Sirt6 promotes tumorigenesis and drug resistance of diffuse large B-cell lymphoma by mediating PI3K/Akt signaling

The Gene Expression Omnibus was used to source microarray datasets GSE32918 and GSE83632. The R-package illumineHumanWGDASLv3.db data probe was used to annotate the GSE32918 data set before its conversion into gene symbols. The classical Bayesian method provided by the Limma package was used to carry out differential expression analysis of GSE83632 gene expression profiles before extraction of Sirt6 gene expression values was carried out.

A total of 70 paraffin-embedded archived tissues that were previously extracted from DLBCL patients (37 females and 33 males; range of ages between 13 and 83 years, median 58 years) between 2011 to 2018 along with 35 control specimens of reactive hyperplasia lymphoid (RHL) were used for these experiments. Established clinical criterion was used to verify the diagnosis of DLBCL [22]. Healthy donors were recruited to donate peripheral blood samples, which were then subjected to the Ficoll-Hypaque density gradient centrifugation method in order to extract peripheral blood mononuclear cells (PBMCs). Primary DLBCL cells were extracted from DLBCL patients (2 females, 43 and 66 years, and 1 male, 56 years). Each human sample was obtained with informed consent in strict compliance to the Declaration of Helsinki. All study protocols were reviewed and approved by the Medical Ethical Committee of Shandong Provincial Hospital affiliated to Shandong University (SPHASU).

Human primary DLBCL cells and human DLBCL cell lines LY1, LY8, Val, and LY3 were maintained in IMDM medium (Gibco, CA, USA) that contained 2 mM L-glutamine, 1% penicillin/streptomycin mixture and 10% heatinactivated fetal bovine serum (HyClone, UT, USA). These cells were left to incubate at 37 °C at a humidified 5% CO₂ atmosphere. All human cell lines were examined for mycoplasma infection periodically and authenticated using small tandem repeat profiling conducted by LGC standards and ATCC. Nicotinamide (S1899), OSS_128167 (S8627), Doxorubicin (Adriamycin, S1208) and Bendamustine (S1212) were procured from Selleck Chemicals (TX, USA). Recombinant human IGF-1 (100–11) was purchased from PeproTech (NJ, USA).

Lentivirus vectors that encoded either shSirt6, lvSirt6 or an empty lentiviral vector (control) were constructed by GeneChem (Shanghai, China). The following RNAi sequences were used: shSirt6 1#, CGGGAACATGTTTGTGGAA; shSirt6 2#, CCGGATCAACGGCTCTATC; shSirt6 3#, CCCTGGTCTCCAGCTTAAA. DLBCL cells were infected with lentivirus vectors according to the manufacturer’s instruction at a multiplicity of infection (MOI) = 100. Stable cells were filtered at 72 h post-transfection using 5 μg/ml puromycin. The efficiencies of each transfection reaction were evaluated with a fluorescence microscope using green fluorescent protein (GFP) and further validated using western blot and qRT-PCR assays.

Extraction of total cell RNA was carried out with the help of RNAiso Plus (TaKaRa, Dalian, China). 1 ~ 2 μg of total RNA per sample (3 stable shSirt6 transfected and 3 shControl transfected LY1 cell sample) was utilized for constructing a KAPA Stranded RNA-Seq Library Prep Kit (Illumina) sequencing library. Afterwards, the constructed library was cross-checked using an Agilent 2100 Bioanalyzer and quantified by qRT-PCR. Lastly, the Illumina HiSeq 4000 (service provided by Kangchen Biotech, Shanghai, China) was used to further sequence the libraries that had different mixed samples.

IHC and H&E staining were operated according to the manufacturers’ instructions as previously illustrated [8, 23]. Detailed information was shown in Supplementary methods. Primary antibody stained human samples was anti-Sirt6 (1:160, HPA071776, Sigma, MO, USA) or anti-Ki-67 (1:200, ab15580, Abcam, Cambridge, UK).

RNAiso Plus (TaKaRa) was used to extract total RNA. Reverse transcription reactions were carried out using reverse transcription reagents (TaKaRa). Amplification reactions were conducted using the SYBR Green Master Mix (TaKaRa) in Light Cycler 480II (Roche, Basel, Swizerland). Sirt6-specific primers used were as follows: forward, 5′-TGTGCCAAGTGTAAGACGCAG; reverse, 5′-TTGCCTTAGCCACGGTGCAG.

