HAT1 Promotes Gemcitabine Resistance by Regulating the PVT1/EZH2 Complex in Pancreatic Cancer

Background: The poor prognosis of pancreatic cancer is primarily due to the development of resistance to therapies, including gemcitabine. PVT1 has been shown to interact with EZH2, promoting gemcitabine resistance in pancreatic cancer. Methods: In this study, we assessed the ability of PVT1/EZH2 targeting to reverse resistance to gemcitabine in pancreatic cancer cells. MTS assay, colony formation assay, and mouse xenotransplantation experiments were used to evaluate the anti-tumor effects of gemcitabine in HAT1 knockdown or overexpressing pancreatic cancer cells. The relationship between HAT1 and PVT1 in pancreatic cancer was determined by RNA sequencing, quantitative real-time PCR, and chromatin immunoprecipitation. Co-immunoprecipitation, pull-downs, western blotting, and immunohistochemistry were used to assess the relationship between HAT1 and EZH2 in pancreatic cancer. Chitosan (CS)-tripolyphosphate (TPP)-siHAT1 nanoparticles were developed to evaluate their effects on the anti-tumor potential of gemcitabine in pancreatic cancer. Student’s t-test, one-way analysis of variance (ANOVA), or two-way ANOVA was used to evaluate statistical signicance. P-values <0.05 were considered statistically signicant. All values were expressed as means ± SD. Results: Our results showed that the aberrant HAT1 expression promoted gemcitabine resistance in pancreatic cancer cells. We also found that HAT1 enhanced the binding of BRD4 to the PVT1 promoter, thereby promoting PVT1 transcription. Moreover, we found that HAT1 prevented EZH2 degradation by interfering with UBR4 binding to the N-terminal domain of EZH2, thus maintaining EZH2 protein stability. Finally, we showed that CS-TPP-siHAT1 nanoparticles augmented the anti-tumor effects of gemcitabine in pancreatic cancer cells in vitro and in vivo. Conclusions: Our ndings suggest that by increasing the levels of the PVT1/EZH2 complex, HAT1

proteomics and next-generation RNA sequencing, suggest that numerous proteins mediate gemcitabine resistance [11]. For instance, aberrant expression of enhancer of zeste homolog 2 (EZH2) in pancreatic cancer cells has been linked to gemcitabine resistance, possibly due to the downregulation of the tumor suppressor p27 Kip1 [12]. Additionally, silencing of the long noncoding RNA (lncRNA) PVT1 increased gemcitabine sensitivity in pancreatic cancer cells [13,14]. Interestingly, PVT1 has been found to form a complex with EZH2, a key step in the development of gemcitabine resistance in pancreatic cancer [14].
We have previously shown that aberrant expression of histone acetyltransferase 1 (HAT1) enhanced PD-L1 expression and promoted pancreatic cancer cell proliferation by modulating the function of BRD4. Herein, we show that HAT1 knockdown in pancreatic cancer cells increases gemcitabine sensitivity and decreases PVT1/EZH2 complex levels, suggesting that HAT1 may represent a promising therapeutic target in pancreatic cancer.

Cell lines and cell culture
Pancreatic cancer cell lines (PANC-1, BxPC-3, and MIA PaCa-2) were purchased from the Chinese Academy of Science Cell Bank. Cell lines were cultured in RPMI 1640 medium (Gibco, USA) containing 10% fetal bovine serum (FBS; Gibco, USA), 100 IU/mL penicillin, and 100 µg/mL streptomycin (Gibco, USA). Plasmocin (InvivoGen) was routinely added to the medium to eliminate mycoplasma. All cell lines were maintained at 37 °C in a 5% CO 2 incubator.

Lentiviral and the construction of stable cell lines
Lentiviral particles carrying gene-speci c shRNAs and negative control shRNA (hU6-MCS-CBh-gcGFP-IRES-puromycin) were used to infect pancreatic cancer cells. Brie y, pancreatic cancer cells were seeded in 6-well plates (2 × 10 5 cells per well). The day after, 1 mL of DMEM medium containing the viral particles and 40 µL of Hitrans G reagent was added to each well. After 16 hours, the viral solution was replaced with complete growth medium. Puromycin was used to select infected cells.
Cells were seeded in 6-well plates (500 cells/well) for colony formation assays. The cell growth medium was replaced every 3 days. After 2 weeks, cells were xed with 4% paraformaldehyde (30 minutes) and stained with crystal violet (20 minutes).

