Increased fatty acyl saturation of phosphatidylinositol phosphates in prostate cancer progression

Phosphoinositides (PIPs) participate in many cellular processes, including cancer progression; however, the metabolic features of PIPs associated with prostate cancer (PCa) are unknown. We investigated PIPs profiles in PTEN-deficient prostate cancer cell lines, human prostate tissues obtained from patients with PCa and benign prostate hyperplasia (BPH) specimens using mass spectrometry. In immortalized normal human prostate PNT1B cells, PTEN deficiency increased phosphatidylinositol tris-phosphate (PIP3) and decreased phosphatidylinositol mono- and bis-phosphate (PIP1 and PIP2), consistent with PTEN’s functional role as a PI(3,4,5)P3 3-phosphatase. In human prostate tissues, levels of total (sum of all acyl variants) phosphatidylinositol (PI) and PIP1 in PCa were significantly higher than in BPH, whereas PIP2 and PIP3 contents were significantly lower than in BPH. PCa patients had significantly higher proportion of PI, PIP1, and PIP2 with 0–2 double bonds in acyl chains than BPH patients. In subgroup analyses based on PCa aggressiveness, mean total levels of PI with 0–2 double bonds in acyl chains were significantly higher in patients with pathological stage T3 than in those with pathological stage T2. These data indicate that alteration of PIPs level and the saturation of acyl chains may be associated with the development and aggressiveness of prostate cancer, although it is unknown whether this alteration is causative.


PTEN deficiency increases PIP 3 levels in the PNT1B normal prostate epithelial cell line.
To assess changes in PIPs levels of prostate cell line associated with the loss of PTEN, which is known to be a tumour suppressor gene as well as a PI(3,4,5)P 3 3-phosphatase, we used non-neoplastic human epithelial PNT1B prostate cells that exhibit wild-type PTEN expression, as previously reported (Fig. 1A, Supplementary Fig. 1) 15 . After doxycycline (dox)-inducible PTEN knockdown (KD) in the PNT1B cells, we confirmed lowered PTEN expression and a higher activation level of AKT (S473 and T308) than in the control cells (Fig. 1B, Supplementary Fig. 2). Cell proliferation at 96 h in the PTEN-KD PNT1B cells was significantly higher than that in the control cells (p < 0.001, Fig. 1C).
We then examined the levels of PIPs by a reverse phase LC/MS/MS method that allows quantitation of PI, PIP 1 (as the sum of PI3P, PI4P, and PI5P), PIP 2 (as the sum of PI(3,4)P 2 , PI(3,5)P 2 , and PI(4,5)P 2 ), and PIP 3 . The most abundant class of PIPs in both PTEN-KD cells and control cells was PI, followed by PIP 2 , PIP 1 , and PIP 3 (Fig. 1D). The mean level of total PIP 3 in the PTEN-KD cells was significantly higher than that in the control cells (8.14 pmol/1 × 10 6 cells vs. 3.67 pmol/1 × 10 6 cells, p < 0.001), whereas the mean levels of PIP 1 and PIP 2 in the www.nature.com/scientificreports www.nature.com/scientificreports/ PTEN-KD cells were significantly lower than in the control cells (p = 0.001 and p < 0.001, respectively). There was no significant difference in total PI levels between the PTEN-KD and control cells. Since PTEN is a PI (3,4,5) P 3 3-phosphatase, increased PIP 3 and decreased PIP 1 and PIP 2 in PTEN-KD cells verified that our method can measure PIPs profiles in prostate cells.

PIPs profiles in malignant and nonmalignant human prostate tissues.
