Aberrant FGFR Tyrosine Kinase Signaling Enhances the Warburg Effect by Reprogramming LDH Isoform Expression and Activity in Prostate Cancer

Metabolic reprograming from oxidative phosphorylation to aerobic glycolysis is a common event in cancer progression. Although glycolysis is less efficient than oxidative phosphorylation for providing energy with respect to the number of ATP per glucose, it meets the demand of rapidly growing cancer cells for building blocks. In addition, increased glycolysis results in glucose deprivation and lactate accumulation in the tumor microenvironment, which suppresses lymphocyte infiltration and compromises anti-immunotherapies (1, 2). In line with those effects, the glycolytic phenotype is associated with prostate cancer progression and aggressiveness (3–10). An understanding of how tumor cells reprogram their metabolism from oxidative phosphorylation to aerobic glycolysis may provide a novel approach to selectively inhibit aerobic glycolysis in tumor cells.

The prostate consists of epithelial and stromal compartments, which maintain active two-way communication through paracrine mechanisms, including fibroblast growth factor (FGF) signaling in which FGF and FGF receptor (FGFR) are partitioned between the two compartments (11). That precisely balanced communication is critical for preserving the tissue homeostasis and function of the prostate. The FGF family consists of 18 ligands that exert their regulatory signals by activating the FGFR tyrosine kinases encoded by 4 homologous genes. FGF and FGFR are expressed throughout the body in a pattern that is spatiotemporally and cell-type specific, controlling embryonic development and maintaining adult tissue homeostasis and function.

There is extensive evidence that ectopic activation of the FGF/FGFR signaling axis is associated with prostate cancer development and progression (11). The acquisition of ectopic FGFR1 expression stands out as the most remarkable change among the FGFR isotypes in prostate cancer. The forced expressions of FGF and FGFR have been shown to induce prostate lesions in mouse models. On the other hand, the ablation of Fgfr1 or Frs2α, which encodes FGFR-substrate 2α (FRS2α), an adaptor protein needed for FGFR kinases to activate ERK and PI3K/AKT pathways, significantly reduced prostate cancer development and progression in mice (12, 13).

FGF signaling promotes aerobic glycolysis by increasing hexokinase 2 (HK2) expression and the tyrosine phosphorylation of multiple enzymes involved in aerobic glycolysis (14). Whether and how ectopic FGF signaling contributes to prostate cancer progression by promoting aerobic glycolysis, remains to be determined. The last step of glycolysis is the reduction of pyruvate to lactate, a reversible conversion catalyzed by lactate dehydrogenase (LDH). LDH is a tetrameric enzyme composed of two types of subunits, LDHA and LDHB. The combination of the two subunits yields five isozymes, which catalyze the conversion between pyruvate and lactate. LDH1 is composed of four LDHB subunits and favors the conversion from lactate to pyruvate, allowing oxidation along the pathway of the tricarboxylic acid cycle. LDH5 is composed of four LDHA subunits and favors the conversion from pyruvate to lactate, allowing the glycolytic pathway to be completed at the formation of lactate (15). Interestingly, hypoxia induces LDHA, while inhibits LDHB, expression (16, 17). Emerging evidence shows that LDHA is required for the survival and proliferation of cancer cells. Although still a matter of debate, current data seem to indicate an association between reduced levels of LDHB and increased malignancy in prostate cancer and other cancers (18–27).

LDHA is tyrosine phosphorylated in cancer cells (28). FGFR1 has been reported to phosphorylate LDHA at multiple tyrosine residues, which enhances tetramer formation and NADH binding and thus the enzymatic activity of LDHA (29). In this study, we demonstrate that FGFR1 signaling promotes aerobic glycolysis by stabilizing LDHA through tyrosine phosphorylation and downregulating LDHB expression through the induction of hypermethylation in the Ldhb promoter. We found that LDHA ablation compromised, whereas LDHB ablation enhanced, the tumorigenicity of DU145 cells. Furthermore, high levels of phosphorylated LDHA and low levels of LDHB in human prostate cancer tissues were associated with short biochemical recurrence and survival times in patients with prostate cancer. Together, our results suggest that ectopically expressed FGFR1 in prostate cancer induces metabolic changes by increasing LDHA levels and lowering LDHB levels, which promotes prostate cancer growth and progression.

