Profile of differentially expressed Toll-like receptor signaling genes in the natural killer cells of patients with Sézary syndrome

Sézary syndrome (SS), an aggressive and leukemic form of cutaneous T-cell lymphoma, usually results in shortened survival. Improving innate immunity in SS by targeting natural killer (NK) cells with Toll-like receptor (TLR) agonists could be an interesting modulatory strategy. We evaluated the NK cell populations in SS patients assessing activating and inhibitory receptors expression and profiled the differential expression of TLR signaling pathway genes in unstimulated NK cells and after TLR7/8 stimulation. We observed preserved CD56bright NK cells and a low percentage of CD56dim NK cells in the peripheral blood of SS patients compared to those in the healthy control group. Both NK cell populations showed down-modulation of NKG2C and NKG2D expression, which was associated with high serum levels of the soluble form of NKG2D ligands. In contrast, an expansion of “memory” CD57+ NKG2C+ NK cells and high cytomegalovirus antibody titers were detected in SS patients. Profiling of the TLR signaling genes in NK cells from SS patients showed an abundance of differentially expressed genes (DEGs) in NK cells in the unstimulated condition, with mostly up-regulation of NFκB/JNK p38 pathway genes, but there was down-regulation of type I (IFN-α/β) and II (IFN-γ) interferon and IL-12A. After activation of NK cells with TLR7/8 agonist, the down-regulated genes correlated with the IFN response, and IL-12 became up-regulated, together with other antitumor factors. NK cell activation with a dual agonist for TLR7 and TLR8 is able to induce the expression of IFN-γ and type I IFN, which can improve immunity in SS patients.


INTRODUCTION
Sézary syndrome (SS) is an aggressive and advanced form of CTCL with circulating malignant T cells. Impaired immunity in SS [1,2] probably is crucial for disease progression [3]. Investigation how to improve SS immunity may contribute to therapeutic strategies.
Natural killer (NK) cells are cytotoxic type 1 innate lymphoid cells with potential for cancer immunotherapy and for treating viral infections. There are two NK cell populations: CD56 dim CD16 bright NK cells are predominant in the peripheral blood, whereas NK cells from secondary lymphoid tissue and from other tissues are primarily CD56 bright NK cells [4]. Moreover, in vivo, human CD56 bright NK cells may undergo progressive differentiation toward CD56 dim NK cells [5].
Major NK cell-activating receptors involved in cancer cell recognition include NKG2D. This activating immune receptor has been identified on NK cells, CD8 T cells, NKT cells, and subsets of γδ T cells [10]. NKG2D expression is variable on NK cells, and in some SS patients, expression is intensely down-regulated [11]. Human MHC class I chain-related genes (MICA and MICB) and ULBP1-5 have been recognized as ligands of NKG2D. At advanced tumor stages, sustained surface expression of NKG2D ligands and shedding of sMICA induces internalization and degradation of NKG2D, thus promoting tumor immune evasion [12]. Some Sézary cells express MICA, MICB and ULBP1 [11], as well as the soluble form of MICA (sMICA), as a possible mechanism to escape the immune system.
Expansion of an NK cell subset carrying an activating receptor heterodimer, comprising CD94 and NKG2C (CD94/NKG2C) could be elicited by human cytomegalovirus (HCMV) infection [13,14]. CD57+NKG2C+ NK cells seem to identify the final stages of peripheral NK cell maturation; the number of these cells increases with age, and they exhibit "memorylike" features [15][16][17]. To date, there is no evidence of the presence of NK cells expressing NKG2C and/or a memory profile in SS or in mycosis fungoides (MF), despite the fact that CMV seropositivity is highly associated with MF and SS [18].
The scientific and clinical interest in TLR7 and TLR8 in cancer biology has originated from the antitumor activities of these molecules in preclinical models [19]. Imiquimod is widely used as a topical treatment for cutaneous tumors, including basal cell carcinomas [20,21], keratoacanthomas [22,23], actinic keratoses [24][25][26], and cutaneous metastases of melanoma [27,28]. Moreover, imiquimod has been successfully used for the treatment of patch-and plaque-stage MF [29][30][31]. Imiquimod preferentially activates TLR7, but it exerts weak agonistic activity by TLR8 [32]. Synthetic substances, such as resiquimod (R848, ligand for TLR7/8), have been extensively studied either as single agents in experimental cancer models or as vaccine adjuvants in clinical trials [33,34]. Resiquimod induces more pronounced cytokine secretion, macrophage activation and cellular immune responses than does imiquimod [35,36]. Topical resiquimod can promote disease regression and enhance T-cell effector functions in cutaneous T-cell lymphoma [37]. Approximately 50% of patients exhibited activation of circulating dendritic cells and NK cells, indicating a systemic immune response, and untreated lesions typically regressed.
An attempt to enhance NK cell function using synthetic agonists of TLRs has been explored in SS. The combined effects of synthetic oligodeoxynucleotides (ODN) with the CpG motif (TLR 9 agonist) and IL-15 on the activation of NK cells in patients with CTCL led to induced activation of NK cells and enhanced activation of CD8 T cells [38]. The CD158k/KIR3DL2 marker may be able to help identify and enumerate neoplastic T-cells in SS, even when present at low levels [39]. Moreover, interaction of KIR3DL2 with CpG ODN can directly promote malignant cell death in SS [40]. Previously we observed a dysfunctional cytokine response in the PBMCs of SS patients upon TLR agonists (TLR2-TLR9) stimulation, emphasizing the role of TLR7/8 agonists (CL097), which partially restored inflammatory cytokine and type I and type II interferon (IFN) secretion [41]. Evaluating the effector response and the differential gene profile in NK cells upon TLR7/8 stimulation on the adaptor and effector molecules and the downstream TLR pathway may lead to the development of novel immunotherapies to enhance immunity in SS patients.
Here, we propose to investigate innate immunity in SS by evaluating NK cell subsets, activation/inhibitory receptor expression, soluble NKG2D ligand levels, "memory" phenotype NK cells, and the effector response induced by a TLR7/8 agonist and differentially expressed genes (DEGs) related to the TLR signaling pathway.

