EZH2 Palmitoylation Mediated by ZDHHC5 in p53-Mutant Glioma Drives Malignant Development and Progression

Glioma is the most common type of primary brain tumor in adults and is associated with poor prognosis (1, 2). Patients with tumors harboring mutant p53 (over 30% of cases) show therapeutic resistance and especially poor outcome (3–6). Mutations in p53 are linked to increased cell survival and growth, decreased apoptosis, and drug resistance (5, 7, 8).

Protein S-palmitoylation is a reversible posttranslational modification in proteins with fatty acids that is regulated by protein acyltransferases (PAT), which are characterized by a conserved catalytic domain containing an Asp-His-His-Cys (DHHC) motif (9, 10). Abnormalities in DHHC proteins and protein palmitoylation have been implicated in neurologic abnormalities in humans such as schizophrenia, X-linked mental retardation, and Huntington disease (11, 12). DHHC proteins and their substrates play important roles in tumorigenesis (13, 14). Palmitoylation of H- and N-Ras facilitates Ras localization to the plasma membrane and is required for protein activity (15, 16), and inhibiting ZDHHC20 palmitoyltransferase leads to the dependence of cancer cells on EGF receptor signaling for survival (17).

The gene-encoding zinc finger DHHC-type-containing (ZDHHC)5 is located in a region of chromosome 11q12.1 (18) associated with a high rate of chromosomal translocation and variation, which has been linked to the development of various cancers (19–22). ZDHHC5 stimulates the proliferation and anchorage-dependent and -independent colony formation of non–small cell lung cancer cell lines, and was found to be required in a subset of these cells for establishment of tumor xenografts in mice (22). ZDHHC5 is highly expressed in a variety of embryonic cells including neural progenitor cells, but not in most adult tissues (23). In cultured neuronal stem cells, induction of neural differentiation led to rapid degradation of ZDHHC5 (23). Gliomas in the brain subventricular zone and cortex can originate from neural precursors with mutations in tumor suppressor genes such as p53 (24, 25). However, the precise role of ZDHHC5 in the development and progression of p53-mutant glioma is not known.

To address this issue, we analyzed the relationship between ZDHHC5 and p53 in human glioma specimens, glioma cell lines, and glioma stem cell (GSC) cultures. We also examined the significance of this interaction with respect to the malignant progression of gliomas.

Sixty glioma tissues were used for qRT-PCR, Western blot analysis, and IHC analysis. Samples were randomly collected from glioma patients who underwent curative resection with informed consent between 2013 and 2015 at Cancer Hospital of Hefei Institutes of Physical Science (Anhui, China) or First Affiliated Hospital of Anhui Medical University (Anhui, China). Study protocols were approved by the the Institutional Review Board of the Cancer Hospital of Hefei Institutes of Physical Science, Chinese Academy of Sciences (CAS), and written informed consent was obtained from patients based on the Declaration of Helsinki. Animal experiments were performed according to the guidelines of the Animal Use and Care Committees at Hefei Institutes of Physical Science, CAS.

The palmitate analogue inhibitor 2-bromopalmitate (2BP) was purchased from Sigma-Aldrich. Antibodies against the following proteins were used for IHC/immunofluorescence analysis (IHC/IF) and Western blotting: ZDHHC5 (1:500 for IHC/IF and 1:2000 for Western blotting); p53 (1:2500 for Western blotting); mutant p53 (1:500 for IHC and 1:2,000 for Western blotting; all from Abcam); enhancer of zeste homolog (EZH)2 (1:1,000 for IHC/IF and 1:4,000 for Western blotting; BD Biosciences); phospho-EZH2 S21 (1:500 for IHC/IF and 1:1,000 for Western blotting; Bethyl Laboratories); (sex-determining region Y)-box (Sox)2 (1:500 for IF and 1:2,000 for Western blotting; R&D Systems); tri-methylated histone H3K27 (1:500 for IHC/IF and 1:1,000 for Western blotting; Millipore); glial fibrillary acidic protein (GFAP; 1:1,000 for IHC/WB; Dako); and β-actin (1:5,000 for Western blotting; Sigma-Aldrich).