To investigate the PIPs profiles of human prostates, we measured the PIPs levels in prostate tissues obtained from patients with PCa who underwent robot-assisted radical prostatectomy with those from patients with benign prostate hyperplasia (BPH) treated with transurethral resection or holmium laser enucleation of the prostate. Figure 2A demonstrates the levels of each PIPs class in human PCa and BPH tissues. The most abundant PIPs in both tissues were PI, followed by PIP 2 , PIP 1 and PIP 3 , which was consistent with the order observed in the human prostate cell line (Fig. 1D). In human prostate tissues, the mean levels of total PI and PIP 1 in the PCa tissues were significantly higher than those in the BPH tissues (PI: 510.4 ± 49.8 pmol/mg vs. 213.2 ± 22.0 pmol/mg, p < 0.001; PIP 1 : 6.3 ± 0.5 pmol/mg vs. 3.4 ± 0.6 pmol/mg, p = 0.001; Fig. 2B). In contrast, the mean levels of total PIP 2 and PIP 3 in the PCa tissues were significantly lower than in the BPH tissues (17.2 ± 1.9 pmol/mg vs. 29.2 ± 4.9 pmol/mg, p = 0.038; 0.07 ± 0.05 pmol/mg vs. 1.67 ± 0.46 pmol/ mg, p = 0.005; Fig. 2B). With respect to the acyl chain pattern of PIPs in human prostate tissues, the C38:4 acyl chain was the most abundant in PIPs from both cancerous prostate and BPH tissues ( Fig. 2A). As was the case with the prostate cancer cell line, PIPs containing C34:1, C36:1, C36:2, and C38:3 acyl chains were also abundant in both cancerous prostate and BPH tissues ( Fig. 2A). Intriguingly, we found that the PCa specimens exhibited a significantly higher proportion of total PI, PIP 1 , and PIP 2 with 0-2 double bonds in acyl chains and lower proportion of total PI, PIP 1 , and PIP 2 with ≥3 double bonds in acyl chains compared with that in the BPH samples (Fig. 3A, Supplementary Table 1). The levels of PIP 3 acyl species in most of the PCa tissues were undetectable ( Fig. 2A).
Finally, we assessed the difference in PIPs profiles between PCa tissues in terms of cancer aggressiveness based on subgroup categories, including preoperative serum prostate-specific antigen (PSA), Gleason grading group, and pathological T stage ( Table 1). The characteristics of the prostate cancer patients investigated in this study are described in Supplementary Table 2. There were no significant differences in levels of total PI, PIP 1 , or PIP 2 among dichotomised groups based on preoperative PSA level and biopsy Gleason grading (Table 1). With respect to the relationship between pathological stage and PIPs levels, the amount of total PI tended to be higher in patients with pT3 than those with pT2 (p = 0.065). Moreover, the quantity of PI with 0-2 double bonds in acyl chains was significantly higher in patients with pT3 than that in patients with pT2 (p = 0.004, Fig. 3B).These results demonstrate that an increased proportion of PI with 0-2 double bonds in acyl chains to that with ≥3 double bonds in acyl chains may be a hallmark of more aggressive cancerous prostate tissues.

Discussion
This is the first report demonstrating the changes in PIPs, including PI, PIP 1 , PIP 2 and PIP 3 , at the acyl species level in human BPH and PCa. Using a mass spectrometry-based approach, we found that deletion of the PIP 3 3-phosphatase, PTEN, in immortalised benign prostate cells resulted in increases in various acyl species of PIP 3 , which validated the applicability of the method to biological samples. The measurements of PIPs in human prostate samples revealed that PCa tissues had higher PI and PIP 1 levels and lower levels of PIP 2 and PIP 3 compared with BPH. In addition, PIPs in cancerous prostate tissues had a higher proportion of PIPs with 0-2 double bonds  www.nature.com/scientificreports www.nature.com/scientificreports/ in acyl chains and a lower proportion of PIPs with ≥3 double bonds in acyl chains than those in BPH. In addition, PCa with capsular or seminal invasion had higher PI with 0-2 double bonds in acyl chains compared with that in organ-confined PCa. These results suggest that PIPs class levels and their acyl chain profiles are altered in cancerous prostate tissues and in aggressive phenotypes of PCa in particular.
There have been several studies investigating phospholipid levels in human prostate tissues 12,13,16 . A study using high-resolution matrix-assisted laser desorption/ionisation imaging mass spectrometry demonstrated that PI (18:0/18:1), PI (18:0/20:3), and PI (18:0/20:2) were significantly elevated in PCa tissues relative to benign epithelium and thus proposed a biomarker algorithm using PI profiles to distinguish cancerous from benign epithelia 13 . These results are concordant with those of the present study showing that the levels of PI C36:1 and PI C38:3 are significantly higher in PCa than in the BPH. The present study also shows that the elevation of PI level is not specific to a particular PI acyl species, since the total PI level, including several PI acyl species mentioned above, was higher in cancerous prostate tissues. In addition to the PI status in human prostates, the present study is the first to successfully evaluate its phosphorylated derivatives, including PIP 1 , PIP 2 and PIP 3 , and showed that the level of total PIP 1 , as well as the level of total PI, is significantly higher in prostate cancer tissues than in BPH. In contrast, the level of PIP 2 and PIP 3 is lower in prostate cancer tissues than in BPH. Since prostate cancer is known to exhibit a higher rate of PTEN alteration, we speculated that PIP 3 levels were elevated in PCa tissues because PCa is known to frequently contain a class I PI3K alteration. The reason for the PIP 3 level decrease in the PCa tissues remains unclear; however, homozygous deletions of the PTEN locus occurred in just 15% of samples in a comprehensive molecular analyses of 333 primary prostate cancers using the TCGA database 17 , and a substantial amount of molecular heterogeneity has been observed in clinically extracted prostate cancer tissues. In addition, although our preliminary results showed that there was no significant association between PTEN immunoreactivity scores and PIPs levels in the tumors ( Supplementary Fig. 3), future studies focusing on the interaction between PIPs class levels and mutation status/expression of PTEN using a larger number of patients are needed.