Mice were housed under the Program of Animal Resources of the Institute of Biosciences and Technology in accordance with the principles and procedures of the Guide for the Care and Use of Laboratory Animals. All experimental procedures were approved by the Institutional Animal Care and Use Committee. Mice were bred and genotyped as described previously (12, 30, 31). For xenograft studies, 2 × 10⁶ DU145 cells were mixed with Matrigel (BD Biosciences) at a 1:1 ratio and subcutaneously injected into the flanks of nude mice (6-week old). The size of xenograft was measured with a caliper and calculated as V = 0.52 × length × width². The xenografts were harvested after the animals were euthanized via CO₂ suffocation.

Prostate tissues and xenografts were fixed, embedded, sectioned as described (32). For immunostaining, the antigens were retrieved by boiling in citrate buffer (10 mmol/L, pH 8.0) for 20 minutes. Rabbit antiphosphorylated LDHA (pLDHA, 1:200) and anti-LDHA (1:200) were purchased from Cell Signaling Technology. Mouse anti-LDHB (1:200), rabbit anti-CD31 (1:200) and anti-Ki67 (1:200), and rat anti-F4/80 (1:200) were purchased from Abcam. A Fluoremetric TUNEL Assay Kit was purchased from Premega Co. The ExtraAvidin Peroxidase System (Sigma Aldrich) and fluorescence-conjugated secondary antibodies (Invitrogen) were used to visualize specifically bound antibodies. For immunofluorescence staining, the nuclei were counterstained with To-Pro 3 before being observed under a confocal microscope (Zeiss LSM 510).

The Massachusetts General Hospital (Boston, MA) prostate cancer tissue microarray (TMA) was used to assess the expression of LDHA, LDHB, and Fgfr1 as well as the level of phosphorylated LDHA in human prostate cancer samples as described previously (33). Immunostaining of prostate cancer cells and that of stromal cells was evaluated separately. The percentage of positive cells was calculated and categorized as follows: 0, 0%; 1, 1%–10%; 2, 11%–50%; 3, 50%–75%; and 4, 75%–100%. The staining intensity was visually scored and defined as follows: 0, negative; 1, weak; 2, moderate; and 3, strong. Final immunoreactivity scores (IRS) were calculated for each case by multiplying the percentage and intensity scores.

Cells or xenografts were homogenized in RIPA buffer as described previously (32). Samples containing 30 μg protein were separated by SDS-PAGE and blotted onto polyvinylidene difluoride membranes. The antiphosphorylated ERK1/2 (1:1,000), antiphosphorylated AKT (1:1,000), anti-ERK1/2 (1:1,000), anti-AKT (1:1,000), antiphosphorylated FRS2α (1:1,000), and anti-HA (1:1,000) antibodies were purchased from Santa Cruz Biotechnology. Anti-pLDHA (1:1,000), anti-LDHB (1:1,000), and anti-LDHA 1:1,000) antibodies and the Glycolysis Antibody Sampler Kit containing antibodies against HK1, PFKP, PFKP2, PFKP3, aldolase, PGAM, PKM1/2, and pyruvate dehydrogenase were purchased from Cell Signaling Technology. The antiphosphotyrosine 4G10 (1:1,000), anti-TET1 (1:1,000), and anti-Flag (1:1,000) antibodies were purchased from Millipore Sigma. The specifically bound antibodies were visualized using the ECL-Plus chemiluminescent reagents. The films were scanned with a densitometer for quantitation.

Total RNA was isolated from cells and tissues using the Ribopure RNA isolation reagent (Ambion), reverse transcripted with SuperScript III (Life Technologies) and random primers, and analyzed with real-time PCR using the Fast SYBR Green Master Mix (Life Technologies) as described. The results were expressed as the mean ± SD as described (32). The primer sequences are listed in Supplementary Table S1.

The lentivirus-based CRISPR-Cas9 system was used to ablate the Ldha, Ldhb, and Fgfr1 alleles in DU145 cells. The sequences of the sgRNAs are shown in Supplementary Table S1. Two days after infection with the lentivirus, the recombinant cells were selected via growth in a medium containing 2 μg/mL puromycin.

The QuikChange Lightning Multi Site-Directed Mutagenesis Kit (Agilent Technologies) was used to generate LDHA-mutant cDNAs. The primer sequences are listed in Supplementary Table S1.