Lentiviral short hairpin RNA (shRNA) clones targeting ZDHHC5 and a scrambled nontargeting control sequence were purchased from Invitrogen. HEK293FT cells were cotransfected with the packaging vectors psPAX2 and pCI-VSVG (Addgene) or the ViraPower Lentiviral Expression System packaging mix (Invitrogen) using Lipofectamine 2000 (Invitrogen) to produce the virus. The efficiency of shRNA-mediated knockdown was evaluated by Western blotting and real-time (RT)-PCR. The shRNA sequences were as follows: shRNA #1, 5′-CAC CGG AAG CAT TAG TGT TGA CTG GCG AAC CAG TCA ACA CTA ATG CTT CC-3′ and shRNA #2, 5′-CAC CGA AAT CAA GCC TGA CGA AGT TCG AAA ACT TCG TCA GGC TTG ATT TC-3′.

For primary cells, glioma tissue specimens were collected from 15 patients (ages 42–75) who underwent curative resection for glioma between 2013 and 2015 at Cancer Hospital of Hefei Institutes of Physical Science, with Institutional Review Board approval. Within hours of surgical removal, glioma specimens were enzymatically dissociated into single cells as described previously (26). For short-term in vitro expansion of GSCs, cells were cultured in Neurobasal medium (Thermo Scientific) with N2 and B27 supplements (Invitrogen), human recombinant basic fibroblast growth factor, and EGF (10 ng/mL each; Invitrogen). Human neural stem cells (hNSC) were obtained from Lonza at 2015 and cultured in a similar manner to GSCs. To induce differentiation of GSCs and hNSCs, cells were cultured in the absence of growth factors or in the presence of 10% FBS (Gibco). U87, D54, A172, CCF-STTG1, U373, U251, U118MGT-98G, and SWO38 glioma cell lines were obtained between 2012 and 2014 from the Cell Bank of Type Culture Collection of the Chinese Academy of Sciences, Shanghai Institute of Cell Biology, Chinese Academy of Sciences (Shanghai, China), where they were characterized by DNA fingerprinting and isozyme detection. All the cell lines were kept at low passages for experimental use, and revived every 3 to 4 months. Cell lines were cultured in DMEM (Hyclone) supplemented with 10% FBS and 1% (100×) penicillin/streptomycin (Hyclone). All cell lines used in this study were regularly authenticated by morphologic observation and tested for the absence of mycoplasma contamination. They were tested for mycoplasma in March 2017 for the last time.

Intracranial transplantation of GSCs into nude mice was performed as described previously (26). Briefly, 36 hours after lentiviral infection, live cells were counted and implanted into the right frontal lobes of athymic nude mice. The animals were maintained until they exhibited neurologic symptoms, and then sacrificed for histologic and IHC analyses of tumors. Brains were perfused with 4% paraformaldehyde by cardiac perfusion and postfixed overnight at 4°C.

Total RNA was prepared using the RNeasy kit (Qiagen) and reverse transcribed into cDNA using a SuperScript III First-Strand Synthesis kit (Invitrogen), and amplified with Platinum Taq Polymerase (Invitrogen) on a LightCycler 480 system (Bio-Rad). Each sample was prepared in triplicate, and target gene expression levels were calculated using the 2^−ΔΔC_t method, with GAPDH serving as the internal control. Forward and reverse primer sequences were as follows: ZDHHC5, 5′-acacctcggcttggctacta-3′ and 5′-gttggctccttcaagctgtc-3′; KRAS, 5′-tgtggtagttggagctggtg-3′ and 5′-tgacctgctgtgtcgagaat-3′; telomerase reverse transcriptase (TERT), 5′-cgtggtttctgtgtggtgtc-3′ and 5′-ccttgtcgcctgaggagtag-3′; p53, 5′-gttccgagagctgaatgagg-3′ and 5′-tctgagtcaggcccttctgt-3′; and β-actin, 5′-ggacttcgagcaagagatgg-3′ and 5′-agcactgtgttggcgtacag-3′.

