The role of DOT1L in the proliferation and prognosis of gastric cancer

Among the most ubiquitous malignancies is gastric cancer (GC), globally occupying the fifth rank in incidence and third in terms of mortality [1]. The number of new diagnoses was 1033701 with 782685 loss of lives in 2018, according to estimates [2]. Although the current comprehensive treatment based on surgery has made great progress, its clinical efficacy in patients is not significant, and the 5-year survival rate even now is below 30% [3]. Due to the lack of obvious clinical symptoms in early GC, most patients manifest advanced stages when diagnosed. Thus, it has been found that tumor markers with high sensitivity and specificity can improve the early discovery rate and improve the prognosis of GC.

With the rapid development of epigenetics, more and more small molecule drugs targeting histones and related regulatory proteins are being used for targeted experimental treatment in various tumors [4]. Disruptor of telomeric silencing-1-like (DOT1L) is an evolutionarily conserved protein, and its homolog DOT1 was first discovered in yeast through genetic screening. It is the only methyltransferase specific for the Lys⁷⁹ of histone H3 (H3K79me_x, x = 1, 2, or 3) [5]. According to recent reports, dot1l-mediated H3K79 methylation not only enables the transcriptionally active gene with the close association of its upstream and downstream regulatory genes along with regulating the cell cycle, chromatin silencing, and DNA damage repair [6–7]. In addition, the vital functioning of DOT1L in the mixed lineage leukemia (MLL)-rearranged leukemia has been shown [8], and thus DOT1L-targeted inhibitors have been used in rearranged leukemia [9]. In solid tumors, DOT1L is also involved in its progression, such as breast cancer [10], ovarian cancer [11], and lung cancer [12]. However, to date, the therapeutic value and the prognostic impact of DOT1L in GC has not been evaluated. This work involved a clinicopathological analysis and correlation with patient prognosis by examining the expression of DOT1L in GC with the observations showing that the molecule may serve as a new target for GC therapy.

All patients (n=101, age in the range 30–89 years, a mean age of 60.33 years) underwent subtotal gastrectomy between December 2012 and December 2013 in the First Affiliated Hospital of Bengbu Medical College. Their GC and corresponding adjacent normal tissues (distance > 4 cm) were taken from the Department of Pathology. The patients showed primary GC with no exposure to preoperative chemotherapy or radiotherapy. The study involved patient consent and received approval from the Ethics Committee of the above-mentioned institution.

Immunohistochemical studies (IHC) employed tissue sections fixed and embedded in formalin and paraffin, sequentially. In short, the slices were submerged in 0.01 mol citrate buffer and subjected to boiling for 10 min to repair the antigens. Normal goat serum (10%) was added for 30 min to block non-specific binding. This was followed by overnight incubation with DOT1L polyclonal antibody, H3K79me2 polyclonal antibody, cyclin-dependent kinase (CDK) 4 (CDK4) polyclonal antibody, and CDK6 polyclonal antibody (1:200; all antibodies were from Abcam; U.S.A.) at 4°C. This was followed by 30 min of incubation with goat anti-rabbit IgG (1:500; Abbkine; U.S.A.) at room temperature. 3,3′-diaminobenzidine tetrahydrochloride solution (DAB Solution; Solarbio, Beijing, China) was used for chemical staining, while the re-dyeing of sections was done with Hematoxylin. After the use of an alcohol gradient for dehydration and xylene, neutral resin was used to seal the slices.

Immunohistochemical scoring criteria [13] that was used incorporated both staining intensity with the staining gradation as (0 = none, 1 = weak, 2 = moderate, 3 = strong) and positive tumor cells represented in terms of % as (0 = none, 1 = 1–25%, 2 = 26–50%, 3 = 51–75%, 4 = 76–100%). The total score used in the analysis was calculated by multiplying the staining intensity and positive cell score, and according to the total score, division was done into a high-expression group (≥4) and a low-expression group (<4). All sections were evaluated by two senior pathologists.