Western blot analysis was carried out as previously interpreted [23, 24]. Primary antibodies in this experiment comprised of: anti-Sirt6 (Sigma), anti-phosphor-PI3 Kinase p110α, anti-phospho-AKT(Ser473), anti-total pan-AKT, anti phospho-mTOR, and anti-total pan-mTOR, anti-PTEN, anti-phospho-4EBP1, anti-FoxO1, anti-HIF-1α, anti-PARP [specific to the full-length (116 kDa) and the cleaved form (89 kDa) of PARP], anti-p27, anti-CDK2, anti-phospho-ATM, anti-phospho-ATR, anti-phospho-Chk1 (Ser345), anti-phospho-Chk2 (Thr68) and anti-β-tubulin (Cell Signaling Technology, MA, USA), β-actin (Zhongshan Goldenbridge, Beijing, China). Details are shown in Supplementary methods.

DLBCL cell viability was evaluated using the Cell Counting Kit-8 (CCK-8; Dojindo, Japan). DLBCL cells with designed treatment were disseminated onto 96-well plates for 24–72 h. Following this, the cells were left to incubate with 10 μl of CCK-8 per well at 37 °C for 4 h. The absorbance at 450 nm was measured using Multiskan GO Microplate Reader (Thermo Scientific, USA).

DLBCL cell apoptosis and cell cycle analysis with the designed treatments were detected by flow cytometry as previous described [23, 24]. Details are shown in Supplementary methods. All assays were carried out using the Navios Flow Cytometer (Beckman Coulter, CA, USA).

Principles of Animal Care and Use Ethics Committee of SPHASU and ARRIE guidelines were strictly adhered to upon conduction of all animal experiments. 6-week-old beige female mice with severe combined immunodeficiency (SCID) were obtained from the Weitong Lihua Laboratory Animal Center (Beijing, China) and reared in a pathogen-free environment. All mice were randomly cohorted into two groups. As previously illustrated [8, 23], DLBCL cells (either transfected with shSirt6 vectors or empty control vectors) were subcutaneously injected to SCID-beige mice to establish xenograft models (n = 8 per group). Details are described in Supplementary methods. Experiments using OSS_128167 involved first subcutaneously injecting the SCID beige mice with 1 × 10⁷ LY1 cells. Eight days after the first injection, mice were administered with intraperitoneal injections of either OSS_128167 (80 mg/kg, n = 4) or a control vehicle (n = 4) every 2 days for 2 weeks. Pathological analysis was then carried out with dissected tumor tissues.

To evaluate the potential role of Sirt6 in DLBCL, we first examined Sirt6 expression in GEO database. As shown in Fig. 1a, Sirt6 was markedly upregulated in DLBCL in contrast to normal samples in a bioinformatic analysis on GSE83632 (n = 163). Kaplan-Meier survival curve analysis of the GSE32918 indicated higher Sirt6 expression in DLBCL (n = 249) was correlated with shorter overall survival times (p = 0.0023, Fig. 1b).

There was noted to be higher Sirt6 expression levels in DLBCL tissues compared to tissues with RHL upon IHC staining (Fig. 1c), with a Sirt6 positive rate of 80% (56 of 70) in DLBCL tissues compared to 8.6% (3 of 35) in tissues with RHL. To clarify the clinical significance of Sirt6 expression in DLBCL, the clinicopathological characteristics of DLBCL patients were analyzed. Sirt6 expression levels were positively related to age (p = 0.009), Ann Arbor stage (p = 0.036), and international prognostic index (IPI, p = 0.045) score of DLBCL patients (Table 1). The survival analysis indicated higher Sirt6 expression in DLBCL was correlated with shorter overall survival times (p = 0.039, Fig. 1d). Both the mRNA and protein levels of Sirt6 were aberrantly higher in DLBCL cell lines than those in the control cells from healthy volunteers (Fig. 1e-f). These results illuminated the prognostic significance of Sirt6 in DLBCL.

The function of Sirt6 was further investigated using lentivirus mediated gain- and loss-of-function assays. Three lentivirus mediated RNA interference (RNAi) vectors carrying GFP against Sirt6 demonstrated effective knockdown of Sirt6 in human LY1, LY8, and Val cells, of which shSirt6#1 exhibited the highest efficacy. Effective knockdown (shSirt6, Fig. 2a-b) or overexpression (lvSirt6, Supplemental Fig. 1a) was verified using western blotting or qRT-PCR experiments. Stable shSirt6 transfected DLBCL cells exhibited growth suppression in contrast to cells with empty vectors (Fig. 2c). However, overexpression of Sirt6 in DLBCL cells had no impact on the proliferative ability of cells, an effect likely being the result of constitutively high Sirt6 levels (Supplemental Fig. 1b).