Activation/inhibition receptor expression in natural killer cell subsets in SS
Disturbance of innate immunity has been described in SS, including a reduction of NK cells [8]. Despite the reduced number and function of NK cells, they are potential effectors cells against Sézary cells [7], which favors the development of strategies to enhance NK cells. Therefore, first we analyzed the expression of activating/ inhibitory markers of NK cell subtypes from peripheral blood. Laboratory characteristics of SS patients are shown in Supplementary Table 1. The gating strategy for CD56+ NK cells is shown in Supplementary Figure 1.
A reduced number of NK cells ( Figure 1A) and CD56 dim NK cells ( Figure 1C), was observed in the peripheral blood of SS patients while the number of CD56 bright NK cells number ( Figure 1B) was not changed compared to HC group. Both NK cell subtypes CD56 bright ( Figure 1B) and CD56 dim ( Figure 1C) showed decreased NKG2D and NKG2C expressions in the SS group compared to HC group, with similar frequencies for NKp46 or NKG2A. We evaluated soluble MHC class I chain-related proteins A and B (sMICA/B) ligands of NKG2D in the serum of the patients. sMICA was detected in only 2/10 SS patients (SS7 and SS11), and sMICB was detected in 8/10 patients, at higher levels compared to HC samples ( Figure 1D). A negative correlation was observed between the percentage of CD56 bright NKG2D+ NK cells and sMICB levels, which may contribute to downregulation of NKG2D expression in the NK cells of SS patients. Also, a decreased percentage of CD3+NKG2D+ cells was revealed by flow cytometry (Supplementary Figure 2). CD56+CD57+NKG2C+ NK cells have been termed "memory" NK cells, and these cells expand in response to infection with CMV [16]. Of note, only 5 SS patients analyzed showed an increased percentage of CD56+CD57+ NKG2C+ NK cells compared to HC subjects ( Figure 2). All clinical evaluations were done with freshly obtained cells; at the time of the study, only 5 patients were available. Moreover, high CMV seropositivity was founded in SS patients (100%) as well as in HC (81%) subjects.
In addition, we analyzed the expression of NKG2C/ NKG2D mRNA and their ligands in unstimulated PBMCs from both analyzed groups. Decreased NKG2C and NKG2D expression were verified in the PBMCs of SS patients (Supplementary Figure 3). Similar mRNA expression levels of ULBP-3 and MICA were observed in both groups, while a decrease in the expression of HLA-E, a ligand for NKG2C and NKG2A, was observed in the PBMCs from SS patients compared to those from HC subjects. The two SS patients that showed up-regulated MICA mRNA expression (SS7, SS11) are not the same patients that showed increased sMICA levels (SS4, SS8) in serum. The data showed a decreased number of CD56 dim NK cells in the peripheral blood of SS patients, impaired expression of NK-activating receptors, such as NKG2D and NKG2C, and expansion of "memory" CD57+NKG2C+ NK cells, together with increased sMICA/MICB, which acts as tumor escape mechanism.
NK cell response to a TLR7/8 agonist Previously, we verified that TLR ligands have the ability to restore, at least partially, the expression of a broad range of cytokines and type I/III IFN in the PBMCs from SS patients [41]. We assessed the NK effector molecules, such as degranulating CD107a, IFN-γ and TNF, induced by TLR7/8 stimulation by means of flow cytometry; PMA plus ionomycin was used as a positive control. Figure 3 shows an increased percentage of CD16+CD56 dim NK cells in the unstimulated condition expressing CD107a and TNF in SS patients compared to HC subjects. After CL097 stimulation, CD56 bright NK cells demonstrated increased CD107a and TNF production compared to HC cells. A decreased IFN-γ response was detected in both NK cell populations with PMA plus ionomycin stimulation, while increased TNF secretion was observed in CD56 bright NK cells and decreased CD107a in CD56 dim NK cells in the SS group.

