NKT-Like (CD3+CD56+) Cells in Chronic Myeloid Leukemia Patients Treated With Tyrosine Kinase Inhibitors

Therapy with Tyrosine Kinase Inhibitors (TKI) aiming stable deep molecular response is the gold standard to treat Chronic Myeloid Leukemia (CML). NKT-like cells (CD3+CD56+) combine characteristics of T and NK cells. The physiopathological role of these cells remains unknown although the literature refers their association with inflammation, autoimmune diseases, and cancer. Since the information regarding the role of NKT-like cells in CML is rare, we aimed at the characterization of these cells in CML patients treated with TKIs. Peripheral blood NKT-like cells from 48 CML patients and 40 healthy donors were analyzed by multiparametric flow cytometry. Functional tests consisting of co-culture with leukemic target cells (K562 cell line) were used to measure degranulation and cytokine production. Our results revealed that NKT-like cells are decreased in treated CML patients, although they present increased expression of activation markers (CD69 and HLA-DR), increased degranulation (CD107a) and impaired IFN-γ production. Significantly alterations on the expression of tumor recognition (NCRs and NKp80), and immune regulation receptors (LAG-3, TIM-3, and CD137) by NKT-like cells were observed in CML patients. Second generation TKIs increased cell activation (CD69) and decreased expression of NKp44 and NKp80 by NKT-like cells from CML patients when compared to Imatinib. CML patients that achieved deep molecular response (MR4.5) presented downregulation of NKp44 and LAG-3. Further studies are needed to clarify the role of these cells as biomarkers of therapy response and also to evaluate their value for discrimination of better candidates for sustained treatment-free remission after TKI discontinuation.


INTRODUCTION
Chronic Myeloid Leukemia (CML) is a myeloproliferative disease that develops from a clonal expansion of hematopoietic cells in the bone marrow (BM) with consequent accumulation of immature cells in peripheral blood (1,2). CML biology is characterized by the expression of constitutively activated tyrosine kinase proteins (3), that are encoded by the BCR-ABL oncogene (4). BCR-ABL is present in the Philadelphia chromosome (5) generated from a reciprocal t (9:22) translocation (6). The introduction of Imatinib and new generations of tyrosine kinase inhibitors (TKIs) represented a shift in chronic phase CML (CP-CML) treatment (7). With TKIs, an higher proportion of patients achieve long-term deep molecular responses (DMR) and the life expectancy of newly diagnosed patients gets close to age-matched normal individuals (8,9).
Is well known that TKIs have off-target immunomodulatory effects, namely on effector and regulatory T cells, NK cells, B cells, and dendritic cells. Moreover, immune reactivation in CP-CML patients has been associated with TKI therapy (10)(11)(12)(13)(14)(15). In the context of immunomodulation, the 2nd generation TKI Dasatinib is the most interesting, since it has targets that are directly implicated in immune regulation (16)(17)(18)(19) and it is associated with large granular lymphocytosis, resulting in expansion of T CD8 and NK cell clones (20).
Natural killer T cells (usually defined as CD3 + CD56 + ), are a poorly known, controversial and heterogeneous population that shares characteristics from both NK and T cells. The classification of NKT cells has been used to define different subpopulations of T cells expressing NK receptors, such as CD1d-restricted cells with invariant TCR (iNKT) or CD8 T cells that acquire NK receptors (NKT-like cells) (21,22). Whereas iNKT frequency decreases, NKT-like cells increase with age in peripheral blood of healthy individuals (22). It has been shown that iNKT cells from chronic phase CML patients show functional deficiencies that are restored upon remission, although their possible contribution to disease control by TKI based therapies is unclear (23).
Considering that CML and TKI therapy induce changes in phenotype and function of immune cells (10)(11)(12)(13)(14)(15), we performed extended immunophenotyping of NKT-like cells, including functional tests (degranulation and IFN-γ production), maturation, activation, and migration status markers and also detailed analysis of NKG2 family receptors, NCRs, NKp80 and immune checkpoints (ICP) expression on NKTlike cells from CML patients treated with tyrosine kinase inhibitors. We identified low numbers of NKT-like cells in peripheral blood from CP-CML patients with cytotoxic potential and differences in the repertoire of receptors. The latter was more evident for receptors linked to activation and immune regulation.

