Overexpressed histone acetyltransferase 1 regulates cancer immunity by increasing programmed death-ligand 1 expression in pancreatic cancer

All pancreatic cancer cell lines including PANC-1, BxPC-3 and MIA PaCa-2 were purchased from the Chinese Academy of Science Cell Bank, and the Panc 02 cells were obtained from Tong Pai Technology (Shanghai, China). These cell lines were cultured in Dulbecco’s Modified Eagle Medium (DMEM) medium (Invitrogen, USA) supplemented with 10% fetal bovine serum (FBS) (HyClone, USA). All cell lines were routinely maintained at 37 °C in a 5% CO₂ incubator.

Mammalian expression vectors for Flag-HAT1 recombinant proteins were generated using the pcDNA3.1 backbone vector. The HAT1 antibody (ab194296) was purchased from Abcam (working dilution 1:2000); beta-tubulin (2128S) was from Cell Signaling Technology - (working dilution 1:5000); BRD4 (ab128874) was from Abcam (working dilution 1:1000); PD-L1 (13684S) was from Cell Signaling Technology (working dilution 1:1000); and H4K5ac (ab17343) was from Abcam (working dilution 1:1000). Ascorbate was purchased from Sigma-Aldrich (Shanghai, China).

The ethics of using human tissue (12 pairs of matched pancreatic cancer/adjacent noncancerous tissues) was approved by the local ethics committee (Tongji Medical College, China), and written informed consent was obtained from patients prior to surgery exactly as described previously [18]. The cells or the tissue specimens were lysed with lysis buffer (Beyotime, China) containing 1% protease and phosphatase inhibitors. The protein concentration was determined with a protein assay kit (Pierce Biotechnology, USA). Equal amounts of protein for each sample were separated using SDS-PAGE gels and transferred onto PVDF membranes (Pierce Biotechnology, USA). The membranes were subsequently blocked in 5% not-fat milk for 1 h at room temperature, followed by incubation with primary antibody overnight at 4 °C. The membranes were then washed with 1x TBST and incubated with a secondary antibody for 1 h. Finally, the membranes were treated with ECL detection reagents and exposed to X-ray films.

Total RNA was extracted from the cells using Trizol reagent (Thermo Fisher Scientific, USA). First strand cDNA was synthesized from 2 μg RNA using a cDNA Reverse Transcription kit (PrimeScript™ RT reagent Kit, Code No. RR037A), and real-time PCR analysis was carried out with a PCR kit (TB Green™ Fast qPCR Mix, Code No. RR430A) according to the manufacturer’s protocols. The two kits were purchased from Takara Bio Inc. (Shigo, Japan). All the values were normalized to actin, and the 2-ΔCt method was used to quantify the fold change. The primers used for RT-qPCR are provided in Additional file 1: Table S1.

ChIP was performed following the manufacturer’s instructions for the Chromatin Extraction Kit (Abcam, ab117152, USA) and ChIP Kit Magnetic - One Step (Abcam, ab156907, USA) [19]. BRD4 (Cell Signaling Technology, 13,440, dilution 1:50) was used for the ChIP assay. The purified DNA was analyzed by real-time PCR with a PCR kit (Takara Bio Inc., Japan) according to the manufacturer’s protocols [20]. The primers for ChIP-qPCR are provided in Additional file 1: Table S2.

The tissue microarray slides were purchased from Outdo Biobank (Shanghai, China) (HPan-Ade060CD-01). The tissue microarray specimens were immunostained with PD-L1 (Cell Signaling Technology, 13,684, dilution 1: 1000) and HAT1 antibodies (Abcam, ab194296, dilution 1:3000) as described previously. Staining intensity was scored in a blinded fashion: 1 = weak staining at 100× magnification but little or no staining at 40× magnification; 2 = medium staining at 40× magnification; 3 = strong staining at 40× magnification [21]. The degree of immunostaining was reviewed and scored by two independent pathologists who were blinded to the clinical details. The scores were determined by the percentage of positive cells multiplied by the staining intensity.