The preparation of Gemcitabine-Resistant PANC-1
Gemcitabine was dissolved in dimethyl sulfoxide (DMSO) at a concentration of 1 mM. Then, gemcitabine solution was diluted to a working concentration of 50 nM with RPMI-1640 medium containing 10% FBS. The pancreatic cancer cell line PANC-1 was cultured with the above con gured medium for 2 weeks in 5% CO2 at 37 °C. Then the cells were cultured for another 2 weeks without gemcitabine. The above process was repeated 3 times. The selected cells could resist gemcitabine and the cells were named GR-PANC-1.
The cell lysate was centrifuged at 12000 rpm and 4 °C, and the supernatant was collected and boiled for 10 minutes at 95 °C. The protein concentration was measured using Lowry method with protein concentration determination kit (Solarbio, PC0030). Subsequently, proteins were resolved by SDS-PAGE and transferred onto PVDF membranes (Pierce Biotechnology, USA). Membranes were blocked with 5% skim milk for 1 hour at room temperature and incubated with primary antibodies overnight at 4 °C. The next day, the membranes were washed 3 times and incubated with the respective secondary antibody for 1 hour at room temperature. Signal was developed using ECL chemiluminescence reagent and X-ray lms.

Quantitative real-time PCR (qRT-PCR)
Total RNA was extracted using TRIzol reagent (15596026, Invitrogen, USA). PrimeScript™ RT kit (Takara, Japan) was used to synthesize cDNA, which was subsequently ampli ed using TB Green™ Fast qPCR Mix (Takara, Japan). Relative gene expression was calculated using the 2 −ΔCq method after normalizing to GAPDH levels. The sequences of the qRT-PCR primers are shown in Table S2. 2.10 Co-immunoprecipitation (Co-IP) and chromatin immunoprecipitation (ChIP) Cells were lysed in western/IP lysis buffer for 30 minutes at 4 °C, and the cell lysate was centrifuged for 10 minutes at 12000 rpm and 4 °C. The supernatant was incubated with protein A/G agarose beads (Thermo Fisher Scienti c, USA) overnight at 4 °C with gentle agitation. The protein A/G agarose beads were collected and washed 7 times with western/IP lysis buffer. Subsequently, proteins were resolved by SDS-PAGE.
According to the manufacturer's protocols, chromatin Immunoprecipitation (ChIP) assay was carried out through utilizing the Chromatin Extraction Kit (Abcam, ab117152, USA) and ChIP Kit Magnetic-One Step (Abcam, ab156907, USA) [15,16]. BRD4 antibody (Cell Signaling Technology, 13440, working concentration 1:50) was used to perform ChIP assay for precipitating the promoter of PVT1. The precipitated DNA fragments were further ampli ed via quantitative real-time PCR with a PCR kit (Takara Bio Inc., Japan) based on the manufacturer's instructions [15,17]. The sequences of the primers used for ChIP-qPCR are provided in Table S3.

RNA immunoprecipitation (RIP) assay
According to the manufacturer's protocols, RNA immunoprecipitation (RIP) assay was performed through using Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore, MA, USA). The supernatant of cell lysate was extracted and then mixed up with treated-beads to incubate for 6 h. Then the RIP wash buffer was used to wash the beads for 7 times. At last, qRT-PCR analysis was conducted to analyze the puri ed RNA.
2.12 RNA pull-down assay mMESSAGE mMACHINE T7 Kit (Ambion, USA) and RNeasy Mini Kit (Qiagen, Valencia, CA) were used to transcribe PVT1 and the antisense RNA in vitro, and Pierce RNA 3′ End Desthiobiotinylation Kit (Thermo Scienti c, USA) was utilized to conduct biotin labeling. 1 mg total protein extracts were added to 50 pmol of biotin-labeled PVT1 for incubating 1 h, then mixed up with 60 µL of Streptavidin Beads (Invitrogen) for another 1 h. Finally, Western blotting analysis was used to detect the proteins.

