STAT3 activation in large granular lymphocyte leukemia is associated with cytokine signaling and DNA hypermethylation

Large granular lymphocyte leukemia (LGLL) is characterized by somatic gain-of-function STAT3 mutations. However, the functional effects of STAT3 mutations on primary LGLL cells have not been studied in detail. In this study, we show that CD8+ T cells isolated from STAT3 mutated LGLL patients have high protein levels of epigenetic regulators, such as DNMT1, and are characterized by global hypermethylation. Correspondingly, treatment of healthy CD8+ T cells with IL-6, IL-15, and/or MCP-1 cytokines resulted in STAT3 activation, increased DNMT1, EZH2, c-MYC, l-MYC, MAX, and NFκB levels, increased DNA methylation, and increased oxidative stress. Similar results were discovered in KAI3 NK cells overexpressing gain-of-function STAT3Y640F and STAT3G618R mutants compared to KAI3 NK cells overexpressing STAT3WT. Our results also confirm that STAT3 forms a direct complex with DNMT1, EZH2, and HDAC1. In STAT3 mutated LGLL cells, DNA methyltransferase (DNMT) inhibitor azacitidine abrogated the activation of STAT3 via restored SHP1 expression. In conclusion, STAT3 mutations cause DNA hypermethylation resulting in sensitivity to DNMT inhibitors, which could be considered as a novel treatment option for LGLL patients with resistance to standard treatments.


Introduction
Large granular lymphocyte leukemia (LGLL) is a rare chronic lymphoproliferative disorder, which in over 80% of the cases results from abnormal clonal expansion of CD3 + CD8 + cytotoxic T cells [1]. Less frequently, LGLL originates from the expansion of natural killer (NK) cells (chronic lymphoproliferative disorder of NK cells, CLPD-NK). Both T-cell LGLL and CLPD-NK are chronic and indolent diseases, but instances of aggressive leukemias have also been reported [2].
LGLL patients with STAT3 mutations have more often neutropenia and rheumatoid arthritis (RA) [3,9,10], and mutations may also confer reduced overall survival [5]. Although STAT3 activation, irrespective of mutation status, is one of the key features and a known mediator of cytokine signaling and inflammation in LGLL [1,[11][12][13], the functional consequences of STAT3 activation in LGLL remain to be fully elucidated.
Cytokines induce differentiation and proliferation of lymphocytes, and aberrant cytokine signaling can be observed in many inflammatory conditions [14]. Interleukin-6 (IL-6) is elevated in RA leading to JAK/STAT3 signaling activation, and IL-6 and IL-6 receptor inhibitors are used in RA treatment [15]. In addition to the well-established role in chronic inflammatory diseases, cytokine dysregulation may also drive tumor initiation and progression [16]. In LGLL mouse model, excessive IL-15 induced DNMT3b upregulation and global DNA hypermethylation, and IL-15 may drive LGLL formation [17][18][19].
As the role of STAT3 in regulating DNA methylation in LGLL is currently unknown, we investigated whether inflammatory cytokine stimulation and STAT3 activation modify the epigenetic machinery in CD8+ T cells and could thereby affect disease pathogenesis in LGLL. Using LGLL patients samples, healthy human CD8+ T cells, and KAI3 NK cell line, we show that cytokine stimulation or overexpression of GOF STAT3 variants lead to increased protein levels of epigenetic regulators, increased global DNA methylation levels, and increased reactive oxygen species (ROS) production. Treatment with DNMT inhibitor abrogated STAT3 mediated changes in SHP1 dependent manner, paving the way for novel targeted therapeutic approaches.

Materials and methods
More detailed descriptions of the methods and analyses are available in the Supplementary Methods.

Study participants
Study participants were recruited at Helsinki University Hospital (HUH), Helsinki, Finland. Healthy controls were collected from Finnish Red Cross Blood Service (Helsinki, Finland). All participants gave written informed consent before participation. The ethics committee in the HUH approved the study, and it was conducted in accordance with the Declaration of Helsinki. Patients' clinical characteristics are shown in Supplementary Table 1 as well as assays in which samples were used.

Plasma cytokine measurements
Plasma cytokine measurement was performed using a Proseek Multiplex Inflammation panel by Olink Proteomics (Uppsala, Sweden) [21]. The relative protein levels are expressed as Normalized Protein eXpression (NPX) in 2-log scale. A larger NPX level indicates a higher protein expression in the sample.