Furthermore, we established a mouse xenograft model using human DLBCL cells to investigate the tumor-promoting effect of Sirt6 in vivo. SCID beige mice received subcutaneous injections of either shSirt6 or shControl transfected LY1 cells (n = 8 per group). Mice bearing shSirt6 were found to have markedly smaller tumor volumes in comparison to the control group, which was consistent with the result obtained in our in vitro experiments (p < 0.01, Fig. 2d). Lower Sirt6 expression levels were confirmed in the xenograft tumor tissues derived from shSirt6 cells by IHC staining. Besides, we also observed lower expression level of proliferative marker Ki-67 [25, 26] in shSirt6 group (Fig. 2e), indicating the positive regulation of Sirt6 on DLBCL cell proliferation.

The pathological effects of Sirt6 in DLBCL were further explored using RNA-seq analysis on stable shSirt6 transfected or shControl transfected LY1 cells. Differentially expressed genes and transcripts were determined for further analysis. As depicted in annotations of gene ontology (GO) analysis in Fig. 3a-c, Sirt6 appeared to be strongly correlated to processes associated with cell cycle, DNA damage, and cell apoptosis.

We next verified the above Sirt6-related biological processes in DLBCL cell lines. Annexin V-PE/ 7AAD apoptotic assays indicated that shSirt6 caused higher rates of cellular apoptosis in both LY1 (8.63 ± 0.44% in Control vs.14.23 ± 0.35% in shSirt6 group, p = 0.0006), LY8 (10.80 ± 1.00% in Control vs. 19.03 ± 1.85% in shSirt6 group, p = 0.0014), and Val cells (10.73 ± 0.76% in Control vs.17.43 ± 0.69% in shSirt6 group, p = 0.0029, Fig. 3d). Conversely, Sirt6 overexpression resulted in reduced apoptosis (LY1: 8.63 ± 0.44% in Control vs. 4.40 ± 0.15% in lvSirt6 group, p = 0.0008; LY8: 10.80 ± 1.00% in Control vs. 2.77 ± 0.32% in lvSirt6 group, p = 0.0016, Supplemental Fig. 1c). The anti-apoptotic effect of Sirt6 was also validated by western blot analysis, as indicated by higher levels of cleaved PARP in shSirt6 cells (Fig. 3f). We then monitored cell cycle by PI staining. Knockdown of Sirt6 induced an obvious increase of DLBCL cell counts in the G2/M phase compared to cells transfected with shControl (Fig. 3e). Besides, expression of p27 and CDK2 were modestly changed in shSirt6 cells (Fig. 3f). Altogether, these results suggested that Sirt6 could promote DLBCL cell survival via acceleration of the cell cycle out of the G2/M phase and reduced rates of apoptosis.

To further confirm the effect of Sirt6 inhibition on DLBCL, the function of OSS_128167, a novel Sirt6-specific inhibitor which can permeate cell membrane and exhibit biologically active in cultured cells, was investigated. OSS_128167 increases H3K9 acetylation and GLUT-1 expression [27]. Studies on OSS_128167 is still in the early stages given its relatively new introduction to the market. Michele Cea et al. suggested that OSS_128167 was able to sensitize primary MM cells, as well as doxorubicin- and melphalan- resistant MM cell lines, towards chemotherapy [18]. Primary DLBLC cells and DLBCL cell lines (LY1, LY8) were exposed to various concentrations of OSS_128167 for 24–72 h, respectively. OSS_128167 decreased cell viability in primary DLBCL cells (Fig. 4a). Cellular proliferation was suppressed in a time- and dose-dependent manner with a reduction by 60% noted at 48h post-exposure to a concentration of 100 μM in DLBCL cell lines (Fig. 4b).

Moreover, flow cytometric analysis revealed that the level of primary DLBCL cells and DLBCL cell lines (LY1, LY8) undergoing early apoptosis were markedly raised after incubation with 80 μM OSS_128167 for 48 h (Fig. 4b). Gradually increasing concentrations of OSS_128167 treatment also impacted the cell cycle progression of DLBCL cells. G2/M phase cell counts elevation was consistent with prior studies (Fig. 4c).

The anti-tumor impact of OSS_128167 on DLBCL cells was further investigated in vivo using a xenograft model of SCID beige mice. Consistent with our in vitro results, obviously decreased tumor growth was noted in the mice treated with OSS_128167 in contrast to the vehicle control group (Fig. 4d). Additionally, lower expression level of the proliferative marker Ki-67 was observed in OSS_128167 exposed mice (Fig. 4e). This experiment for the first time performed intraperitoneal inoculation of OSS_128167 onto mice.

Doxorubicin, a DNA damaging agent, is a common agent frequently incorporated into chemotherapeutic regimens against DLBCL. Bendamustine is another DNA damaging drug used for chronic lymphocytic leukemia (CLL), MM, and more recently for non-hodgkin’s lymphoma (NHL) [28, 29]. Nevertheless, there is still a large proportion of patients that remain refractory to all these agents, representing a gap in current chemotherapeutic efficacies. Previous research has shown that Sirt6 played important roles in response to DNA damage agents and in drug resistance [18, 30‐32].