ZDHHC5 cDNA was cloned in-frame into the pGEX6p-1 vector to obtain the glutathione S-transferase (GST) fusion protein GST-ZDHHC5 for the GST pull-down assay. GST and GST-ZDHHC5 fusion proteins were immobilized on glutathione–sepharose beads (Sigma-Aldrich) and incubated with U251 cell lysates overnight. The beads were washed four times, and bound proteins were eluted with SDS loading buffer containing 5% β-mercaptoethanol and resolved by SDS-polyacrylamide gel electrophoresis (PAGE) and Western blotting.

Cells were collected and lysed in radioimmunoprecipitation assay buffer supplemented with protease inhibitors, incubated on ice for 30 minutes, and cleared by centrifugation at 12,000 rpm and 4°C for 15 minutes. Total protein lysate (500 μg) was subjected to immunoprecipitation overnight at 4°C with agarose-immobilized antibody (1 μg of anti-ZDHHC5, EZH2). Immunoprecipitated and coimmunoprecipitated proteins were analyzed by SDS-PAGE and Western blotting.

Equal amounts of cell lysate were resolved by SDS-PAGE and transferred to a polyvinylidene difluoride membrane (Millipore) that was probed with primary antibodies for 16 hours at 4°C, then blocked with 5% skim milk/0.1% Tween-20 in Tris-buffered saline for 1 hour at room temperature. Horseradish peroxidase-conjugated secondary antibody was applied, followed by enhanced chemiluminescence detection (Pierce).

Cells, neurospheres, or tumor sections were fixed with 4% paraformaldehyde, washed with PBS, and incubated in blocking buffer (1× PBS containing 5% normal goat serum and 0.3% Triton X-100) for 1 hour. Samples were then incubated with primary antibodies for 16 hours at 4°C followed by detection with Alexa 488 goat anti-mouse (Invitrogen) and Alexa 568 goat anti-rabbit (Invitrogen) secondary antibodies. Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (Beyotime Institute of Biotechnology), and samples were covered with coverslips that were fixed using fluorescent mounting medium (Beyotime Institute of Biotechnology). Images were acquired with a fluorescence microscope (IX71; Olympus) and adjusted for brightness and contrast using Image-Pro Plus 6.0 software (Media Cybernetics).

GSCs seeded in 6-well plates at 100,000 cells per well were infected with lentivirus expressing shRNA for 48 hours. To determine the fraction of cells in each phase of the cell cycle, cells were fixed with 75% ethanol and stained with propidium iodide followed by flow cytometry (FACScan; BD Biosciences). To detect apoptotic cells, Annexin V-fluorescein isothiocyanate (FITC) staining was performed using the Annexin V-FITC Apoptosis Detection kit (BD Biosciences) according to the manufacturer's instructions. Cells were fixed in 70% ethanol, centrifuged, and resuspended in a propidium iodide solution (50 mg/mL propidium iodide, 10 mmol/L Tris, 5 mmol/L MgCl₂, 10 mg/mL ribonuclease A, and 1 mL/mL of Nonidet P-40).

Cells (1 × 10⁵) were loaded into the top chamber of Transwell dishes (8-μm pore size; Corning Inc.). FBS (10%) was placed in the bottom chamber as a chemoattractant. After 24 hours, cells in the bottom chamber were fixed and stained with 0.005% (w/v) crystal violet. The number of migrated cells was quantified by counting those in five random fields of each membrane.

Glioma patient data were publicly available in deidentified form and were obtained from the National Cancer Institute Repository for Molecular Brain Neoplasia Data (REMBRANDT) database. Differences between the groups were analyzed with the log-rank P value. Up- and downregulation among glioma specimens were taken as a ≥2-fold increase and decrease, respectively, in ZDHHC5 expression compared with specimens from patients without glioma.

Data in the graphs represent mean ± SD of at least three independent measurements. Differences in mean values were evaluated with the Student t test. Intergroup comparisons were made with the paired two-sample t test. Differences were considered significant at P < 0.05.