RNA sequencing
Total RNA was extracted from PBMCs of T-LGLL patients and healthy controls. RNA-sequencing using Illumina Hiseq 2000 and further bioinformatic processing were performed as previously reported [22]. siRNA transfection siRNAs were purchased from Dharmacon (Chicago, IL, USA). Cells were transfected with 50 nmol/L siRNA using Lipofectamine™ 2000 Transfection Reagent (Thermo Fisher Scientific) in 12-well plates according to manufacturer's protocols. siRNA sequences are described in Supplementary Table 2.

Drug sensitivity and resistance testing
Drug sensitivity and resistance testing were performed with primary CD8+ T cells of both T-LGLL patients harboring STAT3 Y640F mutation and healthy controls. Azacitidine (HY-10586, CAS 320-67-2) was purchased from Med-ChemExpress (Monmouth Junction, NJ, USA). The experiment was performed as previously described [23].

Data analysis and statistics
Statistical analysis was performed using GraphPad Prism (Version 8.4.3) and R 4.0.3. Comparisons between the two groups were made using unpaired two side t-test and Mann-Whitney U test. P value < 0.05 was considered as statistically significant.

Plasma inflammatory cytokine levels are increased in LGLL patients
We measured plasma cytokine levels using Olink inflammation assay panel [21], including 91 proteins in nine T-LGLL patients (STAT3 Y640F , STAT3 D661Y , STAT3 Y657ins , and without STAT3 mutations) and eight healthy controls. 34 out of 91 cytokines were higher in T-LGLL patients compared to healthy controls (Fig. 1A, Supplementary Table 3), suggesting broad cytokine upregulation. Specifically, IL-6 and IL-15RA were higher in T-LGLL patients compared to healthy controls irrespective of STAT3 mutation status (Fig. 1A). Of note, IL-15 was not included in the Olink inflammation panel used (Supplementary Table 3).
Monocyte chemoattractant protein 1 (MCP-1) levels were also upregulated in patients with STAT3 mutations, whereas for example IL-2 and IL-4 levels were similar between patients and healthy controls (Fig. 1A).
To examine whether STAT3 regulates DNMT activity and DNA methylation in T-LGLL, we quantified DNMT protein levels in patient samples. Interestingly, DNMT1 and DNMT3B levels were significantly higher in CD8+ T cells of the LGLL patients with STAT3 Y640F GOF mutation compared to healthy controls (Fig. 2B). The protein level of EZH2, a histone H3K27 methyltransferase, was also higher in T-LGLL patients, particularly in patients with STA-T3 Y640F mutation. In addition, we observed high level of MYC protein in T-LGLL patients harboring STAT3 Y640F GOF mutation ( Supplementary Fig. 2).
To further investigate whether STAT3 Y640F mutation can upregulate epigenetic regulators, we used KAI3 NK cells previously engineered to overexpress exogenous STAT3 wildtype (STAT3 WT ) and hyperactive STAT3 variant (STAT3 Y640F ) [20] found in T-LGLL patients [3][4][5]. We also tested KAI3 cells expressing STAT3 G618R variant [20] to further examine the effect of STAT3 GOF mutations. DNMT1 and DNMT3B were significantly upregulated in KAI3 NK cells overexpressing STAT3 GOF variants compared to cells overexpressing STAT3 WT (Fig. 2C, Supplementary Fig. 3A). In addition, EZH2 protein level and levels of H3K27me3, the end-product of histone methylation catalyzed by EZH2 as part of the PRC2 complex [25], were increased in cells overexpressing STAT3 GOF variants (Fig. 2C, Supplementary Fig. 3A). DNMT3A protein was not detected in KAI3 NK cells ( Supplementary Fig. 3B). Similar to patient cells, the expression of MYC family proteins, including C-MYC and L-MYC, were increased in KAI3 NK cells overexpressing STAT3 GOF mutations (Fig. 2C). Further, MYC-associated factor X (MAX) [26] was upregulated in KAI3 NK cells expressing STAT3 variants (Fig. 2C). The protein level of p65, a subunit of NF-κB reported to be upregulated in MYC-dependent manner upon IL-15 stimulation [17], was also increased in hyperactive STAT3 KAI3 NK cells (Fig. 2C). Although HDAC1 has been reported to be upregulated by excessive production of IL-15 in T cells [27], we did not find any difference in HDAC1 levels according to the STAT3 status (Fig. 2C, Supplementary Fig. 3A).
Overexpression of GOF STAT3 variants have been reported to promote cell proliferation in KAI3 NK cells under limiting IL-2 concentrations [20]. To further validate that this was due to STAT3 activation, we performed si-RNA mediated STAT3 knockdown in KAI3 NK cells (Fig. 2D). While STAT3 GOF expression led to increased cell proliferation in KAI3 NK cells, STAT3 knockdown rescued this phenotype (Fig. 2E). We additionally confirmed that STAT3 knockdown attenuated C-MYC (Supplementary Fig. 4), further supporting a link between STAT3 and MYC pathway.