Of note, we found a higher proportion of PIPs with 0-2 double bonds in acyl chains and a lower proportion of PIPs with ≥3 double bonds in acyl chains in human PCa. In the current study, we could not distinguish two acyl chains in each PIP separately, which means that the clear classification of acyl chains into saturated/monounsaturated fatty acid (SMFA) and polyunsaturated fatty acids (PUFA) was not performed. However, we confirmed the higher levels of SMFA and lower levels of PUFA in PIPs of cancerous tissues, after excluding C34:2 and C36:2 acyl species from the statistical analysis ( Supplementary Fig. 4). Accordingly, our study revealed that the lipid saturation in PIPs increased in PCa tissues, especially in advanced PCa.
In a comprehensive lipidomic analysis of 267 human breast tissues, Hilvo et al. demonstrated that saturated-fat-containing phosphatidylcholines were elevated in tumors compared with normal breast tissues and associated with clinical outcomes 18 . Guo et al. conducted a lipidomic study of 134 tissue samples from six different types of malignancies (breast, lung, colorectal, oesophageal, gastric, and thyroid) using matrix-assisted laser desorption/ionization Fourier-transform ion cyclotron resonance mass spectrometry and demonstrated that monounsaturated fatty acids and monounsaturated phosphatidylcholines levels increased significantly in the cancer microenvironment compared with that in the adjacent normal tissue 19 . In PCa, a study of phospholipid composition by ESI-MS/MS showed that tumours with increased expression of the lipogenic enzyme FASN had a consistent increase in SMFA acyl chains and decrease in PUFA acyl chains of phosphatidylcholine in prostate tumour tissues compared with matching normal tissues 12 . Furthermore, PUFAs were easily influenced by lipid peroxidation, which affects oxidative stress-induced cell death in PCa LNCaP cells. The present study is the first to reveal a high level of three PIPs classes (PI, PIP 1 , and PIP 2 ) containing 0-2 double bond acyl chains in prostate cancer tissues, and especially in aggressive prostate cancer, and along with these other lines of evidence, suggests that cancerous tissues contain elevated levels of phospholipids with SMFA acyl chains and that the SMFAs/ PUFAs balance in tumours may be associated with cancer development and progression. In fact, a study using human glioblastoma and breast cancer cell lines to assess the impact of PUFA and SMFA supplementation on the growth of cancer cells showed that PUFA supplementation inhibited cancer cell growth, whereas SMFA enhanced their proliferation 20 . Moreover, related enzymes such as Stearoyl-CoA desaturase (SCD; a rate-limiting enzyme in the biosynthesis of monounsaturated fatty acids) are highly expressed in human prostate and breast cancers 21,22 , whereas a lower level of delta-6-desaturase (the main determinants of PUFA levels) has been observed in breast tumours compared with normal tissues 23 . The actual SMFA and PUFA profiles in two acyl chains of each PIP in PCa tissues should be evaluated in future studies to clarify key lipogenic enzymes in prostate cancer aggressiveness. Taken together, these lipogenic features in cancer, including PCa, provide specific biomarkers for the detection of aggressive phenotypes of prostate cancer and suggest that these enzymes are potential targets for the exploration of a novel treatment strategy of lipid desaturation to overcome treatment resistance and the aggressive phenotype of PCa.