We first investigated the clinical significance of ZDHHC5 expression levels in glioma patients with p53 mutation using expression data retrieved from the REMBRANDT database. Patients were divided patients into “low” and “high” expression groups based on the median ZDHHC5 level. We found an inverse correlation between tumor ZDHHC5 expression and overall survival (P < 0.0001; Fig. 1A) and disease-free survival (P < 0.0001; Fig. 1B). We then compared ZDHHC5 expression and p53 mutational status in 60 glioma tissue, including low-grade astrocytoma (LGA, grade II; n = 23), anaplastic astrocytomas (AA, grade III; n = 6), and glioblastoma multiforme (GBM, grade IV; n = 31), and 10 normal brain tissue samples. The rate of p53 mutation was 26.09% in LGA, 50.00% in AA, and 64.52% in GBM (Supplementary Table S1). Both the rate of p53 mutation and the percentage of cells positive for p53 mutation were higher in grades III and IV than in grades I and II gliomas (Fig. 1C; Supplementary Table S1). Consistent with these results, ZDHHC5 level in glioma was elevated relative to normal brain tissue, and was positively correlated with the degree of malignancy (Fig. 1C–E). Moreover, ZDHHC5 expression was correlated with p53 gene mutations in glioma (χ² = 6.75, P < 0.01; Supplementary Table S2).

To investigate the correlation between p53 mutation and ZDHHC5 expression in glioma in greater detail, we compared ZDHHC5 levels in wild-type and mutant p53 glioma cell lines. Western blotting revealed that ZDHHC5 protein level was higher in glioma cell lines harboring mutant as compared with wild-type p53 (Fig. 2A). Moreover, RT-PCR analyses revealed elevated levels of ZDHHC5 transcript in p53-mutant glioma cells (Fig. 2B). The most common frameshift and missense mutations in the p53 gene result in a gain of function and interaction of p53 with nuclear transcription factor (NF)-Y, which leads to aberrant transcriptional regulation of oncogenic factors (27, 28). The 1-kb upstream region of the ZDHHC5 promoter contained four potential NF-Y–binding sites (Supplementary Fig. S1); chromatin immunoprecipitation (ChIP) revealed direct association between mutant p53 or NF-Y protein and these regulatory sequences (Fig. 2C). Moreover, the double ChIP assays were performed, and showed both proteins mutant p53 and NF-Y were corecruited onto the ZDHHC5 promoter (Supplementary Fig. S2A). To assess the functional significance of this interaction, we performed a transactivation assay in which SWO-38 cells (p53 Arg280) were transiently cotransfected with plasmids encoding wild-type p53 or p53 siRNA along with the ZDHHC5 promoter reporter plasmid (pCCAAT-B2LUC). ZDHHC5 promoter activity was inhibited in the presence of wild-type p53 or by p53 knockdown (Fig. 2D). This activity was also abolished for the ZDHHC5 promoter harboring three mutated CCAAT boxes (pmutCCAAT-B2LUC; Fig. 2D). In agreement with these findings, ZDHHC5 promoter activity was enhanced by expression of mutant p53 His175 in wild-type p53 U87 cell lines (Fig. 2E). Cotransfected with expression mutant, p53 and NF-Y could synergistically activate the ZDHHC5 promoter activities (Supplementary Fig. S2B). These results demonstrate that mutant p53 acts in conjunction with NF-Y to regulate ZDHHC5 expression.