Cytokines activate STAT3 and epigenetic regulator signaling in healthy CD8+ T cells
To address whether high cytokine levels are sufficient to activate STAT3 and the epigenetic machinery in CD8+ T cells, we isolated primary CD8+ T cells from healthy blood donors. These cells were cultured with IL-6, IL-15, and MCP-1, which were selected based on their higher levels in T-LGLL patients relative to healthy controls (

STAT3 directly binds epigenetic regulator proteins
To understand the connection between STAT3 and epigenetic modifiers, we performed Co-immunoprecipitation (Co-IP) in CD8+ T cells isolated from healthy controls and stimulated with IL-6, IL-15, MCP-1, or IL-15 plus MCP-1. STAT3 was able to bind DNMT1, EZH2, and HDAC1 ( Fig. 3B-E). EZH2 directly interacted with STAT3 but not within a higher-order complex including HDAC1 or , T-LGLL patients harboring STAT3 Y640F mutation (STAT3 Y640F , n = 5) and without STAT3 mutations (No mut, n = 5) were extracted. A Left, data are shown as phosphorylated STAT3 (p-STAT3) and STAT3. Right, quantitative presentation of Western blot assays using Image J (Version 2.0.0). Protein levels were first normalized to LAMINB1, a loading control. The p-STAT3 level was normalized by dividing intensity of p-STAT3 by total STAT3. B Left, DNMT1, DNMT3B, and EZH2 protein expression levels. Right, quantitative presentation of the protein levels. DNMT1, DNMT3B, and EZH2 protein levels were normalized with LAMINB1. Each dot represents the protein expression level from one individual. Error bar present mean ± SD (n = 5 per group), and statistically significant difference was evaluated using Mann-Whitney U test. *P < 0.05; ***P < 0.001; ****P < 0.0001. C KAI3 NK cells harboring STAT3 WT (WT), STAT3 G618R mutation (G618R) and STAT3 Y640F mutation (Y640F) were serum starved for 12 h followed by nucleus extraction. Protein samples were analyzed via Western blot analysis using the indicated antibodies. Data are representative of three independent experiments. Protein quantification is presented in Supplemental Fig. 3A. D KAI3 NK cells (STAT3 WT , STAT3 G618R , and STAT3 Y640F ) were transfected with STAT3-siRNA (si-STAT3) and control siRNA (si-con) for 72 h followed by serum starvation for 12 h Western blot analysis was performed with p-STAT3 and STAT3 specific antibodies. LAMINB1 served as loading control and data are representative of three independent experiments. E Cell proliferation assay was peformed using STAT3 siRNA-mediated knockdown in KAI3 NK cells. KAI3 NK cells stably expressing STAT3 WT and gain-of-function STAT3 variants (STAT3 G618R and STAT3 Y640F ) were cultured with STAT3-siRNA (si-STAT3) and control siRNA (si-con) for 48 h The cells (10,000 cells/well) were incubated in RPMI1640 with reduced IL-2 (25IU/ml) with extra si-RNAs for indicated days, and the cell proliferation was measured by DNA fluorescence-based assay. The cell numbers were calculated with a standard curve based on the fluorescence intensity. Error bar present mean ± SD of three independent experiments, and statistically significant difference was evaluated using unpaired t-test (STAT3 Y640F si-con vs STAT3 Y640F si-STAT3, STAT3 G618R si-con vs STAT3 G618R si-STAT3). *P < 0.05; **P < 0.01; ***P < 0.001. DNMT1 ( Fig. 3F-G). Consistent with findings in stimulated healthy CD8+ T cells, we observed stronger interaction between STAT3 and DNMT1 as well as between STAT3 and HDAC1 in KAI3 NK cells overexpressing GOF STAT3 compared to STAT3 WT (Supplementary Fig. 7).
We then evaluated whether STAT3 inhibition can suppress DNA methylation using KAI3 NK cells and a small interfering RNA (siRNA) targeting STAT3. Consistent with results from a prior study [28], DNMT1 levels decreased in response to STAT3 inhibition (Fig. 4D). Global methylation Fig. 3 Cytokines activate STAT3 that binds epigenetic regulators in healthy CD8+ T cells. A Western blot analysis of CD8+ T cells of healthy controls. Cells were cultured with IL-6, IL-15, MCP-1, IL-15 plus MCP-1 and without cytokine (con) in RPMI1640 medium for 12 h (100 ng/ml for each cytokine). Nuclear fractions of the cells were used to detect the indicated proteins. LAMINB1 was used as loading control. B-F Physical interaction of STAT3 with DNMT1, EZH2, and HDAC1. CD8+ T cells of healthy controls were cultured with IL-6, IL-15, MCP-1, and IL15 plus MCP-1 in RPMI1640 medium for 12 h to induce STAT3 activation. Co-Immunoprecipitation was performed with the nuclear protein lysates of CD8+ T cells. B The nuclear extracts (10%) of the cells were used as input control. Co-IP was performed with antibody against C STAT3, D DNMT1, E HDAC1, and F EZH2. Rabbit IgG Isotype control (IgG) was used as a negative control for non-specific binding for Co-IP. Immunoprecipitated complexes were probed using the antibodies as indicated. Data are representative of three independent experiments. G Simplified scheme of the complex of STAT3, DNMT1, EZH2, and HDAC1 using BioRender (Toronto, Canada). levels decreased with STAT3 siRNA in KAI3 NK cells overexpressing STAT3 mutants, but not in STAT3 WT cells (Fig. 4E), suggesting that the upregulated DNA methylation is caused by STAT3 activity and DNMT1 (Fig. 4E, F).