Profile of the TLR signaling pathway in NK cells
A dysregulation in TLR signaling can compromise the immune response. Therefore, we evaluated the expression pattern of central genes in TLR-related transduction pathways upon activation with a TLR7/8 agonist (CL097). We assessed purified NK cells to verify the direct effect of TLRs activation on NK cells. By PCR array, 84 genes of the TLR signaling pathway (Supplementary Table 3) were assessed in purified NK cells from the peripheral blood of SS patients (SS9, SS2, and SS20) and HC subjects (n=three/per group) in the unstimulated condition and after stimulation with CL097. The three patients selected for our study were among the few who survived. Figure 4 shows the profile of the differentially expressed genes (DEGs) in unstimulated NK cells. Among the 84 genes analyzed, 48 genes were over-expressed in SS patients in relation to HC, including several members of the NF-kB pathway and the JNK/p38 family ( Figure  4A and 4B). There were 19 under-expressed genes, notably those related to the IFN pathway, such as IRF3, TLR7, IFNA1, TLR3 and IL-12A ( Figure 4B). Moreover, only 17 genes showed similar expression between both groups. These findings show that even in the unstimulated condition, NK cells from SS patients have abundant differentially expressed TLR signaling genes. Figure 5A shows a heat map of the TLR signaling pathway genes after TLR7/8 stimulation in NK cells from SS and HC subjects. Again, a different profile of genes was detected in stimulated NK cells from SS patients, as shown by the volcano plot ( Figure 5B), in which 28 genes were up-regulated in SS patients, 29 were downregulated and 27 genes have similar expressions in both groups. Due to the clinical heterogeneity in SS patients, no statistical significance was detected in our analysis; however, the analyzed patients showed consistent expression profiles. Figure 5B shows some selected DEGs upon stimulation of TLR7/8. Notably, some genes that were repressed under normal circumstances in SS patients, especially those associated with IFN signaling, were up-regulated following CL097 stimulation, such as the cytokines IL-12A, IFNA1, IFNB1, IFN-γ and the IFN-γinducible chemokine CXCL10. In addition, the induction of IKBKB, which degrades the NFκB inhibitor IκB, may contribute to the activation of the NFκB pathway. In parallel, TLR7/8 stimulation also induced regulatory genes, such as IL-10, and anti-tumor factors, such as highmobility group box 1 protein (HMGB1) [42].
The ratio between up/down-regulated genes changed from 2.52 in the unstimulated condition to 0.96 after TLR7/8 activation, revealing that in the unstimulated condition most of the DEGs were up-regulated in NK cells from SS patients.