Patients and Healthy Donors
Peripheral blood (PB) samples were collected in heparin tubes, in average 12 h after the drug intake and analyzed within 24 h. The study group consisted of 48

Functional Tests: Degranulation and IFN-γ Production
In aseptic conditions, effector cells (peripheral blood mononuclear cells, PBMCs) were isolated from heparinized blood samples by a standard "Ficoll-Hypaque" (Histopaque, Sigma Aldrich, UK) protocol. PBMCs were counted to adjust the concentration to 50 × 10 6 cells/mL. Target cells (K562 cell line; ATCC CCL-243, Manassas, VA, USA) were maintained in culture, viability was determined and cell concentration adjusted to 1 × 10 5 cells/mL. Into 12 × 75 mm tubes, 75 µL of effector and 100 µL of target cells were mixed in a 25:1 effector ratio, with 2 µL of anti-CD107a-PE antibody (BD Pharmingen, San Jose, CA, USA) and 1 µL of Brefeldin A (Sigma, 10 µg/mL) in a final volume of 200 µL with RPMI 1640 medium (Invitrogen, Carlsbad, CA, USA). Tubes were placed in a humidified 5% (v/v) CO 2 incubator in the water reservoir at the bottom, for 4 h at 37 • C. At the end of incubation cells were washed and resuspended in 100 µL of 1× phosphate-buffered saline (PBS; Invitrogen, Carlsbad, CA, USA) and the extracellular antibodies were added, anti-CD56-PerCP-Cy5.5 (BD Pharmingen, San Jose, CA) and anti-CD3-V500 (BD Horizon, San Jose, CA, USA). After 15 min of incubation in the dark at room temperature (RT), suspensions were treated with Fix and Perm A solution (Invitrogen, Carlsbad, CA, USA) for 15 min, in the dark at RT. Cells were centrifuged at 453 g for 5 min and the supernatant discarded. Next, cells were incubated with Fix and Perm B solution (Invitrogen, Carlsbad, CA, USA) and the intracellular antibodies anti-IFN-γ-V450 (BD Horizon, San Jose, CA, USA) and anti-Granzyme-B-FITC (BD Pharmingen, San Jose, CA, USA) for 20 min, in the dark at RT. After centrifugation at 453 g for 5 min and supernatant discarded, cells were resuspended in 1× PBS and acquired in FACS Canto II with DIVA software (Becton Dickinson, San Jose, CA, USA). For each sample, in the same conditions, effector cells were incubated without target cells to detect spontaneous degranulation and cytokine production.

Immunophenotyping
White blood cells from fresh peripheral blood samples were counted in the hematological counter (A c T diff, Beckman Coulter, Brea, CA, USA). 100 uL or up to 1 × 10 6 cells were added to 12 × 75 mm tubes and stained with the extracellular antibodies for 15 min, in the dark at RT. After incubation red blood cells were lysed with BD Lysing Solution (Becton Dickinson, San Jose, CA, USA), for 10 min. Cell suspensions were centrifuged and washed one time with 1× PBS in same conditions. In the end, cell suspensions were acquired in FACS Canto II with DIVA software (Becton Dickinson, San Jose, CA, USA). The antibodies used for cellular staining are described in Supplementary File 2.

Data and Statistical Analysis
Flow cytometry data was analyzed with FlowJo v.10.7 (Tree Star, Ashland, OR, USA). The gating strategy used for the functional assays and immunophenotyping is similar. Lymphocytes were discriminated based on morphologic parameters SSC-A and FSC-A and doublets excluded using FSC-H and FSC-A. Gated on lymphocytes, NKT-like cells were identified by the expression of both CD3 and CD56 receptors. Within the NKT-like population, it was determined the frequency of positive cells and the median fluorescence intensity (MFI) of CD11b, CD16, CD27, CD56, CD57, CD62L, CD69, CD107a, CD137, CD137L, CRACC, HLA-DR, LAG-3, NKG2A, NKG2C, NKG2D, NKp30, NKp44, NKp46, NKp80, PD-1, TIM-3, and the production of Granzyme-B and IFN-γ. The cut-off for extracellular receptors or intracellular proteins was defined using isotype controls specific for each marker (described in Supplementary File 2). The statistical   analysis was performed with GraphPad Prism version 8.0.1 for Windows (GraphPad Software, San Diego, CA, USA) and data is presented as mean ± standard deviation. Paired ttest and Mann Whitney U-test was used to compare means between two groups. One-way ANOVA followed by Dunn's multiple comparisons test was used to compare more than two groups and Spearman correlation to evaluate the association between two parameters. ClustVis was accessed online (https:// biit.cs.ut.ee/clustvis) for visualizing clustering of multivariate data using Principal Component Analysis and heatmaps (40).
Original values were ln(x+1)-transformed. Unit variance scaling was applied to rows; singular value decomposition (SVD) with imputation was used to calculate principal components. X and Y axis show principal component 1 and principal component 2 that explain the indicated percentages of the total variance. Prediction ellipses are such that with probability 0.95, a new observation from the same group will fall inside the ellipse. Clustering distances were obtained using Pearson correlation subtracted from 1. Ward linkage method was calculated using sum of squared differences from points to centroids as the distance.