This study has several limitations. Regarding in vitro study, we found a time lag between the difference in PTEN expression and the impact of cell proliferation. We speculated that differences in the expression and/or activation of down-stream targets of PTEN are also important in modulating cell proliferation. In fact, the activation of AKT in PTEN-KO cells at 96 h was higher than that at 72 h. Moreover, other potential downstream targets, such as p27 and Bad, may influence the difference in cell proliferation. In human study, we did not assess saturation status of other membrane lipids other than PIPs in this study, and it would be intriguing to know whether or not this higher lipid saturation status is specific to PIPs. We also did not distinguish between regioisomers of PIPs, for example, PI(3,4)P 2 and PI(4,5)P 2 . The levels of each regioisomer and its functional impact on PCa should be clarified in a future study. Additionally, information on the dynamic changes of the PIPs profile in PCa remains elusive because we only measured the PIPs profile at specific time points in preclinical and clinical PCa. Furthermore, in humans, we only compared PIPs levels in pathogenic prostate tissue with BPH or PCa. Therefore, further study is needed to investigate the difference in PIPs profiles in other prostate tissues, including normal prostate from healthy subjects. Finally, we cannot assess the pure expression profiles of prostate epithelial (2019) 9:13257 | https://doi.org/10.1038/s41598-019-49744-3 www.nature.com/scientificreports www.nature.com/scientificreports/ cells because the location of cancerous regions is quite heterogeneous in the prostate, which poses difficulties for the isolation of cancerous cells. The expression profiles in cancer cell and the cancer stromal environment has therefore yet to be clarified.
In conclusion, our study successfully measured PIPs class levels and acyl chain profiles in preclinical prostate cells and clinical PCa specimens. PTEN deletion resulted in increased PIP 3 levels in prostate cells with declines in PIP 2 and PIP 1 . The total levels of PIPs and acyl chain components differed in human PCa compared with BPH, and advanced PCa possessed specific PIPs profiles. Evaluating PIPs profiles in PCa may provide an understanding of the detailed mechanism of PCa initiation and progression as well as novel biomarkers for screening and prognostication. In particular, high PI/PIP 1 and the targeting of saturation status, respectively, have the potential to predict cancer progression and become a novel treatment strategy for PCa. Human prostate tissues. Fresh human prostate tissues were obtained from patients who underwent radical prostatectomy without any preoperative hormonal or chemotherapeutic treatment at Akita University Hospital between 2015 and 2017 24 . All patients were histologically diagnosed with localised PCa and had large, visible suspected tumour regions upon preoperative MRI imaging. The patients' characteristics are described in Supplementary Table 2. The prostates were first coronally dissected into four pieces at even intervals. Slices of the prostates, including ≥50% of the cancer regions with pathological diagnosis of their mirror images, were then snap-frozen in liquid nitrogen and stored at −80 °C until needed. All slides were reviewed by a pathologist (T.Y.), who was blinded with respect to each patient's clinical background 24 . Human benign prostate tissues were obtained from patients treated by transurethral resection of the prostate or by holmium laser enucleation of the prostate (HoLEP) at hospitals affiliated with Akita University School of Medicine. The resected pieces of the prostates were snap-frozen in liquid nitrogen and stored at −80 °C until needed. This study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Institutional Review Board of the Akita University School of Medicine (IRB #1034). Written informed consent was obtained from each participant prior to obtaining prostate tissues.
Western blotting. Prostate cells were homogenised in lysis buffer (50 mM Tris-HCl, pH 8.0, 100 mM NaCl, 1 mM EDTA, 1% Triton X-100, 1 mM DTT, 1 mM Na3VO4, 30 mM sodium pyrophosphate, 50 mM sodium fluoride, and protease inhibitor cocktail [Roche Applied Science, Mannheim, Germany]). After centrifugation at 20,000 g for 15 m, supernatants (10 µg total protein) were subjected to standard SDS-PAGE and immunoblotting as previously described 25 . Establishment of a doxycycline-inducible PTEN-knockdown PNT-1B cell line. The sh-RNA plasmid against PTEN was purchased from OpenBiosystems (USA). Lentiviral production was performed per the supplied protocol and as used in a previous study 25 . After a 72 h incubation at 37 °C, lentivirus-containing supernatant was harvested and applied to PNT1B cells. Once stable in culture, puromycin selection was initiated at a concentration of 1 μg/mL. To induce sh-RNA expression, doxycycline was added to culture medium at a concentration of 1 μg/mL and replaced every 24 h 26 .
Histological analysis. Human prostate tissues were fixed in 10-20% formalin neutral buffer solution and embedded in paraffin. Sections (3-5 μm) were cut and stained with hematoxylin and eosin in accordance with standard procedures.