Increased ROS levels associated with STAT3 activation and cytokine stimulation
Since a link between DNA methylation and oxidative stress has been suggested [29], we measured ROS levels in CD8+ T cells from T-LGLL patients using an in vitro ROS/RNS (reactive nitrogen species) assay. ROS/RNS levels were significantly higher in T-LGLL patients harboring STA-T3 Y640F mutation (median = 3070) compared with both healthy controls (median = 1528) and the patients without STAT3 mutations (median = 2117) (Fig. 5A). Interestingly, KAI3 NK cells overexpressing STAT3 (either wildtype or STAT3 GOF) showed increased levels of both ROS and superoxide compared to the control KAI3 cells. Moreover, ROS levels in cells carrying STAT3 GOF mutations were higher compared to KAI3 NK cells expressing STAT3 WT (Fig. 5B).
(4.3%) had no effect on the production of superoxide. Cytokine stimulation also increased p-ATM and γH2AX in CD8+ T cells (Fig. 5D, Supplementary Fig. 8A), indicating concomitant activation of the DNA damage response pathway [30]. Ser-276 phosphorylation of NF-κB p65 was also noted in cytokine-stimulated CD8+ T cells (Fig. 5D,   Supplementary Fig. 8A), suggesting ROS-dependent NF-κB signaling activation [31]. In addition, these proteins were upregulated in CD8+ T cells of T-LGLL patients harboring STAT3 GOF mutation (Fig. 5E, Supplementary  Fig. 8B). Taken together, these results suggest that STAT3 driver mutations and higher p-STAT3 activation are LGLL patients harboring STAT3 Y640F mutation (STAT3 Y640F ) and without STAT3 mutations (No mut) as well as healthy controls (HC) (n = 3). Error bars expressed as mean ± SD, and statistically significant difference was evaluated using unpaired t-test. *P < 0.05. B Flow cytometry analysis to examine the expression of total ROS and superoxide in KAI3 NK cells harboring STAT3 WT , STAT3 G618R , STAT3 Y640F mutation and control KAI3 NK cells (Mock). Cells were cultured in RPMI1640 with reduced IL-2 (25IU/ml) for 72 h Then, the cells were labeled with ROS-fluorescent dye (Ex/Em= 650/675 nm) and superoxide-fluorescent dye (Ex/Em = 520/605 nm), and then analyzed using flow cytometer. KAI3 NK cells expressing empty vector (Mock) were used as a control. C Flow cytometry analysis to examine the expression of total ROS and superoxide in untreated CD8+ T cells (con) and those incubated with IL-6, IL-15, MCP-1, or IL15 plus MCP-1. CD8+ T cells were incubated with the indicated cytokines (100 ng/ml) for 12 h in RPMI1640. Cells were labeled with ROS-fluorescent dye and superoxidefluorescent dye, and then analyzed using flow cytometer. D p-ATM, ATM, γH2AX, H2AX, p65 and p-p65 protein expression in healthy CD8+ T cells stimulated with IL-6, IL-15, and/or MCP-1 for 12 h. Quantification is presented in Supplementary Fig. 8A. E p-ATM, ATM, γH2AX, H2AX, p65, and p-p65 protein expression in CD8+ T cells of T-LGLL patients harboring STAT3 Y640F mutation (STA-T3 Y640F ) and without STAT3 mutations (No mut) as well as healthy controls (HC). Quantification is presented in Supplementary Fig. 8B. associated with higher ROS production and activation of the DNA damage response and NF-κB pathways, potentially contributing to LGLL phenotype.