Flow cytometry
Annexin V-FITC/PI kit (AntGene, China) was used to assess apoptosis. Cells treated with different agents were harvested and washed with phosphate buffer saline (PBS). Cells were incubated with annexin V-FITC and PI according to the manufacturer's instructions. Flow cytometry was performed on BD FACSCelesta (BD Biosciences, USA), and the data were analyzed using FlowJo 2.14 Confocal imaging Pancreatic cancer cells were seeded into slide chambers, and after overnight incubation, cells were incubated with chitosan (CS)-tripolyphosphate (TPP)-siHAT1 (10 µg/mL) for 4 hours. Cells were washed with PBS, xed in 4% paraformaldehyde, and permeabilized with 0.2% Triton X-100 for 5 minutes.
Subsequently, cells were treated with 100 nM FITC-phalloidin for 1 hour at 37 °C and in the dark. Cell nuclei were counterstained with Hoechst 33342 for 5 minutes. Stained cells were observed under a laser confocal microscope.

Subcutaneous xenotransplantation tumor model
Nude mice (18-20 g, 5 weeks old) were purchased from Vitalriver (Beijing, China) and maintained under speci c pathogen-free (SPF) conditions. Mice were randomly divided into groups (5 mice per group) and subcutaneously injected with 3 × 10 6 PANC-1 cells. The volume was measured every 3 days (volume = (L × W 2 )/2). The treatment protocols with drugs were as follows: When the tumors reached to 50mm 3 , experimental groups were injected intraperitoneally gemcitabine (10 mg/kg) or CS-TPP-siHAT1 (50 µM) with every three days, control groups were injected intraperitoneally equal DMSO or CS-TPP. After 30 days, mice were sacri ced, and tumors were excised and weighed. All animal experimental procedures were approved by the Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology.

Statistical analysis
All data were analyzed using GraphPad Prism 7.0. Student's t-test, one-way analysis of variance (ANOVA), or two-way ANOVA was used to evaluate statistical signi cance. P-values < 0.05 were considered statistically signi cant. All values were expressed as means ± SD.