The lentivirus-based control and gene-specific shRNAs were purchased from Sigma-Aldrich. Lipofectamine 2000 was used to transfect 293 T cells with shRNA plasmids and viral packaging plasmids (pVSV-G and pEXQV). Twenty-four hours after transfection, the medium was replaced with fresh DMEM, containing 10% FBS and 1 mM of sodium pyruvate. Next, 48 h post transfection, the virus culture medium was collected and added to the PANC-1, MIA PaCa-2 and BxPC-3 cells supplemented with 12 μg/ml of polybrene. Twenty-four hours after infection, the infected cells were selected with 10 μg /ml of puromycin. The shRNA sequence information is provided in Additional file 1: Table S3.

The pTsin lentiviral expression vector was used to generate lentiviral plasmids for pTsin-Flag-HAT1. Lipofectamine 2000 was used to transfect 293 T cells with the pTsin expression plasmid and viral packaging plasmids (pHR’ CMVδ 9.8 and pVSV-G). Twenty-four hours after transfection, the medium was replaced with fresh DMEM, containing 10% FBS and 1 mM of sodium pyruvate. Next, 48 h post transfection, the virus culture medium was collected and added to PANC-1 cells supplemented with 12 μg/ml of polybrene. Twenty-four hours after infection, the infected cells were selected with 10 μg/ml of puromycin.

Cell viability was evaluated using the MTS (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt) assay according to the manufacturer’s instructions (Abcam, USA). Briefly, the pancreatic cancer cells (1 × 10³ cells) were seeded in 96-well plates with 100 μl of culture medium. The cells were treated with serial concentrations of small molecular inhibitors. After 72 h, 20 μl of MTS reagent (Abcam, USA) was added to each well of the cells and incubated for 1 h at 37 °C in standard culture conditions. The absorbance was measured in a microplate reader at 490 nm.

The BALB/c-nu mice (4–5 weeks of age, 18–20 g) were purchased from Vitalriver (Beijing, China) and randomly divided into two groups (n = 7/group) for the subcutaneous inoculation with 5 × 10⁶ of PANC-1 cells infected with shControl or shHAT1 lentivirus in the left dorsal flank of the mice. The tumors were examined every other day for 21 days; the length and width measurements were obtained with calipers to calculate the tumor volumes by using the eq. (L x W²)/2. On day 21, the animals were euthanized, and the tumors were excised and weighed. All the animal experimental procedures were approved by the Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology.

The online database Gene Expression Profiling Interactive Analysis (GEPIA, http://​gepia.​cancerpku.​cn/​index.​html.​) [22] was used to analyze the RNA sequencing expression data related to our project based on The Cancer Genome Atlas (TCGA) and the Genotype-Tissue Expression (GTEx) projects. GEPIA performs survival analyses based on gene expression levels and uses a log-rank test for hypothesis evaluation. GEPIA performs a pairwise gene correlation analysis for any given set of TCGA and/or GTEx expression data using Pearson correlation statistics.

Six-week-old C57BL/6 mice were purchased from Charles River Laboratories (Wuhan, China). All the animal experimental procedures were approved by the Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology. Panc 02 cells (5 × 10⁶ in 100 μl 1 × PBS) infected with shControl or shHAT1 lentivirus were injected s.c. into the right flank of mice. The volume of xenografts was measured every other day and calculated using the formula LxW²x0.5. After the xenografts reached a size of approximately 50 mm³, mice carrying similar types of tumors were randomized into different groups and treated with anti-PD-1 (BioXcell, Clone RMP1–14)/IgG (BioXcell, Clone 2A3) (200 μg, i.p., given at days 0, 3, 6); or anti-PD-L1 (BioXcell, Clone 10F.9G2)/IgG (BioXcell, Clone MPC-11) (200 μg, i.p., given at days 0, 3, 6). Mice were euthanized and tumors were collected from all animals once the tumors reached a volume of 200 mm³.