ZDHHC5 is highly expressed not only in neural progenitor cells (23) but also in p53 mutant glioma. To determine whether aberrant ZDHHC5 expression is associated with the occurrence of glioma, hNSCs were cotransfected with expression plasmids encoding constitutively activated (CA)-KRAS, hTERT, dominant-negative (DN)-p53, and ZDHHC5. KRAS, TERT, and p53 mutation and abnormal expression are common in malignant glioma (29, 30). The expression and activity of these genes in hNSCs were examined by RT-PCR and Western blotting (Supplementary Fig. S3A and S3B). hNSCs harboring CA-KRAS/hTERT/DN-p53/ZDHHC5 exhibited stem cell characteristics (Supplementary Fig. S3C) and had higher proliferative capacity than control hNSCs (Fig. 3D). However, the differentiation potential of these hNSCs was altered, as evidenced by upregulation of GFAP relative to control hNSCs (Supplementary Fig. S3E and S3F). We also evaluated the role of ZDHHC5 in the development of glioma with the tumor formation assay. hNSCs expressing CA-KRAS/hTERT/DN-p53 but not those expressing CA-KRAS or hTERT alone generated tumors. However, the rate of tumor formation was higher for hNSCs expressing CA-KRAS/hTERT/DN-p53/ZDHHC5 (Fig. 3A and B), yielding tumors with a 23-fold higher volume than CA-KRAS/hTERT/DN-p53 hNSCs. In addition, GFAP expression level and percentage of proliferating cells were higher for CA-KRAS/hTERT/DN-p53/ZDHHC5 than for CA-KRAS/hTERT/DN-p53 hNSCs (Fig. 3C and D). Consistent with these results, injection of ZDHHC5 siRNA into hNSCs harboring CA-KRAS/hTERT/DN-p53 formed much smaller tumors than negative control siRNA (Fig. 3E and F). These results indicate that ZDHHC5 contribute to the development of glioma.

Immunofluorescence analysis was used to examine ZDHHC5 expression in p53-mutant glioma samples. A significant fraction of ZDHHC5⁺ cells (24%–57%) expressed the neural stem cell marker Sox2, and 19%–38% of Sox2+ cells were positive for ZDHHC5 (Supplementary Fig. S4A and S4B). Accordingly, ZDHHC5 was highly coexpressed with Sox2 in cultured GSCs (Supplementary Fig. S4C).

Given the elevated levels of ZDHHC5 in GSCs, we examined whether ZDHHC5 is essential for GSC self-renewal with the single-cell neurosphere formation assay. We prepared shRNA-expressing lentiviruses to target ZDHHC5 expression. Under free-floating neurosphere culture conditions, ZDHHC5 knockdown reduced the neurosphere formation capacity of GSCs (Fig. 4A and B). A flow cytometry analysis revealed that the rate of apoptosis was increased in ZDHHC5 shRNA-transfected as compared with negative control-transfected cells or nontransfected control cells (Fig. 4C). The 5-bromo-2-deoxyuridine incorporation assay confirmed that GSC proliferation was decreased by ZDHHC5 knockdown (Fig. 4D). We also analyzed GSC cycling by flow cytometry. After 2 days of culture, there were more cells in G₀–G₁ phase and fewer in S-phase in ZDHHC5-depleted as compared with negative control cultures, suggesting that ZDHHC5 deficiency caused G₀–G₁ arrest (Fig. 4E and F). The stemness markers cluster of differentiation (CD)133 and Sox2 were also downregulated by ZDHHC5 knockdown (Fig. 5A). When ZDHHC5 expression was allowed to recover to initial levels, Sox2 and CD133 levels returned to control values (Supplementary Fig. S5A). What is more, overexpression of ZDHHC5 could enhance self-renewal of p53 mutant GSCs, as the expression level of Sox2 and CD133 was increased by ZDHH5 overexpression (Supplementary Fig. S5A). ZDHHC5 might regulate GSC stemness dependent-transcriptional network, as showing the changed levels of Oct4, Nanog, and Snail1 (Supplementary Fig. S5B). However, given that GSCs are truly a heterogenous mix of stem and progenitor cells with different proliferative capacities, we used a serial dilution assay, to allow for the differential analysis of stem and progenitor populations. In this assay, it was noted that p53-mutant GSCs transduced with control shRNA or nontransfected produced a significantly greater number of neurospheres at every dilution tested compared with the ZDHHC5-deficient GSCs (Supplementary Fig. S5C). These results indicate that ZDHHC5 is required for GSC maintenance.