Aberrant HAT1 expression promotes gemcitabine resistance in pancreatic cancer cells
HAT1 is often upregulated in pancreatic cancer and promotes pancreatic cancer cell proliferation by regulating PD-L1 expression; however, the oncogenic role of HAT1 in pancreatic cancer remains poorly understood. Several studies had found that the expression of histone acetyltransferase (HAT) genes was related to drug resistance [18][19][20]. As is well-known, pancreatic cancer was easily tolerated by chemotherapy, while up-regulated HAT1 might led to the results. Thus, we explored the in uence of abnormal HAT1 expression on the sensitivity of several commonly used drugs in pancreatic cancer cells, which showed the signi cantly reduced IC50 for gemcitabine when HAT1 was knocked down, while the sensitivity of other drugs was slightly reduced or unchanged (Fig. 1a). Furtherly, HAT1 silencing increased three pancreatic cancer cell lines sensitivity to gemcitabine; HAT1 overexpression had the opposite effect ( Fig. 1b and Fig. S1a). The expression of HAT1 also was up-regulated obviously in pancreatic cancer cells compared to the normal cells (Fig. S1b), which might be the potential cause of gemcitabine resistance.
Additionally, MTS and colony formation assays revealed that HAT1-knockdown cells grew slower than control cells in the presence of gemcitabine (Fig. 1c, d and Fig. S1c, d). Moreover, HAT1 silencing in pancreatic cancer cells enhanced apoptosis in response to gemcitabine treatment (Fig. 1e, f and Fig.  S1e). Consistently, HAT1 knockdown enhanced the tumor-suppressive effects of gemcitabine in tumor xenografts (Fig. 1g). Speci cally, tumors with HAT1 knockdown exhibited the slowest tumor growth, the highest caspase-3 levels, and the lowest Ki67 levels in response to gemcitabine treatment ( Fig. 1h-1j).
Besides, we constructed the gemcitabine resistant PANC-1. The cell morphology changed signi cantly and the sensitivity reduced by 7.79 times compared to ordinary PANC-1 ( Fig. S1f and S1g). GR-PANC-1 was also used to carry out MTS, clone formation and caspase-3 activity detecting (Fig. S1h-S1j). The results showed that silencing HAT1 could recover the gemcitabine sensitivity. Moreover, the expression of HAT1 increased in GR-PANC-1, which further illustrated HAT1 promoted the gemcitabine resistance of pancreatic cancer cells (Fig. S1k). These ndings suggest that aberrant HAT1 expression affects the response of pancreatic cancer cells to gemcitabine in vitro and in vivo.
3.2 HAT1 enhances PVT1 expression by facilitating BRD4 binding to PVT1 promoter to promote gemcitabine resistance Despite evidence of the critical role of HAT1 in pancreatic cancer progression and gemcitabine resistance, the underlying mechanism remains elusive. RNA sequencing of HAT1 knockdown and control PANC-1 cells yielded 945 differentially expressed genes, including 490 upregulated and 455 downregulated genes ( Fig. 2a and Fig. S2a). A signi cant portion of these genes was directly or indirectly related to cancer (Fig.  S2b). Importantly, the lncRNA PVT1, a potential downstream target gene of HAT1, was downregulated upon HAT1 silencing, whereas HAT1-overexpressing cells exhibited elevated PVT1 levels ( Fig. 2d and Fig.  S2b, c). Furthermore, analyses using the GEPIA tool and ENCORI Pan-Cancer Analysis Platform indicated a positive correlation between HAT1 and PVT1 expression levels (Fig. 2b, c), suggesting that HAT1 regulates PVT1 expression in pancreatic cancer cells. A previous genome-wide screen identi ed PVT1 as a regulator of gemcitabine sensitivity in pancreatic cancer cells [13]. MTS assay con rmed that PVT1 was a drug-related resistant gene (Fig. S2d). To verify the relevance of PVT1 in HAT1-mediated gemcitabine resistance, we performed MTS assay and found that cells with HAT1 silencing and PVT1 overexpressing proliferated signi cantly faster than HAT1-knockdown cells (Fig. S2e). Thus, HAT1 promotes gemcitabine resistance by enhancing PVT1 expression.
Although our ndings demonstrate that HAT1 promotes PVT1 transcription, the underlying molecular mechanism is poorly understood. HAT1 could catalyze H4 acetylation (Fig. S3a, b), which is essential for the binding of the transcription activator BRD4 to histone H4 [21]. Besides, we have previously shown that HAT1 promoted PD-L1 expression in pancreatic cancer cells in a BRD4-dependent manner. We analyzed available ChIP-seq data of BRD4 [22] and identi ed a BRD4-binding peak in the promoter of PVT1 (Fig. 2f). This result was con rmed in pancreatic cancer cells by ChIP-qPCR (Fig. 2h). Moreover, PVT1 expression levels were decreased or increased after BRD4 silencing or overexpression, respectively ( Fig. 2g and Fig. S2f), suggesting that BRD4 may regulate the transcription of PVT1. Furthermore, analysis using GEPIA indicated a strong positive correlation between BRD4 and PVT1 levels in pancreatic cancer specimens (Fig. 2e). To assess the relevance of BRD4 in the HAT1-mediated regulation of PVT1 expression, we silenced BRD4 in combination with HAT1 knockdown or HAT1 overexpression. Interestingly, BRD4 silencing attenuated the ability of HAT1 to regulate PVT1 expression (Fig. 2i). Similarly, treatment with the BRD4 inhibitor JQ1 abrogated the ability of HAT1 silencing or overexpression to regulate PVT1 expression levels (Fig. 2j). The ability of BRD4 to bind to the PVT1 promoter was also signi cantly reduced upon JQ1 treatment, even in HAT1 overexpressing cells (Fig. 2k). Besides, the therapeutic effects of gemcitabine were signi cantly enhanced whether using JQ1 to inhibit BRD4 function or directly knocking down BRD4 (Fig. S3c, d), which further indicated that BRD4 could regulate the expression of PVT1. Collectively, these data suggest that the ability of HAT1 to induce PVT1 expression in pancreatic cancer cells requires BRD4 (Fig. 2l).