The PANC-1, MIA PaCa-2 and BxPC-3 cells infected with control or HAT1-specific shRNAs were harvested and washed with PBS. Cells were fixed in 4% paraformaldehyde for 15 min. After washing with PBS, cells were incubated with ice-cold 100% methanol for 30 min on ice. Cells were washed with PBS and incubated with PD-L1 antibody (Biolegend, APC anti-human CD274, clone 29E.2A3) or isotype IgG (Biolegend,APC anti-human IgG Fc Antibody, clone HP6017) for 15 min at room temperature. After washing three times with PBS, the cells were resuspended in PBS and analyzed by flow cytometry.

For flow cytometry analysis of the mouse tissue samples, the tumors were cut into small pieces and digested with 2 mg/ml collagenase (Sigma, USA) in DMEM for 1 h at 37 °C. Cells were filtered through a 70 μm nylon strainer and resuspended in red blood cell lysis buffer (Biolegend) for 3 min at room temperature. The cells were then suspended in PBS with 2% BSA and costained with the following antibodies: CD45 (Biolegend, 103,112, APC conjugated); CD4 (Biolegend, 100,510, FITC conjugated); CD8 (Biolegend, 100,708, PE conjugated); CD11b (Biolegend,101,212, APC conjugated); and Gr1 ((Biolegend, 108,406, FITC conjugated)). After incubation with antibody for 15 min, the cells were washed with PBS and analyzed by flow cytometry.

To investigate the expression level of HAT1 in pancreatic cancer, we first analyzed HAT1 mRNA levels in pancreatic cancer and nontumor pancreatic tissues by using the GEPIA web tool [22]. We found that the mRNA levels of HAT1 in pancreatic cancer tissues were higher than in nontumor pancreatic tissues (Fig. 1a). Then, we sought to determine the HAT1 protein levels in human PDAC specimens via using the TMA (tissue microarray) approach. We examined the protein level of the HAT1 by immunohistochemistry (IHC) in PDAC specimens obtained from a cohort of patients (n = 25 normal pancreatic specimens, n = 41 PDAC TMA specimens). The IHC staining score was evaluated by measuring both the percentage of cells that stained positive for the marker and the staining intensity [20]. We showed that HAT1 was significantly overexpressed in PDAC specimens compared with normal pancreatic tissues (Fig. 1b and c). Similarly, we examined the protein level in PDAC and paired adjacent nontumor pancreatic tissues in our hospital via Western blot analysis and demonstrated that HAT1 was upregulated in PDAC compared to adjacent non-tumor pancreatic tissues (Fig. 1d and e). To further identify the clinical relevance of HAT1 in pancreatic cancer, the survival rate of PDAC patients associated with HAT1 expression was determined through GEPIA web tool and The Human Protein Atlas. Our data indicated that the high expression of HAT1 was closely correlated with poor prognosis in PDAC patients (Fig. 1f-g). Together, these data suggest that HAT1 is upregulated in PDAC and associated with poor prognosis in PDAC patients.