Azacitidine abrogates STAT3 activation, DNA methylation, and ROS production
To explore whether DNA hypermethylation associated with STAT3 activation could be therapeutically targeted, we evaluated the effects of azacitidine, a hypomethylating agent widely used to treat myeloid malignancies. Ex vivo drug sensitivity testing showed that T-LGLL CD8+ T cells were more sensitive to azacitidine at higher concentrations compared with CD8+ T cells from healthy controls (Fig. 6A). Next, we investigated whether azacitidine decreases the protein levels of STAT3, p-STAT3, DNMT1, and EZH2. CD8+ T cells of a T-LGLL patient harboring STAT3 Y640F mutation were cultured in the presence or absence of azacitidine (1 and 10 μM) for 24, 48, and 72 h. Azacitidine inhibited p-STAT3, total STAT3, and DNMT1 in dose-and time-dependent manner (Fig. 6B, Supplementary Fig. 9). EZH2 levels were not affected by azacitidine. To validate these results, we treated KAI3 NK cells with azacitidine. While no statistically significant differences in cell viability were observed between cells overexpressing STAT3 WT or GOF STAT3 variants (Fig. 6C), protein levels of pSTAT3, total STAT3, and DNMT1 were decreased with azacitidine treatment (Fig. 6D), consistent with results observed in CD8+ T cells. EZH2 level was also decreased after 72 h of 10 μM of azacitidine treatment in KAI3 NK cells (Fig. 6D, Supplementary Fig. 10).
In line with the hypomethylating mechanism of action of azacitidine, KAI3 NK cells showed decreased global methylation levels in time and dose-dependent manner after azacitidine treatment (Fig. 6E, Supplementary Fig. 11). ROS levels were decreased in KAI3 NK cells expressing either STAT3 WT and STAT3 GOF variants (Fig. 6F, Supplementary Fig. 12A). In addition, azacitidine reduced Ser-276 phosphorylation of p65 in CD8+ T cells of LGLL patients (Fig. 6G) and KAI3 NK cells (Fig. 6H). Given that ROS can stimulate overall tyrosine phosphorylationdependent signaling [32], we additionally assessed the effect of azacitidine on total phosphotyrosine levels in KAI3 NK cells. Azacitidine significantly reduced whole cell tyrosine-phosphorylated proteins (25-260 kDa), including p-STAT3 (Supplementary Fig. 12B). Together, these results suggest that DNMT inhibitors can influence epigenetic programming and ROS production in STAT3 activationdependent manner, with potential clinical applicability in LGLL.

Azacitidine inhibits STAT3 activation by SHP1 restoration
Previous studies have shown that SHP1 negatively regulates JAK/STAT3 signaling pathway [33,34]. Basal expression of SHP1 was lower in CD8+ T cells of T-LGLL patients harboring STAT3 Y640F mutations compared to both healthy controls and T-LGLL patients without STAT3 mutations (Fig. 7A, Supplementary Fig. 13). Furthermore, azacitidine treatment led to increased SHP1 levels in CD8+ T cells isolated from T-LGLL patients (Fig. 7B, Supplementary  Fig. 14). SHP1 levels were lower in KAI3 NK cells overexpressing GOF STAT3 mutations compared to STAT3 WT cells at baseline; however, azacitidine treatment increased SHP1 levels in GOF STAT3 cells (Fig. 7C). Next, KAI3 NK cells were transfected with control and SHP1 siRNA for 48 h followed by azacitidine treatment. Knockdown of SHP1 upregulated p-STAT3 and DNMT1 in STAT3 WT cells at baseline (Fig. 7D). In addition, SHP1 knockdown attenuated p-STAT3 and DNMT1 downregulation during azacitidine treatment in KAI3 NK cells. SHP1 knockdown promoted cell proliferation in STAT3 WT but not in GOF STAT3 cells, providing further support for the link between STAT3 upregulation and cell proliferation (Fig. 7E). We also confirmed that siRNA-mediated STAT3 knockdown restored SHP1 expression (Fig. 7F). Together, these findings suggest that SHP1 controls STAT3 activity and inhibition of STAT3 during azacitidine treatment (Fig. 7G).