HAT1 stabilizes EZH2 to promote gemcitabine resistance by competing with UBR4 for binding to the N-terminal domain of EZH2
Several researches showed that EZH2 was a drug-related resistant gene [14,23,24], our research also con rmed EZH2 made pancreatic cancer cells become insensitive to gemcitabine (Fig. S4a). PVT1 had also been shown to bind EZH2 [25,26], it was consistent in pancreatic cancer cells. PVT1 did not change the mRNA level of EZH2 (Fig. S5a), but could bind to EZH2 protein (Fig. S5c, d). Thus, we tried to detect the correlation between HAT1 and EZH2. Firstly, the appropriate working concentration and duration of GSK126 (10 µM and 3 days) were screened in pancreatic cancer cells (Fig. S5b). We found that EHZ2 inhibition with GSK126 suppressed HAT1-mediated gemcitabine resistance ( Fig. 3a and Fig. S4b, c), suggesting that EZH2 is required for the ability of HAT1 to promote gemcitabine resistance. We also found that the protein but not the mRNA levels of EZH2 were decreased after HAT1 silencing in pancreatic cancer cells. Conversely, forced HAT1 expression increased the protein level of EZH2, although EZH2 mRNA levels remained unchanged (Fig. 3b, c and Fig. S4d, e). Tissue microarray of pancreatic cancer (n = 31) was used to conduct IHC analysis and ascertain the relationship between HAT1 and EZH2 in pancreatic cancer (Fig. S4f). The IHC score of HAT1 and EZH2 was calculated and summarized in a heatmap (Fig. 3d). We observed a positive correlation between the protein levels of HAT1 and EZH2 in pancreatic cancer tissues (Spearman correlation coe cient r = 0.5899, P = 0.0005; Fig. S4g).
Since HAT1 primarily regulated EZH2 expression at the post-transcriptional level, we hypothesized that HAT1 may regulate EZH2 protein stability. Although EZH2 protein levels were reduced after HAT1 silencing, treatment with the proteasome inhibitor MG132 restored EZH2 protein levels in HAT1 knockdown pancreatic cancer cells (Fig. 3e). Additionally, the half-life of EZH2 protein was signi cantly shorter in pancreatic cells with HAT1 knockdown than in control cells; HAT1 overexpression extended the half-life of EZH2 protein (Fig. 3f). Further, HAT1 silencing increased the polyubiquitination levels of EZH2, and HAT1 overexpression decreased EZH2 polyubiquitination (Fig. 3g). Co-immunoprecipitation assays revealed an interaction between endogenous HAT1 and EZH2 in pancreatic cancer cell lines (Fig. 3h). To identify the EZH2-binding domain of HAT1, we constructed two GST-EZH2 recombinant proteins: a wholelength GST-EZH2 and an N-terminus-lacking GST-EZH2 (GST-ΔN) (Fig. 3i). GST pull-down revealed that EZH2 lacking the N-terminal domain failed to interact with HAT1, indicating that the N-terminal domain of EZH2 is required for its binding to HAT1 (Fig. 3j). The N-terminal WD repeat domain of EZH2 is believed to be a ubiquitin ligase-binding motif [27,28]; as an E3 ubiquitin ligase, UBR4 may bind to the N-terminus of EZH2 to promote EZH2 polyubiquitination [29]. Thus, we hypothesized that HAT1 binds to the N-terminal domain of EZH2, competing with UBR4 binding. Supporting this hypothesis, HAT1 silencing increased the ability of UBR4 to bind EZH2, whereas HAT1 overexpression had the opposite effect (Fig. 3k). These data suggest that HAT1 competes with UBR4 for binding to the N-terminal domain of EZH2, thereby stabilizing EZH2 (Fig. 3l).

HAT1 stabilizes EZH2 by competing with UBR4 for binding to EZH2
We also found that UBR4 silencing profoundly increased the protein levels of EZH2 (Fig. 4a, b), as well as extended its half-life (Fig. 4c) and decreased its polyubiquitination levels (Fig. 4d, e), regardless of HAT1 silencing or overexpression. These data further con rmed that HAT1 stabilizes EZH2 by interfering with the binding of UBR4 to the N-terminus of EZH2.