Shanghai Cell Bank of the Chinese Academy of Sciences was the source of the human GC cell lines sgc7901, bgc823 and mgc803. Dulbecco’s modified Eagle’s medium (DMEM, HyClone, U.S.A.) plus 10% fetal bovine serum (FBS, Gibco, U.S.A.), 50 U/ml penicillin, and 50 μg/ml streptomycin was utilized for culture of cells in a humidified incubator containing 95% air and 5% CO₂ at 37°C. The cells were transduced with pLKO.1 puro lentiviral particles (GeneChem, Shanghai, China) while DOT1L was knocked down by shDOT1L (shRNA; same brand): 5′-CCGGGCCCGCAAGAAGAAGCTAAACCT CGAGGTTTAGCTTCTTCTTGCGGGCTTTTTG-3′. Lentiviral particles of shDOT1L or shCON (control) were utilized to transfect SGC7901 and MGC803 with puromycin utilized for selection.

EPZ-5676 (Selleckchem; U.S.A.) (Pinometostat; (2R,3R,4S,5R)-2-(6-amino-9H-purin-9-yl)-5-((((1r,3S)-3-(2-(5-(tert-butyl)-1H-benzo[d]im-idazol-2-yl)ethyl)cyclobutyl)(isopropyl)amino)methyl)tetra hydrofuran-3,4-diol), is a potent and selective inhibitor of DOT1L with a Ki of ≤80 pM and >37000-fold selectivity over a panel of other histone methyltransferases (HMTs). In this experiment, GC cells were cultured with different concentrations of EPZ5676 to observe the changes in cell proliferation. In addition, the effect of EPZ5676 on the growth inhibition of GC cells in vivo was observed after administering the drug in nude mice.

A 1:100 of RIPA lysis buffer (Sigma, U.S.A.) with proteinase inhibitor was utilized for extracting the total protein whose concentration was quantified using the BCA kit (Thermo Scientific, U.S.A.). Protein extracts (50 μg) were electrophoresed on 8% polyacrylamide gel and then transferred to a PVDF membrane. This was followed by 120 min of soaking in 5% skim milk at room temperature, followed by an overnight treatment with 1:1000 of rabbit anti-human monoclonal antibodies against DOT1L, H3K79me2, CDK4 and CDK6 (Abcam; U.S.A.) at 4°C along with 1:2000 of mouse anti-human monoclonal antibody H3 and β-actin (Abbkine, U.S.A.). The next step was incubation of 1:5000 of the corresponding goat anti-rabbit IgG and goat anti-mouse IgG (Abbkine, U.S.A.) at room temperature for 2 h. Finally, the images were developed using the ECL chemiluminescence kit and analyzed using the Ge Gene gel imaging system (Syngene, Cambridge, U.K.).

GC cell total RNA was isolated for Trizol™ Plus RNA Purfication kit (Invitrogen, Thermo Scientific, U.S.A.), followed by reverse transcription to cDNA with Revert Aid First Strand cDNA short kit (Thermo Scientific, U.S.A.). The primers were supplied by Shanghai GeneChem for PCR and their sequences were as follows: DOT1L-F 5′‐CATCACTATGGCGTCGAGAAA‐3′; DOT1L-R 5′‐CGCCTCTCTCCAATGTGTATT‐3′; β-actin-F 5′‐CAGGAAGGAAGGCTGGAAG‐3′; β-actin-R 5′‐CGGGAAATCGTGCGTGAC‐3′. SYBR Premix Ex Taq II (Takara Biotechnology Co., Dalian, China) was employed to perform triplicate quantitative reverse transcription polymerase chain reaction (qRT-PCR) using Fast Real-time PCR 7500 system (Applied Biosystems, Foster City, CA, U.S.A.) in accordance with the prescribed protocols of the manufacturer. The PCR conditions were as follows: (a) denaturation at 95°C for 15 s, (b) annealing at 52.4°C for 30 s, (c) extension at 72°C for 34 s, for a total of 40 cycles. The internal reference was β-actin. The 2^−ΔΔC_t approach was employed to assess gene levels.