RESULTS
No statistically significant differences were observed for CML patients concerning absolute and relative frequencies of NKT-like cells according to sex, age, sample time after diagnosis (Supplementary File 3) and clinical risk scores Sokal (41), EURO (42), EUTOS (43), and ELTS (44) (Supplementary File 4).
Similarly, no statistically significant differences were observed between individual TKIs and after analysis of CML patients according to therapy response (Supplementary File 5). Nevertheless, we further analyzed NKT-like cells parameters considering TKI generation and deep molecular response (DMR). The gating strategy to identify NKT-like cells is shown in Figure 1A.

NKT-Like Cells Are Decreased in CML and Present a Different Receptor Repertoire According to TKI Generation and Molecular Response
We analyzed the relative and absolute frequencies of lymphocyte and NKT-like populations in peripheral blood from CML patients and healthy donors (HD) (Figure 1B). Our results revealed no significant differences in the relative and absolute frequencies of lymphocytes in treated patients when compared to healthy donors. The relative and absolute frequency of NKTlike cells from treated CML patients [(7.5 ± 4.0 vs. 5.1 ± 4.9; p = 0.0004) (150 ± 91 vs. 103±129; p = 0.0007), respectively] are diminished compared to healthy donors, with strong statistical value ( Figure 1B). However, patients treated with 2nd generation of TKIs revealed a significant increase in the absolute count of lymphocytes when compared to those treated with 1st generation TKIs (2,391 ± 917 vs. 1,737 ± 587; p = 0.0391) ( Figure 1C). No significant difference was observed in NKT-like cells for CML patients treated with 2nd generation TKIs ( Figure 1C). Similarly, no correlation was observed in lymphocytes neither in NKT-like cells for CML patients that achieved deep molecular response to therapy ( Figure 1D).
Principal Component Analysis using relative frequency of NKT-like cell subsets, according to the markers used in this study (Figure 2A) demonstrated that CML patients cluster with HD, although more dispersed than those control samples. Similarly, combined analysis of TKI generation and molecular response (Figure 2B), cluster 1st generation TKI CML patients that achieved DMR closer to HD. The figure also shows vicinity of CML patients treated with 2nd generation TKIs, independently of molecular response.
Cluster analysis revealed by heatmaps identified different patterns for the density of receptors expressed by NKT-like cells ( Figure 2C) when comparing CML patients with HD. Combined analysis of TKI generation and molecular response ( Figure 2D) revealed different patterns of NKT-like cell expression in the analyzed groups.

Increased Degranulation and Decreased IFN-γ Production by NKT-Like Cells in Treated CML Patients
In order to evaluate NKT-like cytotoxicity and cytokine production, we performed an indirect cytotoxic assay (CD107a, degranulation test) after 4 h co-culture with K562 cell line (Figure 3). Adding Brefeldin A, we also measured Granzyme-B and IFN-γ accumulated at the Golgi complex/endoplasmic reticulum of NKT-like cells in the same conditions. Although our results do not revealed significant differences in the percentage of NKT-like degranulating cells, we observed a significant increase of CD107a expression by NKT-like cells when comparing treated patients with HD (MFI: 1,109 ± 495 vs. 1,525 ± 315; p = 0.0002) ( Figure 3B). Contrarily, IFN-γ produced by NKT-like cells from treated patients were decreased compared to healthy donors (MFI: 427 ± 164 vs. 118 ± 149; p < 0.0001) (Figure 3B). Granzyme-B produced by NKT-like cells in vitro was not affected in CML treated patients (Figure 3B). No effect of 2nd generation TKIs and deep molecular response was observed for CD107a, Granzyme B, and IFN-γ (data not shown). generation TKI (n = 20 and n = 9, respectively) and 2nd generation TKI (n = 10). Mann Whitney U-test was used for statistical analysis and the charts represent the mean ± standard deviation. HD, Healthy donors; CML, treated CML patients; TKI, Tyrosine Kinase Inhibitor; 1st gen. TKI, 1st generation TKI CML patients; 2nd gen. TKI, 2nd generation TKI CML patients; MFI, Median Intensity Fluorescence; p-value <0.05*, <0.01**, <0.001***, or <0.0001****.