Preparation and characterization of chitosan (CS)tripolyphosphate (TPP)-siHAT1
CS-TPP-siRNA nanoparticles were prepared using an ionic gelation method and through cross-linking the negatively charged phosphate groups of TPP with the positively charged amino groups of CS [30]. At a 5:1 ratio, CS-TPP formed small, positively charged nanoparticles [31]. Hence, we prepared CS-TPP-siRNA nanoparticles by adding 250 µg of siHAT1 to a solution containing 1 mg TPP. This mixture was added dropwise to a solution containing 5 mg CS, yielding CS-TPP-siHAT1 nanoparticles (Fig. 5a). Transmission electron microscopy indicated that CS-TPP-siHAT1 particles had a spherical structure (Fig. 5b). The cellular uptake of CS-TPP-siRNA was also investigated by laser scanning confocal microscopy; strong uorescent signals were observed in the cytoplasm (Fig. 5c). To assess the ability of CS-TPP-siHAT1 to suppress HAT1 expression, we incubated pancreatic cancer cells with different concentrations of CS-TPP-siHAT1 and for different times. A high concentration of CS-TPP-siHAT1 and prolonged incubation provided superior HAT1 downregulation (Fig. 5d, e).
3.6 CS-TPP-siHAT1 augments the ability of gemcitabine to inhibit pancreatic cancer cell growth Next, we assessed the ability of CS-TPP-siHAT1 to enhance the cytotoxic effects of gemcitabine and found that the combination of CS-TPP-siHAT1 with gemcitabine was more potent in inhibiting cell proliferation than CS-TPP-siHAT1 or gemcitabine alone (Fig. 6a, b and Fig. S6a, b). Additionally, Annexin-V/PI staining revealed that CS-TPP-siHAT1 augmented the pro-apoptotic effects of gemcitabine (Fig. 6c). In nude mice, CS-TPP-siHAT1 and gemcitabine were used to treat the nude, the detailed information was as follows (Fig. S6c). The combination of CS-TPP-siHAT1 with gemcitabine exhibited the most potent tumor-suppressive effects among all groups (Fig. 6d-f). The delaminated tumors were also extracted proteins to analyze the expression of HAT1 (Fig. S6d). Additionally, tumors from mice treated with CS-TPP-siHAT1 combined with gemcitabine showed the highest caspase-3 levels and lowest Ki67 levels (Fig. 6g). Moreover, GR-PANC-1 was also used to further attest the synergistic effect, which showed that gemcitabine suppressed the growth of GR-PANC-1 and promoted the apoptosis of GR-PANC-1 after inhibiting HAT1 with CS-TPP-siHAT1 ( Fig. S6e-g). These data suggested CS-TPP-siHAT1 and gemcitabine have a synergistic anti-tumor effect in pancreatic cancer.