We next examined whether ZDHHC5 is required for migration and invasion of GSC-derived cells. In the transwell migration assays, human umbilical vein endothelial cells (HUVEC) and GSCs were seeded on the bottom surface of Boyden chambers and on insert membranes, respectively, in the presence of TGFβ. In invasion assays, the top of the insert membrane was coated with Matrigel prior to GSC seeding. ZDHHC5-depleted GSCs showed reduced migration and invasion than control cells (Fig. 5B and C). Moreover, medium from ZDHHC5-depleted GSC cultures reduced the ability of HUVECs to form vessel-like tubular structures (Fig. 5D), and xenograft brain tumors derived from ZDHHC5-deficient GSCs showed less vasculature (Supplementary Fig. S6A and S6B). Thus, ZDHHC5 is essential for preserving the migratory, invasive, and angiogenic potentials of GSCs.

We investigated whether ZDHHC5 is required for GSC tumorigenicity. An analysis of GSC survival curves generated following exposure to various doses of X-ray (Fig. 6A and B) revealed that ZDHHC5 knockdown increased the sensitivity of GSCs to radiation as compared with cells transfected with a scrambled sequence. The colony formation assay showed that loss of ZDHHC5 reduced the number of colonies formed by GSCs (Fig. 6C and D). Similarly, the intracranial tumor formation assay revealed that while injection of GSCs transduced with control shRNA resulted in animal death within 18 to 46 days (Fig. 6E and F), injection of ZDHHC5-deficient GSCs extended the survival rate (Fig. 6F). With as few as 100 cells, GSCs transduced with control shRNA could form tumors in nude mice, whereas ZDHHC5-deficient GSCs were unable to do so at the same dilution (Supplementary Table S3). These results suggest that loss of ZDHHC5 reduced the number of tumor-initiating GSCs, which inhibited tumor development.

An anti-ZDHHC5 antibody was used to purify a ZDHHC5 complex in p53-mutant gliomas, and the components of the complex were analyzed by LC/MS-MS. Peptide sequences of EZH2 were identified, suggesting that it is a binding partner of ZDHHC5 (Fig. 7A). The physical interaction between ZDHHC5 and EZH2 was examined in vitro using recombinant GST-ZDHHC5 and p53-mutant glioma lysates. In a GST pull-down assay, EZH2 protein in the lysate bound directly to GST-ZDHHC5 (Fig. 7B). This interaction was confirmed by immunoprecipitation analysis of 293T cells transfected with Flag-tagged EZH2 and GFP-tagged ZDHHC5 (Fig. 7C). We analyzed this interaction by constructing a series of ZDHCH5 deletion mutants, and observed that ZDHHC5 bound to EZH2 via a 141–715-residue region located at the C terminus (Fig. 7C). This interaction was independent of the PAT activity of ZDHHC5, as both wild-type and ΔDHHC ZDHHC5 proteins immunoprecipitated with EZH2 (Fig. 7C). Besides, EZH2 inhibitor GSK126 could well suppress the growth and invasion of ZDHHC5-mediated GSCs (Supplementary Fig. S7A and S7B).

We next examined whether ZDHHC5 modulates the expression and histone methyltransferase activity of EZH2. ZDHHC5 overexpression did not affect EZH2 protein level; however, trimethylation of histone 3 at lysine 27 (H3K27me3 level) was decreased (Fig. 7D). This change in EZH2 histone methyltransferase activity was dependent on the 141–715-residue region and DHHC motif of ZDHHC5 (Fig. 7D). EZH2 S21 phosphorylation level was also important for the ZDHHC5-mediated methyltransferase activity of EZH2, as this was reduced by ZDHH5 overexpression (Fig. 7D). These findings are in agreement with our IF and IHC results. ZDHHC5 knockdown induced the phosphorylation of EZH2 S21, but reduced H3K27me3 level in GSCs (Fig. 7E and F). H3K27me3 was reduced by ZDHHC5 knockdown (Fig. 7G). These results indicate that EZH2 activity mediates ZDHHC5-dependent malignant progression of glioma.