Cells (2 × 10⁷) were suspended in 450 μL methanol/chloroform (2:1) and lysed by ultrasound. The lysates were mixed with 450 μL chloroform/H₂O. The supernatants were lyophilized and then resuspended in 500-μL D₂O containing 0.25 mmol/L sodium trimethylsily 1 propionate-d4. All NMR spectra were recorded on a Bruker AVANCE III 600 MHz NMR spectrometer.

Cells were isolated with the Qiagen Kit and ultrasound sheared into smaller fragments (200–600 bp). The sheared gDNA was denatured by incubation in 1 mol/L NaOH/25 mmol/L EDTA at 95°C for 12 minutes, and then immunoprecipitated with anti-5-methylated cytosine (5mC) antibodies. The bound 5mC-containing DNA was then purified and used to make the sequencing library with the NEB (E6240S) Library Prep Kit, which was subsequently sequenced by the Agrelife Genomics and Bioinformatics Service, Texas A&M University (College Station, TX).

Genomic DNA was isolated using the Qiangen Genomic DNA Kit, followed by bisulfite conversion using the EpiJET Bisulfite Conversion Kit (Thermo Scientific). The bisulfite-specific primers for the PCR amplification were listed in Supplementary Table S1. The PCR products were cloned into pDrive Cloning Vectors (Qiangen) and sequenced.

Stable MEF transfectants expressing HA-tagged LDHA were treated with cycloheximide for the indicated times. The abundance of LDHA was examined by Western blot analysis. The relative level of endogenous or HA-tagged LDHA was quantitated using ImageJ. The GraphPad software was used to compare the slopes of each curve.

Statistical analysis was performed using the two-tailed t test. For protein stability assay, the GraphPad software was used to compare two curves. P < 0.05 is considered statistical significant. Error bars indicate SDs.

We first determined whether FGF signaling regulated cell metabolism in mouse embryonic fibroblasts (MEF), which expressed FGFR1, FGFR2, and FRS2α. We generated MEFs that were devoid of Fgfr1, Fgfr2, and Frs2α by infecting MEFs bearing floxed Fgfr1, Fgfr2, and Frs2α alleles with adenovirus carrying the Cre-GFP coding sequence. The cells, designated MEF^ΔF, did not express Fgfr1, Fgfr2, and Frs2α mRNA (Fig. 1A). Interestingly, real-time RT-PCR also revealed that at the mRNA level, MEF^ΔF increased Ldhb, but not Ldha expression. However, Western blot analyses revealed that at the protein level, compared with parental MEFs, the MEF^ΔF cells had reduced LDHA expression and increased LDHB expression (Fig. 1B). Those results suggest that FGF signaling promotes LDHA expression at the protein level and suppresses LDHB expression at the mRNA level. Western blot analysis also showed that FGF2 failed to induce phosphorylation of FRS2α, ERK, and AKT, indicating successful abrogation of FGF signaling in the cells.

LDHA and LDHB favor opposite direction of the conversation between pyruvate and lactate (15). To determine whether abrogation of the FGF signaling axis in MEFs changes the cell metabolism, we compared the lactate production and oxygen consumption in MEFs and MEF^ΔF cells (Fig. 1C). The abrogation of FGF signaling reduced lactate production and increased the oxygen consumption rate. In contrast, overexpression of LDHA in MEF^ΔF cells significantly increased lactate production (Supplementary Fig. S1). In addition, NMR analyses revealed that the abrogation of FGF signaling reduced the concentrations of isoleucine, valine, lactate, and acetate and increased the concentrations of glutamate and succinate (Fig. 1D). The reduced isoleucine and valine levels in MEF^ΔF cells suggest that the synthesis of branched chain amino acids (BCAA) was compromised. The dysregulation of BCAA synthesis in MEFs lacking FGF signaling is in line with evidence that BCAAs promote glucose uptake (34). Those results demonstrate a shift in energy metabolism from aerobic glycolysis to oxidative phosphorylation in MEF^ΔF cells, suggesting that energy metabolism in MEFs is regulated by FGF signaling.