Each group of cells in log growth phase was digested with trypsin, and 1000 cells/well were spread on a six-well plate. After 10 days of culturing in the incubator, methanol was utilized to fix the cells for 20 min, stained with Crystal Violet (Sigma, U.S.A.) for 5 min followed by washing. Microscopy was employed for enumeration of cell colonies that were >50 per well.

This was checked by cell counting kit-8 (CCK-8, Abbkine, U.S.A.); 96-well plates were used to seed 1 × 10³ GC cells per 100 μl and culture conditions as previously mentioned for 24–96 h. After the fresh medium was changed, CCK-8 reagent was added followed by 60 min of incubation, and spectrophotometry to assess the OD₄₅₀ value.

Each cell was fixed in ice-cold 70% ethanol in 4°C o/n followed by suspension in PBS with 0.1 mg/ml RNase A and 5 mg/ml propidium iodide (Sigma, U.S.A.). According to the cell DNA content, flow cytometry (BD Accuri™ C6-plus, U.S.A.) was utilized for analyzing the cell cycle. The percentages of G₁, S, G₂/M phase cells were ascertained by ModFit LT 5.0.

Twelve 4-week-old NOD-SCID male mice were purchased from the Cavens Laboratory Animal Center (Changzhou, China) and housed in the SPF Animal Center in Bengbu Medical College. Random assignment of the animals was done into one of two groups: a control group and EPZ5676 group. Resuspension of log-phase SGC7901 cells in PBS was done at 1 × 10⁷/ml, of which 0.2 ml was injected into the underarm of each mouse. EPZ5676 at 5 mg/kg was administered intraperitoneally on a weekly basis in the test group for 4 weeks. The control group was injected with an equal volume of solvent. After anesthetizing the mice with 3% pentobarbital sodium, tumors were isolated for studies. The experiments were in accordance with the Ethics Committee of the Institute.

The application of Chi-square test was done to check the association of DOT1L expression with that of GC clinicopathological factors. Overall survival (OS) curves were analyzed by Kaplan–Meier, and then the bilateral log-rank test was performed. Univariate and multivariate Cox proportional hazard models were made for assessing the prognostic significance.

A nomogram was built based on multivariable Cox regression results, using the R version of 3.3.01 RMS package. Using the Akaike information criterion, backward stepdown decline was applied to select the final model. The accuracy of the model was assessed using concordance index (C-index). By drawing the corresponding nomogram to predict survival probability, the 5-year OS was assessed after calculation using the nomogram of the calibration curve.

For basic experiments, the Student’s t test was utilized to compare the mean ± standard deviation (SD) data with Chi-square test to assess categorical data, while analysis of variance (ANOVA) was applied for three or more groups.

IHC of 101 GC cases and its surrounding tissues (Figure 1A) showed high DOT1L expression in 62 (61.4%) GC samples and in 16 (15.8%) cases of paracancer tissues (malignant vs borderline, P<0.0001) (Figure 1B).

After 5 years of follow-up, 80 (79.2%) of the 101 patients had died, with a median OS of 28 months (95% CI: 26.4–29.6 months) and an average survival time of 34.3 months (range: 11–60 months).

To clarify the prognostic function of DOT1L in GC pathology, we analyzed the DOT1L levels in 101 GC patients. DOT1L expression correlated with differentiation degree (P=0.04), lymph node metastasis (P<0.01) and TNM stage (P<0.01), but had no significant relationship with age (P=0.727), gender (P=0.188), tumor size (P=0.165), tumor location (P=0.452), depth of invasion (P=0.065), invading vessel (P=0.924) and invading nerve (P=0.748) (Table 1).

