of ribosomal S6 protein (RPS6) phosphorylated at S235/S236 or S240/S244, the first two bands seen with BKIDC-1553 and the only bands seen with BKIDC-1553-N-Me. The response to BKIDC-1817 was comparable to the one to DMSO. (C) Time course and dose dependent studies of BKIDCs in LNCaP95 and PC3. Time- and dose-dependent decrease in phosphorylated RPS6 signals in RPPA in LNCaP95 but not PC3 as early as at 4 h after treatment with BKIDC-1553 and 1553-N-Me.

Figure 3.. BKIDCs are antiglycolytic agents.

(A) AMPK activation in LNCaP95 treated with BKIDC-1553-N-Me. AMPK activity in total LNCaP95 lysates was measured after 1 h exposure to BKIDC-1553-N-Me or 1817. One-way ANOVA followed by Tukey’s test for multiple comparison: **p <0.01. (B) Phosphorylated acetyl-CoA carboxylase (P-ACC) at Ser79 was increased after treatment with BKIDC-1553 but not 1817. Metabolic inhibitors 2-deoxyglucose (2-DG) and oligomycin (oligo) served as positive controls for AMPK activation to increase P-ACC. (C) Effects of BKIDCs on intracellular ATP levels in LNCaP and LNCaP95 cells. Cells were treated with BKIDC-1553-N-Me, glycolysis inhibitor 2-DG, and OXPHOS inhibitors (OXPHOSi) singly or in combination for 120 min. The data shown are representative of at least two independent experiments, normalized to the values obtained from cells grown in medium containing 0.1% DMSO vehicle and presented as mean ± SD (n=2). Rot: rotenone, FCCP: carbonyl cyanite-4 (trifluoromethoxy) phenylhydrazone, AA: Antimycin A. One-way ANOVA followed by Dunnett’s test for multiple comparison (vs. Oligomycin alone). (D) Extracellular acidification rates (ECAR) were determined in cultured cells by the Seahorse XFe24 analyzer. Cells were pre-treated with 20 µM BKIDC for 1 h prior to application to Seahorse analyzer. Increase in ECAR by glucose injection (glycolytic rates) was inhibited by BKIDC-1553-N-Me in LNCaP95 but not PC3. Maximum glycolytic capacity was evaluated by oligomycin injection. 2-DG injection shuts down glycolysis as reflected by return of ECAR to the basal levels. (E) The effect of BKIDC-1553-N-Me on glycolytic rates. The data shown are normalized to the values obtained from cells pre-treated with DMSO vehicle and presented as mean ± SD (n=3). One-way ANOVA followed by Tukey’s test for multiple comparison: **p <0.01. (F) Cells were exposed to BKIDC at the indicated time by injection at the final concentration of 10 µM. Decrease in ECAR was observed in LNCaP95 but not PC3 immediately after injection of BKIDC-1553-N-Me. There was a compensatory increase in oxygen consumption rates (OCR) which reflects mitochondrial oxidative respiration.

Figure 4.. Glucose phosphorylation is the primary target for BKIDC as an antiglycolytic agent.