To determine the role of FGFR1 in the regulation of LDHA at the protein level, we treated the MEFs and MEF^ΔF cells with cycloheximine to block protein synthesis. Western blot analyses revealed that the abundance of LDHA declined faster in MEF^ΔF cells than in the parental MEFs (Fig. 2A). Those results indicate that the half-life of LDHA is shorter in MEF^ΔF cells than in the parental MEFs, suggesting that FGF signaling enhances the stability of LDHA.

LDHA consists of four tyrosine phosphorylation sites (Fig. 2B). In line with a previous report that FGFR1 directly phosphorylates LDHA (29), Western blot analyses with the antiphosphotyrosine antibody 4G10 showed that LDHA was tyrosine phosphorylated (Fig. 2C). To identify which tyrosine phosphorylation sites were involved in the regulation of LDHA stability, we employed site-directed mutagenesis to generate LDHA mutants with individual, double (2F), or quadruple (4F) mutations. The individual mutations at each of the four tyrosine-phosphorylation sites did not affect the phosporylation LDHA. Note that weak expression of FGFR1 in the Y83F group might account for the relatively low phosphorylation of Y83F. Only the 4F mutant failed to be phosphorylated by FGFR1. The other mutants displayed only partially reduced phosphorylation, comfirming the previous report that all four tyrosine residues are phosphorylated by FGFR1.

Western blot analyses revealed that the individual mutations at each of the four tyrosine phosphorylation sites and the two 2F mutants did not affect the stablility of LDHA (P > 0.05). However, the 4F mutant showed a statistically significant reduction in stabilitity (Fig. 2D). Together, those results indicate that phosphorylation of the four tyrosine residues promotes the stability of LDHA.

The Ldhb promoter is heavily methylated in prostate cancer, which inhibits the transcription of LDHB (22). To determine whether FGF signaling suppresses LDHB expression via promoter methylation, we employed methylated DNA immunoprecipitation to pull down the methylated DNA for high-throughput sequencing (Fig. 3A). The results showed that the CpG islands in the Ldhb promoter were less methylated in MEF^ΔF cells than in the parental MEFs. Further bisulfite sequencing of the Ldhb promoter area confirmed that DNA demethylation was reduced in the MEF^ΔF cells (Fig. 3B). Consistent with the data that ablation of FGF signaling did not affect LDHA mRNA expression, no differences were observed in ldha promoter methylation between MEF and MEF^ΔF cells (Supplementary Fig. S2).

Because DNA demethylation is catalyzed by the three TET enzymes (TET1–3), we compared the expression levels of TET1–3 in MEF^ΔF cells and parental MEFs. We found that the expression of Tet1 at both mRNA and protein levels was significantly increased in the MEF^ΔF cells (Fig. 3C and D), suggesting that FGF signaling suppressed Tet1 expression. Consistantly, Ldhb, but not ldha, expression was higher in MEFs bearing Tet1^null alleles than in parental MEFs (Fig. 3E), further indicating that the expression of LDHB was reduced by DNA methylation.

DU145 cells highly expressed Fgfr1 (Fig. 4A). To investigate whether ectopic FGF signaling contributes to metabolic reprogramming in prostate cancer, we used CRISPR/Cas9 gene editing to ablate the Fgfr1 alleles in DU145 cells that expressed high levels FGFR1. Although it did not affect cell growth in the medium with 10% FBS, ablation of Fgfr1 compromised cell growth in the medium with 1% serum (Supplementary Fig. S3). In line with the data from MEFs, the expression of Ldha mRNA was not affected in the Fgfr1^null DU145 (DU145^ΔR1) cells, whereas that of Ldhb mRNA was increased (Fig. 4B). We then determined the expression profiles of the LDH isozymes in the parental and DU145^ΔR1 cells by Western blot analysis (Fig. 4C). The ablation of Fgfr1 reduced LDHA expression and increased LDHB expression at the protein level. Interestingly, expression of FGFR1 protein was increased by FGF2 treatment. However, the underlying mechanism remains to be determined. Together with Fig. 4B, the results further indicate that FGFR1 regulates LDHA at either the translational or posttranslational level and LDHB at the transcriptional level. Consistent with the report that FGFR1 phosphorylates LDHA, Western blot analyses revealed that the phosphorylation of LDHA at the Y10 residue was reduced in the DU145^ΔR1 cells. Consistently, the ablation of Fgfr1 reduced glucose uptake and lactate production and increased O₂ consumption (Fig. 4D). Similarly, treating DU145 cells with FGFR inhibitor AZD4547 also suppressed glycolysis (Supplementary Fig. S4). Together, the data suggest that the ablation of Fgfr1 in prostate cancer cells reprograms the cell energy metabolism from oxidative phosphorylation to aerobic glycolysis.