DISCUSSION
The innate lymphocytes NK cells are essential for innate immune responses against cancer and viral infections. Indeed, a genetic NK deficiency causes severe herpes virus infections and affects tumor clearance [43]. In SS, impaired immunity is one factor involved in disease progression. We verified that SS patients showed a reduced number of total NK cells and CD16+CD56 dim NK cells, while the CD56 bright NK population was preserved. The equilibrium percentage of CD56 bright NK cells in SS must be crucial in attempting to replace the number of terminally differentiated CD56 dim cells, which seems to be due to an impaired output of NK cells from bone marrow. It is known that the percentages and absolute numbers of NK cells increase during healthy aging, but the decrease in the CD56 bright subset is interpreted as a lower output of such NK cells from the bone marrow [44]. Moreover, the CD56 dim NK cell population is preferential for causing apoptosis in cancer [45].
NK cell receptors provide signals, which determine whether an NK cell becomes activated or inhibited. We detected a down-modulation of NK-activating receptors, such as NKG2D and NKG2C, regardless of NK population, while the inhibitory receptor NKG2A remained unchanged in SS subjects. In addition, decreased NKG2D mRNA expression was detected in the PBMCs from the SS group with wide NKG2D down-modulation, including by CD3+ T cells. Down-regulation of NKG2D could be due to factors such as TGF-β or metalloproteinases [46] and/or to NKG2D ligands, the production of which are a tumor evasion mechanism. TGF-β1 secreted by tumors is responsible for the poor NK lytic activity via downregulation of NKG2D [47]. TGF-β serum levels in SS patients were similar to those in the HC group (data not shown), whereas increased serum sMICB levels were found in the majority of SS patients. The high sMICB serum levels found in SS patients could originate from several cells, including Sézary cells. In fact, MICB rather than MICA is highly expressed by neoplastic cells from CTCLs, including those in SS patients [48]. In addition, MICB transcript levels were strongly correlated to those of T-plastin [48], a factor characteristically expressed by neoplastic T-cells of CTCL [49]; the correlation suggests a relationship to blood tumor burden.
A broad variability of NKG2D expression has been described in some SS patients (2/6), some with severe down-regulation [11]. In our SS cohort, NKG2D expression was heterogeneous, and at least 50% of patients showed half the level of the control group. In addition, an inverse correlation between the percentage of CD56 bright NKG2D+ NK cells and sMICB levels was detected. This finding suggests that NKG2D ligands may interact with receptors to induce internalization and degradation, contributing to its down modulation in SS patients.
An increased percentage of CD56+CD57+ NKG2C+NK cells was found in SS patients, together with high seropositivity to CMV. The subset of CD57+NKG2C+ NK cells has been described as exhibiting memory-like features and has potent effector functions [17]; HCMV infection is able to induce the expansion of CD94/NKG2C+ NK cells [13,50]. CMV seropositivity is highly associated with SS and MF, even in the earliest stages of the disease, and is significantly higher than that of healthy and immune compromised controls [51]. The immunosuppressive status seems to be responsible for CMV reactivation [52]. Interaction of CD94/NKG2C with its ligand HLA-E was shown to drive the expansion of the CD94-NKG2C+ NK cell population [14]. We verified decreased HLA-E expression in the PBMC of SS patients, whereas we did not assess other cells from tissue, such skin cells. The expansion of this mature CD57+NKG2C+ NK subset detected in SS patients could be due to memory for CMV infection, and the fact that this population display potent function may be beneficial for the patients.
Employment of adjuvants such as TLR agonists has been widely studied in APCs and other cells, including NK cells. Human NK cells express TLR3, TLR7, and TLR8 [53]. Previously, we verified that the compound CL097, a TLR7/8 dual agonist, was able to induce type I and II IFNs and other cytokines in the PBMCs of SS patients [41]. For  (HC 1-3). www.impactjournals.com/oncotarget NK cells, we observed that activation through the TLR7/8 pathway induced mainly degranulating factor CD107a and TNF by CD56 bright NK cells. Therefore, to better investigate the NK response following TLR7/8 activation, we assessed DEGs related to the TLR signaling pathway in NK cells. We used purified NK cells to verify the direct role of TLR activation, and we did not split the NK cells into subtypes due to the scarcity of NK cells in patients with SS. Of note, unstimulated NK cells from SS patients showed abundant DEGs of the TLR signaling pathway (48/84 genes). These up-regulated genes included NFκB, JNK/p38 and MyD88-dependent genes and revealed the activated status of NK cells. Among the TLRs, the most intensively expressed was TLR2, which was possibly a consequence of bacterial infection; this is not surprising considering that sepsis is a common cause of death among SS patients. This aligns with a previous finding that some patients with SS showed up-regulation of TLR2 mRNA expression in unstimulated PBMCs [41].
Other up-regulated genes expression in unstimulated NK cells from SS patients were prostaglandin synthase 2 (PTGS2/COX2), which exerts a negative effect on NK cell function via PGE2 through PGE2 receptor EP4 [54]. Moreover, PGS2 is a growth factor for malignant T cells in MF [55]. The expression of some pro-inflammatory cytokines, such as TNF and IL-1β, were up-regulated in NK cells demonstrating again the activated profile of NK cells from SS patients. In contrast, it is possible that TNF can also exert a pathogenic role. Evidence based on MF skin lesions has shown that TNF is involved in MF tumorigenesis because a cDNA microarray study demonstrated the promotion of antiapoptotic signaling by the deregulation of TNF signaling and an autocrine TNF feedback loop [56]. Although DEGs were mostly up-regulated in NK cells from SS patients under normal conditions, the downregulated DEGs included genes in the IFN pathway, such as IRF3, TLR7, IFNA1, TLR3 and IL-12A, and the IFNγ-inducible chemokine, CXCL10. Importantly, NK cells seem to contribute to the deficit in IFN-γ, type I IFN and IL-12 genes, which are well characterized in SS patients [2,29,57], for whom IFN-α and IFN-γ are recommended as a treatment [58,59].
Notably, the profile of DEGs after TLR7/8 stimulation showed that several inhibited genes became up-regulated, including those related to the type I and II IFN responses (IFNA1, IFNB1, IFNG, CXCL10) and IL-12 (IL-12A). We verified that CL097 was an interesting adjuvant to enhance the activity of NK cells in SS. Included in the NK genes up-regulated after CL097 activation was HMGB1, which can play an important role against tumor cells. NK cells invoke this strategy by employing HMGB1 protein, which specifically targets glycolysis in cancer cells and opens up new perspectives for cancer immunotherapy [42]. Moreover, NK cells, upon interaction with melanoma cells, can release a chemotactic form of HMGB1 protein that is capable of attracting additional activated NK cells [60]. Furthermore, another negatively regulated gene that becomes up-regulated in SS is IKBKB, which regulates the degradation of the NF-κB complex and favors the NF-κBsignaling pathway in NK cells.
However, some genes are down-regulated after stimulation, including IRF-1, a pivotal factor in the regulation of NK cells during infection, inflammation and metastasis [61], while PTGS2 is also down-regulated in NK cells, which could be favorable for NK function in SS.
Moreover, our data with TLR7-8 as adjuvant to potentiate the innate immunity, as NK cells, aligns with the data that topical resiquimod, an agonist of TLR7/8, is safe and effective in patients with stage IA-IIA CTCL, that can clear both treated and untreated skin lesions [37].
Our findings showed that NK cells in SS display different features than those in healthy subjects, with one aspect being decreased CD56 dim NK population number, down-modulation of NKG2D, and impairment of IFN-γ response, and the other revealing a preserved CD56 bright NK population and the presence of mature, memory CD57+NKG2C+ NK cells. In addition, NK cells exhibited a constitutively activated status with abundant DEGs; after CL097 activation, up-regulation of crucial genes of the IFN response and anti-tumor response occurred. These findings emphasize the use of TLR agonists as adjuvants to improve innate immunity with NK cells in CTCL.