Discussion
Despite recent progress in cancer therapeutics, the prognosis of pancreatic cancer remains poor, primarily due to the development of gemcitabine resistance. The mechanisms underlying resistance to gemcitabine remain unclear and involve alterations in drug transporters, proteases, transcription factors, and drug metabolism enzymes [32,33] [7]. These alterations can be internal to pancreatic cancer cells or induced by components of the tumor microenvironment [34].
HAT1 was rst identi ed as a classical B type histone acetyltransferase mediating the acetylation of histone H4 N-terminus [35]. As an epigenetic modi er, acetyl moieties can be found on lysine residues of cellular proteins (eg, histones, transcription factors, nuclear receptors, and enzymes). In addition to regulating gene expression, protein acetylation plays a critical role in replication-dependent chromatin assembly and DNA damage repair [36,37]. HAT1 mutations are frequent in tumors, leading to abnormal gene expression and promoting resistance to chemotherapeutic agents. For example, HAT1 was found to promote liver cancer cell proliferation and induce cisplatin resistance [38]. In melanoma, HAT1 was shown to catalyze histone H4 acetylation, thereby driving resistance to BET inhibitors [21]. In addition to HAT1 mutations, HAT1 overexpression is also frequent in multiple cancer types, including colon cancer [39], esophageal cancer [40], and lymphoma [41], exacerbating tumor malignancy. Hence, HAT1-mediated resistance to chemotherapy is an internal characteristic of tumor cells. We have previously shown that HAT1 was overexpressed in pancreatic cancer and that HAT1 silencing reduced the expression of PD-L1 on the surface of pancreatic cancer cells in a BRD4-dependent manner, improve the therapeutic e cacy of immune checkpoint blockade [42]. These ndings suggest that HAT1 may represent an important therapeutic target in pancreatic cancer.
In this study, we found that HAT1 overexpression in pancreatic cancer cells promoted gemcitabine resistance and that HAT1 silencing restored sensitivity to the anti-tumor effects of gemcitabine. To acquire further insight into the molecular mechanism underlying HAT1-mediated gemcitabine resistance, we knocked silenced HAT1 expression in PANC-1 cells and performed RNA sequencing, which indicated PVT1 as a potential HAT1 target gene. PVT1 is a lncRNA located on the human chromosome 8q24 [43,44] near the oncogene MYC gene, the expression of which is enhanced by PVT1 [14,44]. Mounting evidence suggests that PVT1 has oncogenic functions in various tumors. Notably, PVT1 enhanced Bcl2 expression in gastric cancer cells, thereby inhibiting apoptosis and promoting resistance to 5-uorouracil [45]. Additionally, PVT1 was found to upregulate the expression of numerous drug resistance-related molecules (eg, MDR1 and MRP1) and inhibit apoptosis signaling, promoting cisplatin resistance in colorectal cancer. Notably, a previous genome-wide screening identi ed PVT1 as a critical regulator of gemcitabine sensitivity in pancreatic cancer [13]. Here, we con rmed the role of HAT1 in regulating PVT1 expression and identi ed BRD4 as a key player of the HAT1-mediated regulation of PVT1 expression.
Previous studies have shown that by forming a complex with EZH2, PVT1 promoted cell proliferation and inhibited apoptosis in liver cancer and thyroid cancer cells [46,47]. The histone methyltransferase EZH2 suppresses gene expression by catalyzing the trimethylation of histone H3 lysine 27 (H3K27me3) [48,49]. EZH2 has also been implicated in multidrug resistance in gastric cancer and ovarian cancer [50,51]. Importantly, EZH2 has been shown to promote drug resistance by suppressing p27 Kip1 expression in pancreatic cancer cells [12]. Our study was consistent with previous research results, which showed EZH2 regulated the gemcitabine sensitivity and PVT1 could bind to EZH2. More importantly, we found that HAT1 enhanced the expression of EZH2 in pancreatic cancer cells in this study. We also found that HAT1 regulated EZH2 expression at the post-transcriptional rather than transcriptional level. Moreover, we found that HAT1 interacted with EZH2, increasing the stability of the latter. Jalan-Sakrikar N et al. found that the E3 ubiquitin ligase UBR4 could bind to the N-terminus of EZH2, promoting EZH2 ubiquitination and subsequent degradation [29]. Consistent with these results, we found that HAT1 could bind to the Nterminus of EZH2, interfering with the ability of UBR4 to interact with EZH2 (Fig. 9).
This study suggests that HAT1 regulates the sensitivity of pancreatic cancer to gemcitabine by regulating PVT1/EZH2 complex, highlighting the importance of HAT1 in the development of gemcitabine resistance in pancreatic cancer. Although we provide strong evidence that HAT1 inhibition may suppress tumor growth, reverse drug resistance, and improve the prognosis of pancreatic cancer, there are currently no HAT1 small molecule inhibitors. High-throughput screening approaches have led to the identi cation of potential drug candidates; however, most of these compounds exhibited moderate e cacy and speci city [52]. Therefore, future studies are urgently needed to develop speci c and potent HAT1 inhibitors. Nanoparticles carrying siRNA have emerged as promising alternatives of small molecule inhibitors [53][54][55]. Here, we used a CS-TPP carrier to deliver siHAT1 and suppress HAT1 expression in pancreatic cancer cells. CS-TPP-siHAT1 nanoparticles have proved effective in inhibiting HAT1 expression and augmenting the anti-tumor effects of gemcitabine in pancreatic cancer. The effectiveness and safety of CS-TPP-siHAT1 nanoparticles required further investigation in a clinical setting.

Conclusion
Our data strongly support that HAT1 upregulates PVT1 and promotes gemcitabine resistance in pancreatic cancer by enhancing BRD4 binding to the PVT1 promoter. We also show that HAT1 prevents EZH2 degradation by preventing UBR4 binding to the N-terminal domain of EZH2. Our ndings suggest that HAT1-induced gemcitabine resistance in pancreatic cancer may be mediated by the PVT1/EZH2 complex. Finally, we show that CS-PTT-siHAT1 nanoparticles suppress HAT1 expression and augment the anti-tumor effects of gemcitabine in pancreatic cancer cells. Collectively, the ndings presented here suggest that HAT1 may be a valuable therapeutic target in pancreatic cancer. Availability of data and materials All data generated or analysed during this study are included in this published article and its supplementary information les.

Competing interests
The authors declare that they have no competing interests.

Funding
This work was supported by grants from the National Natural Science Foundation of China (Grant No. 81702374 (X.J.)).
Authors' contributions XJ and HW conceived and designed this project, and reviewed and wrote the nal manuscript; YS, JS and DR designed and conducted all experiments, analyzed the data; JZ conducted RNA-sequence, generated the data and provided the results of RNA-sequence; YZ and BW generated the immune-histochemistry data and labelled the image; YG provided CS-TPP nanoparticles, generated the CS-TPP-siHAT1 mode pattern and took photos with electron microscope and confocal microscopy. PF conducted western blotting for H4 acetylation analysis