To investigate the functions of the LDH isoforms in the tumorigenic activity of prostate cancer cells, we employed CRISPR/Cas9 to delete the Ldha or Ldhb alleles in DU145 cells, designated DU145^ΔLdha and DU145^ΔLdhb, respectively. Western blot analyses of the xenografts revealed that the ablation of Ldha increased PGAM levels, while the ablation of Ldhb increased HK1, PFK3, and aldolase levels (Fig. 5A). Those results indicate that LDHA depletion compromised, whereas LDHB depletion enhanced, glycolysis in DU145 cells. Although no difference was observed in growth rates (Fig. 5B), when grafted subcutaneously into the flanks of nude mice, DU145^ΔLdha cells generated smaller tumors compared with parental DU145 cells (Fig. 5C). In contrast, grafts of DU145^ΔLdhb cells generated larger tumors than those of parental DU145 cells. Consistantly, similar results were derived from PC3 cells (Supplementary Fig. S5).

LDHA deletion and LDHB deletion both stimulated compensatory upregulation of glycolytic-related proteins but had opposite effects on tumor growth, suggesting that there might be unidentified proteins that mediate LDH-regulated tumor growth. Noticebly, ablation of Ldha increased expression of LDHB and vice versa. Whether these changes are because of feedback regulation to compensate the loss of LDH remains to be investigated.

Although there were no significant differences in tissue histology among the parental DU145, DU145^ΔLdha, and DU145^ΔLdhb xenografts, the DU145^ΔLdha xenografts had less, and the DU145^Δldhb xenografts had more, Ki67⁺ cells than the parental DU145 xenografts, indicating that the ablation of Ldha reduced, and the ablation of Ldhb increased, the number of proliferating cells in the xenografts (Fig. 6A and B). Furthermore, there were more apoptotic cells in the DU145^ΔLdha xenografts than in the parental DU145 xenografts, whereas the opposite relation was observed between the DU145^ΔLdhb xenografts and the parental DU145 xenografts (Fig. 6C). To investigate the impact of Ldha or Ldhb ablation on the tumor microenvironment, we stained DU145^ΔLdha, DU145^ΔLdhb, and parental DU145 xenografts with anti-CD31 antibodies and anti-F4/80 antibodies to examine the densities of microvessels and macrophages, respectively. There were fewer CD31⁺ and F4/80⁺ cells in the DU145^ΔLdha xenografts than in the parental DU145 xenografts. In contrast, there were more CD31⁺ and F4/80⁺ cells in the DU145^ΔLdhb xenografts than in the parental DU145 xenografts. Those results demonstrated that Ldha ablation compromised, whereas Ldhb ablation stimulated, angiogenesis and macrophage infiltration and thus the tumorigenic activity of DU145 cells.

We performed IHC staining to determine the expression patterns of LDHA and LDHB in a human prostate TMA that comprised 225 prostate cancer samples and 27 benign prostate samples (33). The results suggested that LDHA levels were higher in prostate cancer tissues than in adjacent prostate tissues (Fig. 7A). FGFR1 can phosphorylate human LDHA at multiple tyrosine residues (29). Compared with noncancerous tissues, prostate cancer tissues had higher levels of Y10-phosphorylated LDHA (pLDHA) and lower levels of LDHB (Fig. 7B).

The samples in the TMA were annotated with detailed information based on a 15-year follow-up of the patients, including PSA recurrence, Gleason Scores, pathologic stages, patients’ age, and survival time. Therefore, we separated the samples into two groups based on the median level. Higher than the median is defined as High and lower than the median is defined as Low. As shown in Fig. 7C and D, the clinical outcomes of the patients with high pLDHA levels (N = 153) were worse than those of the patients with low pLDHA levels (N = 48), while those of patients with low LDHB levels (N = 126) were worse than those of patients with high LDHB levels (N = 96); the patients with high pLDHA levels and low LDHB levels (N = 13) clearly had shorter biochemical recurrence-free times than those with low pLDHA levels and high LDHB levels (N = 86). Those results suggest that pLDHA and LDHB expression have the potential to serve as biomarkers for prostate cancer prognosis.