Kaplan–Meier plots show that the high level of DOT1L in GC patients was associated with lower OS against those with lesser DOT1L expression (Figure 1C). In addition, Cox regression was followed to assess the prognosis of OS. Univariate analysis showed the following prognostic factors: DOT1L expression (P<0.001), invasion depth (P=0.005), lymph node metastasis (P<0.001), TNM stage (P<0.001) and the degree of differentiation (P=0.01). Meanwhile, the result of multivariate analysis points out at high DOT1L expression (P=0.014), lymph node metastasis (P=0.023) and TNM stage (P=0.030) as independent prognostic factors of OS in GC (Table 2). In summary, the involvement of DOT1L to function as a patient prognostic marker with proto-oncogenes is strongly indicated.

A nomogram model was established using regression analysis (Figure 1D). The nomogram C-index was 0.8 and suggested that the TNM stage made the largest contribution to prognosis, followed by DOT1L expression and lymph node metastasis. Assignment of score on the point scale was done for these independent prognostic variables. The scores of all indicators were added to obtain the total score, and the total score corresponded to the predicted value of risk. Here, a calibration plot of 5-year survival shows an ideal consistency between prediction and observation of nomogram (Figure 1E).

According to the above studies, the high expression of DOT1L is closely associated with GC malignancy. Therefore, we used a loss-of-function method to study the function of DOT1L. The DOT1L expression in different GC cell lines was evaluated by Western blot (Figure 2A). Compared with other cell lines, SGC7901 and MGC803 had higher DOT1L expression. Thus, we selected these two cell lines, for knocking down DOT1L by shRNA lentiviral particles and EPZ5676. Lowered expression was observed by qRT-PCR and Western blot (Figure 2B,C). The clone formation assay was used to observe the effect of DOT1L on GC proliferation. The results showed that in comparison with controls, the clone formation ability of transfected SGC7901 and MGC803 was significantly reduced. Consistent with knockdown of DOT1L expression using shRNA, the GC proliferation also reduced because of inhibition of DOT1L enzymatic activity pharmacologically with EPZ5676 treatment (Figure 2D). Intriguingly, through the CCK-8 experiment, it was found that the inhibition of proliferation of SGC7901 and MGC803 gradually increased with time and the concentration of EPZ5676 (Figure 2E). To clarify the cause of this slowdown in GC cell proliferation caused by DOT1L inhibition, we evaluated cell cycle by flow cytometry. Compared with the empty vector control group, the number of transfected SGC7901 and MGC803 G₁ phase cells increased, and the S and G₂/M phase cells decreased (Figure 2F). Similarly, we found the same results with EPZ5676 drug treatment of SGC7901 and MGC803. These results indicated that inhibition of DOT1L in GC may cause arrest at G₁ and block cell proliferation in vitro.

According to inhibition of DOT1L, GC cell proliferation was slowed down and G₁ arrest was induced. The role of H3K79 is suggested on account of DOT1L functioning as a specific methyltransferase of H3K79, suggestive of its regulation of the cell cycle by CDKs [14,18]. CDK4/CDK6 complex can promote cell cycle, from G₁ to S phase [15]. This led us to estimate its levels by Western blot (Figure 3A). Transfected SGC7901 and MGC803 cells were significantly reduced in H3K79me2 compared with the empty vector control group, CDK4 and CDK6 were significantly decreased (Figure 3B,C). These results suggest that DOT1L modulates CDK4 and CDK6 by altering H3K79me2, leading to G₁ phase stagnation of GC cells.

To further investigate whether the DOT1L gene can also inhibit GC growth in vivo, we injected SGC7901 cells into the armpit of nude mice to construct an ectopic tumor formation model. After 28 days, it was found that mice treated with EPZ5676 grew more slowly than mice that did not receive any treatment (Figure 4A). The isolated tumors were analyzed, and it was evident that the size and weight of the untreated group were larger than those of the treatment group (Figure 4B,C). Furthermore, immunohistochemical images showed that mice treated with EPZ5676 had reduced expression of H3K79me2, CDK4 and CDK6 in mouse tumors compared with the control group (Figure 4D). These results indicate that EPZ5676 could inhibit tumor growth in GC.