Given that HAT1 functioned as a negative prognostic biomarker in PDAC, we wanted to explore the specific role of HAT1 in pancreatic cancer. First, we knocked down HAT1 with a specific lentiviral short hairpin RNA in PANC-1, MIA PaCa-2 and BxPC-3 cells (Fig. 2a). The MTS assay and colony formation assay indicated that the knock down of HAT1 significantly impeded the cell growth of the PANC-1, MIA PaCa-2 and BxPC-3 cells (Fig. 2b and c). On the other hand, we also found that the overexpression of HAT1 promoted the proliferation of PANC-1 and BxPC-3 cells (Additional file 1: Figure S1a and b). The above data were consistent with the data reported for liver, nasopharyngeal and lung cancer cells [15‐17]. Moreover, to investigate the role of HAT1 in the tumor growth of PDAC in vivo, PANC-1 cells infected with control or HAT1-specific shRNAs were injected subcutaneously into the right flank of nude mice for the xenograft assay. We found that the knockdown of HAT1 blocked the growth of PANC-1 xenografts in mice (Fig. 2d-f). Then, xenografts were subjected to IHC analysis for Ki-67 expression, the most commonly used indicator to evaluate cell proliferation (Fig. 2g). We found that the knockdown of HAT1 resulted in a decrease in Ki-67 staining compared with the control group (Fig. 2h). Furthermore, the PANC-1 cells infected with pTsin-EV or pTsin-Flag-HAT1 used to establish the control or HAT1-overexpressing pancreatic cancer stable cell lines, respectively, were injected subcutaneously into the right flank of nude mice for the xenograft assay. Our data demonstrated that overexpressed HAT1 promoted pancreatic cancer growth in vivo (Additional file 1: Figure S1c-e). Taken together, our findings indicate that HAT1 acts as a growth promoting protein in pancreatic cancer.

Since HAT1 was upregulated in PDAC and promoted cell proliferation in pancreatic cancer cells, the more biological role of HAT1 needs to be explored. It has been documented that PD-L1 was inversely correlated with prognosis in pancreatic cancer [23]. Given the poorly understood nature of its regulation [24], we sought to determine whether HAT1 is involved in PD-L1 regulation. Strikingly, the knockdown of HAT1 decreased the protein and mRNA expression of PD-L1 in the PANC-1, MIA PaCa-2 and BxPC-3 cells (Fig. 3a-c). Conversely, using a gain of function approach, we demonstrated that the ectopic expression of HAT1 led to the up-regulation of PD-L1 expression in both PANC-1 and MIA PaCa-2 cells (Fig. 3d and e). A recent study reported that HAT1 was downregulated by ascorbate through the TET-mediated DNA hydroxymethylation pathway [25]. Similarly, the ascorbate treatment decreased the protein and mRNA levels of both PD-L1 and HAT1 in the PANC-1 and BxPC-3 cells (Fig. 3f and g). Then, we sought to examine whether the effect of ascorbate on PD-L1 expression is mediated by HAT1 or not. The ascorbate treatment decreased PD-L1 expression, and this effect was diminished after HAT1 knockdown in BxPC-3 cells (Fig. 3h and i), which indicated that HAT1 played a key role in modulating the ascorbate-induced PD-L1 down-regulation of PD-L1. Taken together, our findings indicated that HAT1 upregulated PD-L1 expression at the transcriptional level.

To further investigate the relationship between PD-L1 and HAT1, we analyzed the mRNA levels of PD-L1 (CD274) and HAT1 in a subset of pancreatic cancer patients (Fig. 4a) [26]. Intriguingly, we found that the overexpression of PD-L1 (CD274) was accompanied by the upregulation of HAT1 (p < 0.05) (Fig. 4a). Then, we analyzed the correlation of the mRNA level between PD-L1 and HAT1 using the GEPIA web tool. Our results indicated that PD-L1 mRNA is positively correlated with HAT1 mRNA in pancreatic cancer patient specimens (Pearson’s product-moment correlation coefficient r = 0.47, p = 4.4e-11) (Fig. 4b). To further determine the correlation between HAT1 and PD-L1 in pancreatic cancer specimens, we examined the expression of these two proteins by performing immunohistochemistry (IHC) on a tissue microarray (TMA) containing a cohort of pancreatic cancer samples (n = 41). The IHC staining index (SI) was calculated by multiplying the percentage of positively stained cells and the staining intensity. Representative images of high and low/no staining of HAT1 and PD-L1 are shown in Fig. 4c. The PD-L1 expression was positively correlated with the HAT1 level (Pearson’s product-moment correlation coefficient r = 0.4776, p = 0.0016) (Fig. 4d), which was consistent with the mRNA level correlation reported above. Together, our results suggest that PD-L1 is positively correlated with HAT1 in PDAC patient specimens.