(A) Schematic flowchart of glycolysis and the metabolites measured in this metabolic flux assay with ¹³C-labeled glucose with BKIDC-1553-N-Me this experiment (dark boxes). (B) Metabolic flux assay with ¹³C-labeled glucose. ¹³C enrichment of glycolytic intermediates was determined by liquid chromatography coupled with mass spectrometry (LC-MS) in DU145, LNCaP, and LNCaP95 cells at 30 min after exposure to [U-¹³C]-glucose in the presence and absence of BKIDC-1553-N-Me. The X-axis is the concentration of BKIDC-1553-N-Me, with 0–20 μM added to the indicated cell lines. Black columns indicate where unlabeled carbon (¹²C, thus M+0 [M=mass expected for ¹²C]) is found and grey columns originate from the ¹³C-labeled glucose (¹³C, thus M+6 or M+3). In LNCaP and LNCaP95 cells, addition of BKIDC-1553-N-Me inhibited glucose being metabolized to 3-carbon glycolytic intermediates, DHAP, PG, PEP, or pyruvate, indicating a block in glycolysis, at or prior to the six-carbon to 3-carbon fructose-bisphosphate aldolase step. Note that in BKIDC non-responsive DU145 cells there is no BKIDC-1553-N-Me effect on hexose-6-phosphate and 3-carbon glycolytic intermediates (DHAP, PEP, PG, or pyruvate). The results trended to indicate that BKIDC blocked effects at hexose-6-phosphate levels. One-way ANOVA followed by Dunnett’s test for multiple comparison for M +3 and M + 6 isotopomers: *p <0.05, **p <0.01, ***p <0.001, ****p <0.0001. (C) To distinguish between glucose uptake and phosphorylation steps, two different radiolabeled glucose analogs were used for radiometric assays in LNCaP and DU145 cells. Both 2-DG and 3-O-methylglucose (3-OMG) are taken up into cells in the same manner as glucose. Unlike 2-DG which is trapped by hexokinase (HK)-mediated phosphorylation and accumulated in the cells as a form of 2-DG-6P, 3-OMG is not phosphorylated, and its intracellular levels reflect the cells’ capability of Glut1-mediated glucose uptake itself. (D) 2-DG 2-[1,2-³H(N)] accumulation was inhibited by the presence of BKIDC-1553-N-Me at 20 μM in LNCaP but not DU145 cell lines. 3-O-[methyl-³H] D-glucose uptake was not affected by BKIDC-1553-N-Me in LNCaP. These data support the conclusion that BKIDC block phosphorylation but not uptake of glucose in BKIDC-susceptible cell line LNCaP cell line. Two-tailed t tests: ****p <0.0001

Figure 5.. The effect of HK deficiency on antiglycolytic actions of BKIDC-1553-N-Me.

(A) CRISPR-Cas9 system was used to generate HK1 and HK2 knock-out (KO) LNCaP clones and two independently derived KO clones for HK1 and HK2 were tested. Intracellular ATP levels, in the presence of a mitochondrial respiration blocker oligomycin (Oligo), were used as a proxy to monitor the activity of glycolysis to generate ATP. 2-Deoxyglucose (2-DG) and sodium oxamate (SO) were employed to block glycolysis at different steps. SO is an inhibitor of lactate dehydrogenase thus blocking the final step of glycolysis conversion of pyruvate into lactate. Note HK2 KO clones displayed remarkable resistance to decreasing ATP when co-treated with Oligo and BKIDC-1553-N-Me while they were sensitive to 2-DG and SO. One-way ANOVA followed by Dunnett’s test for multiple comparison (vs. Oligomycin alone). (B) Western blot validation of HK1 and HK2 deficiency in LNCaP KO clones. GAPDH served as a loading control.

Figure 6.. Preclinical anti-prostate cancer activity of BKIDC-1553

(A) Growth Curves of Prostate Cancer Xenografts Implanted in SCID Mice. The BKIDC-resistant human prostate cancer cell line PC3 and a human patient-derived xenograft (PDX) model (LuCaP 35) were injected subcutaneously in SCID mice. Mice were treated for 4–5 weeks with 20 mg/kg BKIDC-1553 po three times a week. The LuCaP 35 tumors responded to BKIDC-1553 to a similar degree as enzalutamide (Enza) and no decrease was seen in PC3 xenografts. The data are presented as mean ± standard error of mean (SEM) (n=7–10). One-way ANOVA followed by Tukey’s test for multiple comparison (LuCaP35) and two-tailed t tests (PC3): *p <0.05, **p <0.01, ns: not significant. (B) Immunohistochemical analysis of P-ACC in LuCaP 35 tumors. Representative images of LuCaP 35 tumors stained with an anti-P-ACC antibody are shown (at 20x magnification). Box and whisker plots show scoring analysis of anti-PACC immunoreactivities in tumors in the specified treatment. One-way ANOVA followed by Dunnett’s test for multiple comparison (vs. Control: *p <0.05). The number of animals used for IHC study was: 6 for control, 8 for 1553, 6 for enzalutamide. IHC score is defined in Materials and Methods. (C) Dose response of LTL331R line (patient-derived in vitro model of neuroendocrine prostate cancer) to BKIDC-1553 and enzalutamide using CellTiter-Glo^® 3D Cell Viability Assay. The data shown are normalized to the values obtained from cells treated with DMSO vehicle and presented as mean ± SD (n=3). One-way ANOVA followed by Dunnett’s test for multiple comparison (vs. Control).