We found that the expression of DOT1L is related to the development and progression of GC. In humans, DOT1L by MLL fusion partner proteins such as ENL, AF4, AF9 or AF10 binding, initiation and maintenance of participation rearrangement leukemia resulting in abnormal H3K79 methylation [16–17]. In malignant tumors of solid organs, Oktyabri et al. [10] found that increasing the expression of DOT1L may increase the expression of BCAT1, leading to changes in the sphere formation and cell migration activity of breast cancer cells. In kidney cancer, Yang et al. [18] suggested a higher level of DOT1L in patients with renal clear cell carcinoma with worse survival prognosis post-surgery than those with a low DOT1L expression. It can be seen that DOT1L is involved crucially to control gene transcription.

To confirm the expression of DOT1L in GC, we collected tissue wax blocks for immunohistochemical staining and found that DOT1L levels were conspicuously elevated in GC compared with the neighboring healthy tissues. In addition, these increased levels were connected with poor differentiation, low TNM stage and lymphatic metastasis. Additionally, the role of DOT1L to serve as an independent factor for the prognosis for OS in the case of GC and its potential role to medically address the malignancy is outlined in this work. Therefore, a nomogram was established to predict the OS over 5 years in GC patients.

For in vitro study, we blocked the expression of DOT1L in GC cells by transfection, and found that the proliferation ability of the cells was significantly decreased. It was previously reported that DOT1L inhibition in ovarian cancer and lung cancer may result in cell cycle G₁ inhibition and the association of DOT1L with DNA replication [12,19]. By examining the cell cycle, we found that the G₁ phase of GC cells was stagnant, thus causing cell proliferation to slow down. By determining the phenotype through gene expression, we found that down-regulation of H3K79 dimethylation due to DOT1L inhibition resulted in a decrease in CDK4 and CDK6 levels, leading to cell cycle inhibition and slowing of cell proliferation. This provides a new mechanism for the involvement of DOT1L in GC pathogenesis.

Dynamic changes in gene transcription lead to changes in cell morphology, and targeted therapy of small molecules greatly expands the treatment options for cancer patients. Pinometostat (EPZ5676) has been established to inhibit DOT1L enzyme activity, to function as a strategy for the treatment of MLL rearrangement leukemia [20]. Early clinical Phase I trials in adults have shown that EPZ5676 has stable metabolism and anticancer effects. EPZ5676 shows synergy with existing chemotherapeutic drugs [21]. We found in vitro that EPZ5676 indeed reduces the activity of DOT1L enzyme in GC cells, and their decreased proliferation activity. In addition, in vivo injection of EPZ5676 in nude mice with ectopic tumors was found to significantly inhibit tumor growth. Therefore, EPZ5676 has a good prospect in the progress of GC treatment.

In conclusion, this work suggests the applicability of DOT1L as a predictor of prognosis in GC, and its role in regulating CDK4 and CDK6 via H3K79me2 to subsequently, affect the cell cycle. We also showed that EPZ5676 can reduce the activity of GC cells by reducing the activity of DOT1L enzyme to exert a therapeutic effect. DOT1L can hence be looked at as an approach to treat GC.

We thank Anhui Key Laboratory of Tissue Transplantation of Bengbu Medical College for its help in experiments.

The authors declare that there are no competing interests associated with the manuscript.

This work was supported by the Natural Science Foundation of Anhui, China [grant number KJ2018A1013].

Conceptualization: Jun Qian and Zaozhi Song. Methodology: Zhuoli Wei and Zaozhi Song. Investigation: Nan Wu and Xue Liu. Formal analysis: Xiaoying Tao and Zaozhi Song. Data curation: Qingkang Wang and Xinxin Zhang. Writing – review and editing, Zaozhi Song.

The present study was granted ethical approval by the Institutional Review Board of Bengbu Medical College. Written informed consent was obtained from all participants involved in the study.

CCK-8

cell counting kit-8

CDK

cyclin-dependent kinase

C-index

concordance index

DOT1L

disruptor of telomeric silencing-1-like

GC

gastric cancer

MLL

mixed lineage leukemia

OS

overall survival

qRT-PCR

quantitative reverse transcription polymerase chain reaction