As described above, HAT1 regulated PD-L1 expression in human pancreatic cancer cells. We sought to examine this phenomenon in vivo. First, we knocked down of Hat1 in the murine pancreatic cancer cell line Panc 02 and showed that reducing the Hat1 resulted in decreasing the Pd-l1 (Fig. 5a and b). These results suggest that the knockdown of Hat1 may enhance the immune checkpoint blockade efficacy in pancreatic cancer therapy. To test this hypothesis, Panc 02 cells infected with shControl or shHat1 lentivirus were injected subcutaneously into immune-proficient mice. Panc 02 tumor-bearing mice were sham-treated or treated with anti-PD-1 antibody or nonspecific control IgG as indicated in Fig. 5c. Consistent with the previous results in human pancreatic cancer cells (Fig. 2), the knockdown of Hat1 slowed down the cells proliferation of murine pancreatic cancer cells in vivo (Fig. 5d and e). In agreement with the observation that Pd-l1 protein was detectable in Panc 02 cells (Fig. 5a), the anti-PD-1-antibody treatment impeded tumor growth and prolonged the survival time of tumor-bearing mice (Fig. 5d and e). Furthermore, the knockdown of Hat1 significantly increased tumor infiltration of immune effectors including CD45⁺CD8⁺ T cells and CD45⁺CD4⁺ T cells, but decreased the infiltration of CD11b⁺Gr1⁺ myeloid cells in tumors (Fig. 5f). In agreement with the effect on tumor regression, treatment with the anti-PD-1-antibody after the knockdown of Hat1 led to a further decrease in tumor growth and an increase in CD45⁺CD8⁺ and CD45⁺CD4⁺ T cell infiltration, but a greater decrease in CD11b⁺Gr1⁺ myeloid cells in tumors (Fig. 5f). Therefore, our findings indicate that the knockdown of HAT1 improves the efficacy of immune checkpoint blockade by decreasing the PD-L1 expression and in vivo.

Although HAT1 transcriptionally increased PD-L1 expression, but the underlying mechanism is not fully understood. It has been documented that various transcriptional factors, such as STAT3, MYC, p65 and BRD4, could directly bind to the promoter of PD-L1 and regulate the transcription of PD-L1 in cancer cells. Moreover, HAT1 catalyzes the acetylation of H4K5 and H4K12, which is essential for BRD4 binding to the histone H4 and initiating the transcription. Herein, we hypothesized that BRD4 was the key protein that mediates the HAT1-induced PD-L1 expression. To test this hypothesis, we knocked down HAT1 in PANC-1 cells and treated them with or without the BRD4 inhibitors JQ1 (Fig. 6a and b). We found that the knockdown of HAT1 repressed PD-L1 expression and that these effects were diminished by JQ1 treatment (Fig. 6a and b). Similarly, the knockdown of the HAT1 induced PD-L1 downregulation was not significant after the knockdown of BRD4 in PANC-1 cells (Fig. 6c and d). Conversely, the ectopic expression of HAT1 significantly increased the PD-L1 expression, but these effects were not apparent after the knockdown of BRD4 in PANC-1 cells (Fig. 6e and f). Notably, just as reported by other groups, we checked the existing BRD4 ChIP-seq data [27] and noticed that there is a BRD4 binding peak in the promoter of the PD-L1 gene (Fig. 6g). Thus, we confirmed the binding of BRD4 by ChIP-qPCR in PANC-1 cells (Fig. 6h). Consistently, JQ1 prevented BRD4 binding to the promoter of PD-L1 and abolished the effect of the HAT1 knockdown-induced reduction in BRD4 binding to the PD-L1 promoter (Fig. 6h). Collectively, these data suggest that HAT1 catalyzes the histone H4 acetylation and that the BRD4 complex binds to the acetylated H4 to initiate the transcription of PD-L1 (Fig. 6i).