Fgfr1 is overexpressed in about 40% of human prostate cancers (35, 36). To determine whether the expression level of Fgfr1 correlates with the abundance of pLDHA, we used in situ hybridization to assess the expression levels of Fgfr1 in the TMA. Fgfr1 expressed was higher in prostate cancer than in benign tissues (Fig. 7D). The expression was associated with short PSA-free survival time (Fig. 7E). Furthermore, the Fgfr1 expression level was positively associated with the level of pLDHA (Fig. 7F). Together, the data imply that ectopically expressed Fgfr1 in prostate cancer deregulates the expression of LDH isozymes and thus changes the glycolysis and metabolism of the cells (Fig. 7G).

There is extensive evidence that ectopic FGF/FGFR1 signaling is a contributing factor in prostate cancer development and progression (11, 36–41), that FGF-mediated glycolysis plays a pivotal role in development (14, 42), and that FGFR1 directly phosphorylates LDHA and enhances its enzymatic activity (29). Herein, we showed that FGFR1 signaling promotes aerobic glycolysis by upregulating LDHA at the protein level and downregulating LDHB at the transcriptional level. The ablation of LDHA compromised, whereas that of LDHB enhanced, the tumorigenic activity of DU145 prostate cancer cells. Furthermore, high levels of phosphorylated LDHA and low levels of LDHB in human prostate cancer tissues were associated with short biochemical-recurrence and survival times of patients. Our results suggest that ectopic FGFR1 signaling contributes to prostate cancer progression by reprograming cell energy metabolism and those high levels of phosphorylated LDHA and high expression levels of LDHB are potential biomarkers for prostate cancer diagnosis and prognosis.

LDHA has four tyrosine phosphorylation sites. Phosphorylation at tyrosine 10 enhances the formation of tetramers and increases enzymatic activity, while phosphorylation at tyrosine 83 enhances the binding of NADH (29). In addition, high levels of FGFR1 expression are associated with high levels of phosphorylated LDHA (43). Tyrosine 10 only exists in human LDHA, suggesting that Y10 phosphorylation is not the only way in which FGFR1 activates LDHA. We showed that phosphorylation of the four tyrosine phosphorylation sites by FGFR1 extended the half-life of LDHA. Ablation of the FGF signaling axis significantly reduced the half-life of LDHA; the substitution of phenylalanine at the four tyrosine residues in LDHA had the same effect (Fig. 2). Together, the results suggest that tyrosine phosphorylation not only increases the enzymatic activity of LDHA but also enhances the stability of LDHA. More research is needed to determine how tyrosine phosphorylation affects the half-life of LDHA.

It has been reported that LDHB expression is silenced by hypermethylation of the LDHB promoter in human prostate cancer (23). We found that Fgfr1 ablation reduced the methylation of CpG islands in the Ldhb promoter (Fig. 3A and B). The ablation of FGF signaling in MEFs increased the expression of Tet1, which catalyzes the conversion of methylated guanidine to hydroxyl-methylated guanidine, the first step of DNA demethylation. Moreover, ablation of Tet1 increased LDHB expression although ablation of LDHA or LDHB did not affect TET1 expression (Supplementary Fig. S6). Thus, our results demonstrate that FGF signaling suppresses LDHB expression by promoting Tet1 expression.

Because LDHA promotes aerobic glycolysis while LDHB facilitates oxidative phosphorylation, LDHA may enhance tumor progression by fueling aerobic glycolysis and LDHB exerts opposite effects in glycolytic prostate cancer cells. Our results in Fig. 4 revealed that LDHA and LDHB have opposite effects on prostate cancer growth further support this hypothesis. Hence, it is essential to develop new strategies to specifically inhibit LDHA without compromising LDHB activity. FGFR1 selectively phosphorylates LDHA on multiple tyrosine residues, which stabilizes LDHA, and concurrently enhances the expression of LDHB. Yet, other signaling pathways may also regulate cell metabolism, which may explain why ablation of FGF signaling did not fully convert LDH expression from LDHA to LDHB. It has been reported that LDHA is degraded via chaperone-mediated autophagy (44). However, no difference in LDHA degradation was observed in DU145 and DU145^ΔR1 cells with or without treating with protease inhibitor leupeptin or autophagy inhibitor bafilomycin (Supplementary Fig. S7), suggesting that FGF signaling protected LDHA degradation via other mechanism. Although both lactate production and glucose consumption are reduced in DU145^ΔR1 cells (Fig. 4), the growth curves of DU145 and DU145^ΔR1 were not significantly different in glucose-free medium with or without lactate supplement, suggesting that deletion of FGFR1 did not confer the cells ability to use lactate as a main energy source. Further efforts are warranted to fully understand how cell metabolism is regulated.