Discussion
Activating STAT3 somatic mutations can be found both in patients with T-cell LGLL (40-50% of cases with CD8 + CD3 + phenotype) and in CLPD-NK (20-30% of patients) [3,6]. STAT3 is also mutated in other T and NK cell neoplasias such as mature T cell lymphomas and aggressive NK cell leukemias [35][36][37][38]. Interestingly, germline STAT3 GOF variants have also been associated with early LGL lymphoproliferation [39,40]. Furthermore, a role for JAK/ STAT activation in LGLL pathogenesis even in patients without STAT3 mutations has been suggested [1,6,12,13]. Nevertheless, the consequences of STAT3 activation in pathological T-and NK-cell proliferations remain elusive. In our study, we show that in primary LGLL cells GOF STAT3 mutations alter autoregulation of p-STAT3 and drive changes in epigenetic regulator levels, global DNA hypermethylation, and ROS production. In addition to LGLL Fig. 6 Azacitidine abrogates DNA hypermethylation and STAT3 activation in KAI3 NK cells. A Drug sensitivity testing was performed on CD8+ T cells from T-LGLL patients harboring STAT3 Y640F mutation (STAT3 Y640F , n = 3) and healthy controls (n = 3). Cell viability inhibition by correlation of drug sensitivity score (DSS) was measured with CellTiter-Glo2.0 (Promega, USA). DSS is a quantitative measurement of a drug response based on the area under the curve (AUC) with further normalization. Higher DSS denote better killing activity. Dose-response curve of LGLL patients (red line) and healthy controls (gray line) for azacitidine. B CD8+ T cells from a T-LGLL patient harboring STAT3 Y640F mutation were cultured in the presence of azacitidine (1 or 10 μM) or absence (DMSO alone) for 24, 48, and 72 h. Cell lysates were immunoblotted with antibodies against p-STAT3, STAT3, DNMT1, and EZH2. LAMINB1, a loading control. Data are representative of three T-LGLL patients harboring STAT3 Y640F and quantification is presented in Supplementary Fig. 9. C Dose-response curve of KAI3 NK cells stably expressing STAT3 WT (gray line), STAT3 G618R (blue line), STAT3 Y640F (red line) for azacitidine (n = 2 per group). DSS scores are presented in Supplementary  Table 5. D KAI3 NK cells stably expressing STAT3 WT , STAT3 G618R and STAT3 Y640F were cultured in the presence of azacitidine (1 or 10 μM) or absence (DMSO alone) for 24, 48, and 72 h in RPMI1640 with reduced IL-2 (25IU/ml). Cell lysates were immunoblotted with antibodies against p-STAT3, STAT3, DNMT1, and EZH2. LAMINB1, a loading control. Data are representative of two independent experiments. Quantification of the proteins is presented in Supplementary Fig. 10. E Global DNA methylation levels in KAI3 NK cells of STAT3 WT , STAT3 G618R and STAT3 Y640F (for each group, n = 3) treated with 1 μM of azacitidine (upper) and 10 μM of azicitidine (lower) at 24, 48, and 72 h in RPMI1640 with reduced IL-2 (25IU/ml). %5-mC were evaluated by fluorescence according to manufacturer's instruction. Error bar present mean ± SD (n = 3 per group), and statistically significant difference was evaluated using Mann-Whitney U test. **P < 0.01; ***P < 0.001; ****P < 0.0001. F ROS expression in KAI3 NK cells stably expressing STAT3 WT , STAT3 G618R , and STAT3 Y640F . Cells were cultured in the presence of azacitidine (1 or 10 μM) or absence (DMSO alone) for 72 h in RPMI1640 with reduced IL-2 (25IU/ml). Cells were labeled with ROS-fluorescent dye then analyzed using flow cytometer. Error bars expressed as mean ± SD, and statistically significant difference was evaluated using unpaired T-test. **P < 0.01; ***P < 0.001. The representative of three independent experiments is presented in Supplementary Fig. 12A. G p-p65 and p65 protein expression level from azacitidine (1 or 10 μM) treated CD8+ T cells of T-LGLL patient harboring STAT3 Y640F mutation. Data are representative of three T-LGLL patients harboring STAT3 Y640F . H p-p65 and p65 protein expression level from azacitidine (1 or 10 μM) treated KAI3 NK cells stably expressing STAT3 WT , STAT3 G618R , and STAT3 Y640F . Data are representative of two independent experiments. patient cells, we used KAI NK cells and cytokinestimulated CD8+ T cells to validate the consequences of STAT3 pathway activation although it needs to be taken into account that due to gain of function STAT3 mutations, the STAT3 activation in T-LGLL is chronic compared to transient changes induced by cytokine stimulation. Our results were well recapitulated in these models, and we observed higher-order protein complex interactions between STAT3 and EZH2, HDAC1, and DNMT1 in both healthy CD8+ T cells with proinflammatory cytokine treatment and KAI3 NK cells overexpressing GOF STAT3 variants, indicating a direct link between STAT3 and epigenetic regulation. Azacitidine, a hypomethylating agent, rescued these phenotypes in SHP1 dependent manner. Together, our results suggest a central role for STAT3 and epigenetic changes to drive LGLL pathogenesis, and provide rationale for testing hypomethylating agents in the treatment of LGLL patients resistant to standard therapies.