Patients
We enrolled patients with Sézary syndrome (SS; n=13, 6 males, 7 females), with a median age of 61 years (range 48-84 years), from the Clinic of Cutaneous Lymphomas of the Department of Dermatology, Hospital das Clínicas, University of São Paulo Medical School, Brazil (HC/FMUSP). Diagnosis of SS was performed according to the revised clinical, histological, and biological criteria proposed by the International Society for Cutaneous Lymphomas (ISCL) and the cutaneous lymphoma task force of the European Organization of Research and Treatment of Cancer (EORTC) [62]. All SS patients have "de novo" erythrodermic; therefore, none of the cases were derived from MF progression. Healthy controls (HC; n=21, 10 males, 11 females), with a median age of 54 years (range 31-74 years), were selected from the Laboratory of Dermatology and Immunodeficiencies (LIM-56). Blood collection was performed before the initiation of treatment. Patients with dermatologic diseases, those using drugs, and those with a history of autoimmune diseases were not included in this evaluation. The exclusion criteria were treatment with immunosuppressant or immune-modifying drugs, pregnancy, and under 18 years of age. This study was approved by the São Paulo University Institutional Use Committee (CAAE, 07965312.0.0000.0068), and informed consent was obtained from all subjects. All experimental protocols within this study were performed in accordance with the Ethics Committee of this institution.