Therefore, we investigated the effects of Sirt6 in drug responses to doxorubicin and bendamustine by cell cytotoxicity assays. shSirt6 cells and control cells were exposed for 48 h to pre-determined concentrations of doxorubicin or bendamusitine. Notably, the cell proliferation of shSirt6 DLBCL cell lines was markedly reduced (p < 0.05, Fig. 5a-b).

DNA damage is closely related to the occurrence, progression and treatment of cancer. Cellular responses to DNA damage encompass complex mechanisms and checkpoint responses that can maintain genome stability and delay cell cycle progression. Master regulators of the DNA damage response (DDR) program are two proteins, the ataxia-telangiectasia mutated (ATM), and ataxia-telangiectasia and RAD3-related protein (ATR), which can phophorylate protein substrates [33]. The most important substrates among them are serine-threonine protein kinases- Chk1 and Chk2 [34]. In vertebrates, the ATM-Chk2 and ATR-Chk1 signaling cascades coordinated the DDR responses and cell cycle checkpoints, the latter pathway is the principal direct effector [35]. Considerable evidence has accumulated to support that, defects in DNA damage responses, especially inhibit Chk1, can result in sensitization to genotoxic stress [36‐38]. Our western blotting experiments showed that, cells transfected with shSirt6 were noted to have reduced expressions of phosphorylated ATR and phosphorylated Chk1 (Fig. 5c). These findings might shine a light on the mechanism of Sirt6 knockdown or inhibition, a process that can sensitize DLBCL cells to chemotherapeutic drugs. By the way, these consequences are echoed in our RNA-seq analysis results.

Next, we futher explored the potential mechanisms involved in the regulation of Sirt6, using gene set enrichment analysis (GSEA) according to RNA-seq data. Sirt6 appeared to be functionally enriched in the PI3K/Akt and FOXO signaling pathways (Fig. 6a). The conserved forkhead box class O (FoxO) transcription factor family is a prominent mediator of the PI3K signaling cascade, Akt is a key regulator of FoxO pathway [39, 40]. The FoxO proteins regulate diverse cellular and physiological processes such as apoptosis, cell proliferation, cell cycle, and DNA repair pathways [41, 42]. Therefore, FoxO1 functions as tumor suppressor in various solid tumors and B cell lymphomas, including DLBCL [43‐45]. The PI3K/Akt/mammalian target of rapamycin (mTOR) pathway is a well-established pathway in carcinogenesis, including DLBCL [46, 47]. Hence, we focused on the protein levels of this signaling pathway in DLBCL cells with Sirt6 knocked down. As shown in Fig. 6b, the phosphorylation levels of PI3K p110α and its downstream targets, Akt (Ser473) and mTOR proteins, were noted to be dramatically decreased in shSirt6 cells compared to control cells. Conversely, the total Akt and mTOR protein levels were rarely significantly different between the two groups. Moreover, increased expression levels of PTEN, a negative regulator of PI3K [48], phosphorylated 4EBP1, and FoxO1 protein were found in shSirt6 cells.

As mentioned above, we established a xenograft mouse model with human DLBCL cells. LY1 cells transfected with shSirt6 or shControl cells were injected subcutaneously into the SCID beige mice. Western blotting confirmed the reduced Sirt6 expression in the xenograft tumor sections derived from shSirt6 cells. Mice bearing shSirt6 cells were noted to have lower levels of phosphorylated PI3K and its downstream targets compared to those in the shControl group (Fig. 6c).

Insulin-like growth factor-1 (IGF-1), a known activator of PI3K signaling [49], was used to further investigate the ability of Sirt6 to modify PI3K signaling in DLBCL cells. DLBCL cells transfected with either shSirt6 or shControl were exposed to IGF-1 or vehicle control in 0.5% FBS culture medium for 24–96 h. shSirt6 cells that did not receive IGF-1 treatment demonstrated lower LY8 and LY1 cellular proliferation in contrast to cells transfected with empty vectors. Cells that received IGF-1 treatment were noted to have a partial restoration in their ability to proliferate. This effect was less in shSirt6 cells in contrast to control cells (Fig. 6d).

LY1 cells were first infected with either shSirt6 or shControl, starved of serum for 48 h and subsequently exposed for 30 min to either IGF-1 50 ng/ml or vehicle control [23]. Cells that were transfected with shSirt6 were noted to have decreased levels of phosphorylated PI3K and its downstream targets, Akt and mTOR (Fig. 6e). To sum up, these findings suggest that Sirt6 may regulate the progression of DLBCL via activating the PI3K/Akt/mTOR and DNA damage response signaling pathway (Fig. 6f).