Prior studies have shown elevated levels of circulating inflammatory cytokines in LGLL [13,[41][42][43]. The exact mechanism and dominance of certain cytokines versus complex action of several cytokines remains largely enigmatic. We confirmed increased cytokine levels in T-LGLL patients compared to healthy controls. Increased global DNA methylation pattern was found both in CD8 + LGLL cells and in response to stimulation with IL-6, IL-15, and MCP-1 in healthy CD8+ T cells. These global epigenetic changes may partly confer a selection pressure on T cell  Supplementary Fig. 13. B SHP1 protein levels in CD8+ T cells of a T-LGLL patient harboring STAT3 Y640F mutation treated with azacitidine (1 or 10 μM) for 24, 48, and 72 h Data are representative of three T-LGLL patients harboring STAT3 Y640F and quantification is presented in Supplementary Fig. 14. C SHP1 expression in KAI3 NK cells (STAT3 WT , STAT3 G618R , and STAT3 Y640F ) treated with azacitidine (1 or 10 μM) for 24, 48, and 72 h in RPMI1640 with reduced IL-2 (25IU/ml) followed by Western blot assay. D KAI3 NK cells stably expressing STAT3 WT , STAT3 G618R , and STAT3 Y640F were treated with SHP1-siRNA (si-SHP1) and control siRNA (si-con) for 72 h followed by azacitidine treatment for 72 h in RPMI1640 including reduced amount of IL-2 (25IU/ml). Western blot was performed with the use of anti-pSTAT3, anti-STAT3, anti-DNMT1, anti-SHP1, and anti-LaminB1 antibodies. E KAI3 NK cells stably expressing STAT3 WT and GOF STAT3 variants (STAT3 G618R and STAT3 Y640F ) were cultured with SHP1-siRNA (si-SHP1) and control siRNA (si-con) for 48 h. Next, siRNAs transfected cells (10,000 cells/ well) were cultured in RPMI1640 with reduced IL-2 (25IU/ml) with extra siRNAs for 72 h, and the cell proliferation was measured by DNA fluorescence-based assay. Cell numbers were calculated with a standard curve based on the fluorescence intensity. Error bar present mean ± SD of three independent experiments, and statistically significant difference was evaluated using unpaired t-test (STAT3 WT si-con vs STAT3 WT si-SHP1). *P < 0.05; **P < 0.01. F SHP1 expression in KAI3 NK cells transfected with STAT3-siRNA (si-STAT3) and control siRNA (si-con) for 72 h. The cells were serum starved for 12 h followed by Western blot assay. G Simplified scheme of the effect of azacitidine. Pathway components discovered to be upregulated or downregulated with functional assays are marked with arrows. The scheme was made using BioRender (Toronto, Canada). clones, facilitating clonal expansion after cytokine stimulation. While the increased total methylation pattern in response to cytokine stimulation in CD8+ T cells was confirmed in the fluorescence-based assay, DNA methylation sequencing analyses are warranted to explore the methylated genomic sites in detail. In addition, further studies using primary LGLL cells could link alterations in cytokine profiles and global DNA methylation with key gene expression profiles.
We observed higher levels of circulating IL-15RA in T-LGLL patients samples, and IL-15 promoted epigenetic changes in healthy CD8+ T cells. IL-15-producing macrophages and dendritic cells have been reported to crossrepresent IL-15 to promote proliferation and cytotoxicity in LGL [19]. IL-15 can initiate LGLL in transgenic mice via the induction of MYC, p65, and HDAC1 with DNMT3B overexpression and DNA hypermethylation [17]. In addition, IL-15 can maintain the survival of leukemic clones in T-LGL and other T malignancies [44]. We also observed the activation of these proteins in healthy CD8+ T cells stimulated with cytokines, and validated these findings in KAI3 NK cells overexpressing GOF STAT3 mutations, indicating that enhanced p-STAT3 may promote LGLL pathogenesis in part via induction of these oncogenic pathways by cytokine stimuli.
STAT3 interacts with many key epigenetic regulators and other transcription factors, coactivators, and corepressors to reprogram transcription [45][46][47]. DNMT1 is closely associated with STAT3 signaling in malignant T cells [48], and our results suggest that STAT3 directly binds DNMT1, EZH2, and HDAC1 to regulate DNA methylation. We showed that STAT3 interacts with DNMT1 and HDAC1 both in CD8+ T cells and KAI3 NK cells, which is in line with previous studies showing STAT3-DNMT1-HDAC1 complex in lymphoma cells [49]. These complexes may contribute to the link between p-STAT3 activation and global DNA methylation.