The DHHC motif of ZDHHC5 is associated with palmitoyltransferase activity. We speculated that ZDHHC5 ΔDHHC impairs EZH2 methyltransferase activity by inhibiting its palmitoylation. To test this hypothesis, GSCs were cotransfected with Flag-EZH2 and GFP-ZDHHC5, and palmitoylation levels were examined with the acyl-biotin exchange assay. EZH2 was palmitoylated by wild-type ZDHHC5 (lane 1; Fig. 7H); however, this was greatly reduced for mutant ZDHHC5 lacking the DHHC domain (Fig. 7H). Attesting to its specificity, EZH2 did not palmitoyalte by other similar DHHC-containing proteins, ZDHHC8, 13 and 17, although ZDHHC8 could associate with EZH2 (Supplementary Fig. S7C). EZH2 has two potential palmitoylation sites that are conserved across species (Supplementary Fig. S8). To determine which of the two EZH2 cytoplasmic cysteines can be palmitoylated, we mutated these cysteines to alanine. Cys 571 and Cys 576 were required for EZH2 palmitoylation (lanes 3–6; Fig. 7H), as mutation of both residues resulted in complete inhibition of palmitoylation (lanes 7 and 8; Fig. 7H). This was accompanied by an increase in EZH2 S21 phosphorylation. Consistent with these results, ZDHHC5 expression and EZH2 palmitoylation level were correlated with p53 gene mutations in glioma, and were higher in grades III and IV than in grades I and II gliomas (Supplementary Fig. S9A). Moreover, the the phosphorylation status of EZH2 was negative correlation of its palmitoylation level. Consistent with these results, overexpression of EZH2 S21D (a mutant that mimics the active EZH2) could recover the inhibitory effect of ZDHHC5 shRNA on Sox2 and CD133 expression (Supplementary Fig. S9B). To further assess whether ZDHHC5 palmitoylation is associated with glioma progression, we established a xenograft tumor model using p53-mutant GSCs. Intratumoral injection of the palmitoylation inhibitor 2BP suppressed tumor growth relative to PBS-injected animal group (Fig. 7I), and obviously affected ZDHHC5-overexpressed group (Supplementary Fig. S9C). Moreover, 2BP-injected tumors showed low levels of H3K27me3 (Supplementary Fig. S9D). These results demonstrate that ZDHHC5 palmitoylates EZH2 and modulates its methylation status in p53-mutant glioma.

Dysregulation of DHHC family proteins contribute to cancer development (13, 14). The ZDHHC5 gene is located in a region of chromosomal instability, and is upregulated in both breast cancer and lung adenocarcinoma (22). However, because of its importance in neural development, there have been no previous studies on the role of ZDHHC5 in gliomagenesis.

The results of this study demonstrate that p53 controls ZDHHC5 promoter transactivation in glioma, suggesting a link between ZDHHC5 expression, p53 regulation, and tumorigenicity. We show here for the first time that ZDHHC5 mRNA and protein levels increase with glioma grade, which is associated with an increased frequency of p53 mutation. To confirm the functional link between ZDHHC5 and p53, we examined whether mutant p53 modulated ZDHHC5 expression, as most cancer-associated p53 mutations are associated with a loss of its transcriptional activity. Mutant p53 was capable of physical interaction with the transcription factor NF-Y; the gain-of-function induced by this protein complex was independent of the type of p53 mutation (27, 28). Moreover, this interaction resulted in aberrant upregulation of ZDHHC5, whose promoter contains four potential NF-Y–binding sites. It is worth noting that the presence of mutant p53 in the ZDHHC5 promoter regions was critical for recruitment of p300, whose acetylase activity may be the key event underlying the transcriptional activity of the mutant p53/NF-Y protein complex (27).