Our data are consistent with reports that loss of LDHB correlated with malignancy (20, 26, 45). However, it is different from the report that LDHB expression is associated with poor survival in patients with uterine cancer (27, 46). One explanation is that in the cancer, such as uterine cancer, which largely rely on the TCA cycle to fuel cellular activities, can be inhibited by LDHB depletion. In other cancers, such as metastatic prostate cancer, however, rely on aerobic glycolysis, and therefore, is benefited from LDHB depletion. Therefore, more studies are needed to understand the role of LDHB in various types of malignancies. Our data show that although the deletion of FGFRs led to an increase in oxygen consumption, it still reduced ATP production. These results suggest that the upregulation of oxidative phosphorylation is not sufficient to make up the energy loss because of FGFR ablation, and that prostate cancer cells derive ATP primarily from aerobic glycolysis.

Although FRS2α is required for the FGF kinase to activate the ERK and PI3K/AKT pathways, we found that deletion of Frs2α alleles did not affect LDHA and LDHB expressions as well as lactate production in MEFs (Supplementary Fig. S8). This is consistent with the fact that FGFR1 directly phosphorylates LDHA. The mechanism by which FGF signaling regulates LDHB promoter methylation remains to be determined.

In summary, we demonstrated that FGF signaling promotes aerobic glycolysis and that ectopic FGF signaling in prostate cancer reprograms cell metabolism, suppressing aerobic glycolysis and promoting oxidative phosphorylation. LDHA deletion suppressed, whereas LDHB deletion promoted, the tumorigenic activity of prostate cancer cells. Furthermore, the LDHA overexpression and LDHB downregulation correlated with short biochemical recurrence and survival times in patients with prostate cancer. Our results shed new light on how the manipulation of ectopic FGF signaling may serve as a strategy for prostate cancer treatment.

No potential conflicts of interest were disclosed.

Conception and design: J. Liu, W.L. McKeehan, C. Wang, W. Zhong, F. Wang

Development of methodology: J. Liu, G. Chen, Z. Liu, S. Liu, Z. Cai, P. You, Y. Ke, Y. Huang, H. Pelicano, C.-L. Wu, C. Wang, W. Zhong, F. Wang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Liu, G. Chen, Z. Liu, S. Liu, Z. Cai, P. You, Y. Ke, L. Zhao, H. Pelicano, C.-L. Wu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Liu, G. Chen, Z. Liu, S. Liu, Z. Cai, P. You, Y. Ke, L. Lai, Y. Huang, H. Gao, L. Zhao, H. Pelicano, P. Huang, W.L. McKeehan, C.-L. Wu, C. Wang, W. Zhong, F. Wang

Writing, review, and/or revision of the manuscript: J. Liu, G. Chen, Z. Liu, S. Liu, Z. Cai, P. You, Y. Ke, Y. Huang, H. Gao, P. Huang, C.-L. Wu, C. Wang, W. Zhong, F. Wang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Liu, C. Wang, W. Zhong, F. Wang

Study supervision: W.L. McKeehan, C.-L. Wu, C. Wang, W. Zhong, F. Wang

We thank Drs. Juha Patanen and David Ornitz for sharing the Fgfr1^floxed and Fgfr2^floxed mice, respectively. This work was supported by the NIH CA96824, TAMU1400302, CPRIT 110555 grants, and a gift from Agilent Technology, Inc. (to F. Wang), the National Natural Science Foundation of China 81101712, 31371470, and 81270761 grants (to C. Wang); and National Key Basic Research Program of China (2015CB553706), National Natural Science Foundation of China (81571427) and the Fundamental Research Funds for the Central Universities 2017PY023 (to W.D. Zhong).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.