Real-time PCR
Total RNA was extracted from PBMCs using an RNeasyPlus Mini Kit (Qiagen, Valencia, CA, USA), and reverse transcription was performed with a Sensiscript Reverse Transcriptase Kit (Qiagen). The primers used in the real-time PCR assay for NKG2C, NKG2D, MICA, HLA-E, and UBLPB-3 are shown in Supplementary Table 2. GAPDH mRNA levels were used to normalize the mRNA content from PBMCs. PCR was performed on an Applied Biosystems 7500 system using specific primers and SYBR Green (Applied Biosystems, Carlsbad, CA, USA) fluorescence detection reagents. The cycling protocol consisted of 10 min at 95°C, followed by 40 cycles of 15 s at 95°C and 60 s at 60°C. The amplification results were visualized and analyzed using Sequence Detection System (SDS) software (Applied Biosystems). Normalized expression was calculated as previously described by Livak [63].

Determination of soluble NKG2D ligands by ELISA
Serum levels of ligands for the receptor NKG2D, such as MICA and MICB, were assessed by ELISA according the manufacturer recommendations (R&D System, Wiesbaden, German). Serology for HMCV was performed using a kit from Virion\Serion (Wurzburg, Germany).

Toll-like receptor signaling pathway PCR array
Purified NK cells from the PBMCs of three SS patients (SS9, SS2, and SS20) and three HC subjects by cell sorting were left in RPMI medium supplemented with 10% human AB serum (Sigma Aldrich, EUA) for 18 h at 37°C. Cells were stimulated with a TLR7/8 agonist (5.0 μg/mL) or with medium only (unstimulated) for 4 hours at 37°C. The cells were stored in RNA later solution (Sigma Aldrich, EUA) at -80°C. Total cellular RNA was extracted from NK cells using an RNeasy Micro Kit (Qiagen). The first strand of cDNA was synthesized based on the instructions for the RT2 First Strand Kit and PreAMP kit (Qiagen, Valencia, CA, USA), and the final cDNA product was used for the RT2 Profiler PCR Array using SYBR Green-based real-time PCR according to the manufacturer's protocol (PAHS-018Z array, Qiagen). The array consisted ofa panel of 84 genes relevant to the TLR pathway (Supplementary Table 3) plus five housekeeping genes (B2M, HPRT1, RPL13A, GAPDH and ACTB), a genomic DNA control, three reverse transcription and three PCR quality controls. Only samples passing the PCR array run quality control, which assessed the absence of genomic DNA contamination and proper amplification of the reverse transcription controls and the positive PCR controls, were further evaluated. ΔCt was calculated as the difference between the Ct of the TLR pathway genes and the geometric average of the Ct of the housekeeping genes. Log-fold differences in expression were reported using the 2-ΔΔCt or 2-ΔCt method [63].

Statistical analysis
The Mann-Whitney U-test was used to compare variables between the groups, the Wilcoxon matched pairs test was used to compare the baseline level with stimulated samples. p≤0.05 was considered significant.

ACKNOWLEDGMENTS
We are grateful to all the individuals who participated in the study. We thank the Core Facility for Scientific Research -University of Sao Paulo (CEFAP-USP) for the FLUIR (Flow Cytometry and Imaging Research).

CONFLICTS OF INTEREST
No conflicts of interest declared.