NSC-rich regions in the brain are more susceptible to malignant transformation (24, 31). For example, GFAP and Nestin-Cre–mediated inactivation of the tumor suppressors neurofibromin 1 and p53 induced glial progenitor proliferation and glioma progression to malignant astrocytomas (31). ZDHHC5 is highly expressed in early embryonic cortical neural precursor cells but is undetectable in most adult tissues, suggesting that it is associated with an undifferentiated state, as in malignant tissues. In this study, we showed that ZDHHC5 overexpression and mutations in KRAS, TERT, and p53 oncogenes in human stem cells was sufficient for full and rapid malignant transformation. We examined whether the growth of established tumors requires ZDHHC5 by knocking down ZDHHC5 in CA-KRAS/TERT/DN-p53 tumors, and found that ZDHHC5 deficiency led to tumor suppression. ZDHHC5 was necessary but not sufficient for cell transformation, as ZDHHC5 overexpression did not in itself induce tumorigenesis (data not shown).

Recent studies indicate that a population of cells with stem-like properties may contribute to glioma repopulation after therapeutic intervention (32–34). Several genes involved in normal stem cell maintenance such as Oct4, Sox2, and CD133 have been shown to increase glioma malignancy (35). In this study, we observed high levels of ZDHHC5 in GSCs, as ZDHHC5 is coexpressed with Sox2 or CD133 in GSCs. ZDHHC5 is highly expressed in neural progenitor cells but is rapidly degraded during differentiation (23). Thus, ZDHHC5 may be involved in the maintenance of glioma cell stemness. Interestingly, we found that EZH2 mediated ZDHHC5-dependent glioma progression. EZH2 is expressed in various malignancies, including glioma, and its silencing in glioblastoma tumor–initiating cells was shown to inhibit its proliferation and tumorigenic potential (36, 37). We demonstrated that ZDHHC5 inhibited EZH2, which activates the expression of other pluripotency-associated transcription factors. Thus, ZDHHC5 prevents the self-renewal of glioma-initiating stem cells via suppression of EZH2.

Given its role as a methyltransferase that targets H3K27, EZH2 is considered as a repressor of tumor suppressors such as p19, B-cell lymphoma-2-interacting protein, p57, E-cadherin, and Runt-related transcription factor 3 (38). Recent proteomic and biochemical studies have reported that EZH2 phosphorylation at S21 inhibits H3K27 trimethylation and consequently, expression of polycomb repressive complex 2 target genes (39). The inhibitory effect of ZDHHC5 on the malignant progression of gliomas is associated with changes in the S21 phosphorylation status rather than the expression level of EZH2. Notably, EZH2 S21 phosphorylation was inversely correlated with the level of EZH2 palmitoylation by ZDHHC5 at Cys 571 and Cys 576. Palmitoylation is required for EZH2 targeting to the Golgi apparatus, where phosphorylation occurs. ZDHHC5 is localized to detergent-resistant microdomains of the plasma membrane (18); palmitoylation is necessary for vesicle trafficking and for the formation of EZH2 multiprotein complexes (40).

In summary, the current study demonstrated that ZDHHC5 was upregulated in glioma as compared with normal brain tissue, which was correlated with p53 mutation. Mutant p53 protein transcriptionally regulated ZDHHC5 in conjunction with NF-Y. Moreover, ZDHHC5 was shown to contribute to the development of glioma by promoting the self-renewal and tumorigenicity of GSCs. ZDHHC5 palmitoylation of EZH2 altered the phosphorylation status of EZH2 S21. These findings indicate that ZDHHC5 is a potential therapeutic target for the treatment of p53-mutated glioma.

No potential conflicts of interest were disclosed.

Conception and design: X. Chen, Z. Fang

Development of methodology:

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H. Ma, Z. Wang, S. Zhang, H. Yang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Z. Wang, H. Yang

Writing, review, and/or revision of the manuscript: X. Chen, Z. Fang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H. Ma, Z. Fang

Study supervision: Z. Fang

The authors thank the Center of Medical Physics and Technology, Hefei Institutes of Physical Science for technical assistance and Dr. Aijun Hao at Shandong University School of Medicine for providing glioma cell lines.

This work was supported in part by grants from the National Natural Science Foundation of China (grant nos. 31501171 and 31571433) and Anhui Provincial Natural Science Foundation (grant nos. 1608085MH180 and 1508085SMC214).

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.