ROS levels were higher in CD8+ T cells of LGLL patients compared to healthy controls, which is consistent with previous studies showing increased level of ROS in hematological malignancies [50,51]. KAI3 NK cells expressing STAT3 GOF variants had higher levels of ROS compared to cells overexpressing STAT3 WT , suggesting that ROS is associated with STAT3 activation. We also observed increased ROS levels after 12-hour cytokine stimulation in healthy CD8+ T cells. Furthermore, we found increased activation of both DNA damage and NF-κB pathways in STAT3 mutated CD8+ T cells of LGLL patients and healthy CD8+ T cells after cytokine stimulation. Rapidly accumulating DNA damage upon stimulation of CD8+ T cells may lead to increased mutagenesis and clonal selection, while chronic elevation in ROS and DNA damage may be markers of increased clonal turnover in T-LGLL patients. Given that inflammatory cytokines are commonly associated with chronic inflammation and ROS production [52], our results suggest that STAT3 activation, high ROS levels, DNA damage, and NF-kB signaling are associated with LGLL phenotype.
While current LGLL treatment protocols rely on immunosuppressive therapies, targeted therapies such as NF-κB inhibitors are under clinical investigations [1]. We demonstrated that DNMT inhibition can potently inhibit the STAT3-DNA methylation pathway in both CD8+ T cells of LGLL patients and KAI3 NK cells expressing activating STAT3 variants. Azacitidine, a therapeutic agent with DNA hypomethylating activity as the primary mechanism of action [53], inhibited ROS dependent NF-κB activation as well as STAT3 activation via restoration of SHP1, which is in line with a previous study using CD34 cells from AML patients [54] and anaplastic large cell lymphoma cells [55]. The negative correlation between SHP1 and p-STAT3 after azacitidine treatment in KAI3 NK cells is consistent with previous studies in other cell types [33,56,57]. SHP1 is primarily expressed in hematopoietic cells and is associated with DNA methylation and leukemogenesis [58]. As STAT3 can promote SHP1 gene silencing by forming complexes with DNMT1 and HDAC1 [49], the observed restoration of SHP1 expression by azacitidine might be explained by downregulated STAT3 activation. Conversely, siRNA targeting SHP1 was sufficient to abrogate the inhibitory effect of azacitidine on STAT3, suggesting that SHP1 is required for STAT3 inhibition by azacitidine treatment. The effect of azacitidine treatment on upregulating SHP-1 was confirmed in CD8+ T cells collected from a LGLL patient harboring STAT3 Y640F mutation, suggesting that azacitidine may be used to target STAT3 mutated cells in LGLL patients. The treatment of LGL cells with demethylating drugs has also been shown to lead to upregulation of SOCS3 providing additional mechanism for observed STAT3 downregulation [13].
In conclusion, our results indicate that proinflammatory cytokine stimulation and STAT3 GOF mutations lead to epigenetic reprogramming, increased global DNA methylation, increased ROS production, and enhanced proliferation that may drive LGLL disease pathogenesis. The activity of this axis could be partially rescued by DNMT inhibition, providing novel targeted treatment options for LGLL. was performed at the Institute for Molecular Medicine Finland FIMM Technology Centre, which is supported by Biocenter Finland. We acknowledge IT Center for Science Ltd for computational resources. Finnish Red Cross Blood Service is acknowledged for providing healthy control samples.
Data sharing and declaration RNA sequencing data are available from the corresponding author upon request. All other data are available in the Supplementary files.
Author contributions DK, MM, GP, and SM designed the study and experiments. JH, BG, and MK performed sequencing analysis. HR provided the clinical data and patient samples. RM provided important idea and critical comments on this study. WC provided study materials and input to study design. DK, MM, and SM wrote the manuscript with the input of other authors. All authors read and approved the final manuscript.
Funding Open access funding provided by University of Helsinki including Helsinki University Central Hospital.

Compliance with ethical standards
Conflict of interest SM has received honoraria and research funding from Novartis, Pfizer and Bristol-Myers Squibb (not related to this study). MM has received honoraria from Celgene and Sanofi. The remaining authors declare no competing financial interests.
Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.