Maturation and Migration Markers Expressed by NKT-Like Cells in CML
We analyzed the expression of surface markers on NKT-like cells, indicative of NK and T cells maturation (CD11b, CD27, and CD57) and migration (CD62L

NKT-Like Cells Express Activation Markers in Treated CML Patients
The expression of surface receptors linked to activation of NK and T cells were also analyzed on NKT-like cells (Figure 4).
The gating strategy to identify activation markers expressed by NKT-like cells and heatmap showing their relative frequency distribution according to disease, TKI generation and molecular response are shown in Figures 4A,B, respectively. When compared to HD, treated CML patients presented a significant increase in the relative frequency of HLA-DR NKTlike cells (Percentage: 7.6 ± 6.2 vs. 16.5 ± 10.6; p = 0.0005) and surface density of HLA-DR (MFI: 101 ± 31 vs. 164 ± 34; p < 0.0001) ( Figure 4C).
No alterations were observed for the expression of CD16 and CD2-like receptor-activating cytotoxic cell (CRACC) receptor, a co-stimulatory molecule, on NKT-like cells (Figure 4B).

NKG2 Receptor Family on NKT-Like Cells in CML Patients
Concerning the NKG2 family, characteristic from NK cells, we evaluated the expression of the inhibitory receptor NKG2A and the activating receptors NKG2C and NKG2D on NKT-like cells from CML patients. No differences were found either in the frequency of NKT-like cells expressing these receptors or in the receptor surface density between the groups analyzed (data not shown).

NCRs and NKp80 Are Significantly Altered on NKT-Like Cells in CML Patients
The gating strategy to identify NCR + and NKp80 + NKT-like cells and heatmap showing their relative frequency distribution according to disease, TKI generation and molecular response are shown in Figures 5A,B, respectively.
CML patients achieving deep molecular response showed ( Figure 5E) a significant reduction of the density of NKp44 at the surface of NKT-like cells (MFI: 48.1 ± 12.9 vs. 67.9 ± 16.8; p = 0.0101).

Immune Checkpoints Expressed by NKT-Like Cells in CML Patients
The gating strategy to identify immune checkpoints expressed by NKT-like cells and heatmap showing their relative frequency distribution according to disease, TKI generation and molecular response are shown in Figures 6A,B, respectively.

DISCUSSION
In CML, the TKI-dependent mechanisms underlying immune regulation are not clear. However, it is widely recognized that TKIs have immunomodulatory effects that influence response to therapy and also according to drug generation (10)(11)(12)(13)(14). Several authors reported the immune restoration of PB lymphocytes levels in treated CML patients and also the particular effect of 2nd generation TKI (14,15,39). Our results corroborate the preceding studies. We observed normal lymphocyte counts in TKI treated CML patients that were more evident for patients undergoing 2nd generation TKIs.
To our knowledge, there is only one study that analyzed NKT-like cells as a whole population in CML and described no differences in PB of TKI treated patients, with exception of those under Dasatinib (39). Our results revealed a decrease in relative and absolute frequency of NKT-like cells in PB of treated CML patients without discrimination between first or second generation TKI. This discrepancy can be explained by the significant difference in the sample time after diagnosis between the two CML studies.
In our study, NKT-like cells from treated CML patients exposed to the leukemic cell line K562 upregulated CD107a expression and decreased IFN-γ production. Lack of other studies for comparisons allows us to speculate that NKT-like cells had increased cytotoxicity as a major TKI-dependent antileukemic effect. No obvious explanation for the significant downregulation of IFN-γ recommends further studies in order to clarify these results.
Concerning CD62L, a molecule correlated with migration, we observed that patients treated with 2nd generation TKIs had lower density of CD62L on NKT-like cells. This is in line with CML studies on T cells that demonstrated downregulation of CD62L after TKI therapy (45). In particular, it has been reported that dasatinib-treated patients had less CD62L-expressing CD4+ and CD8+ T cells (39,46) whereas in nilotinib-treated patients, the expression levels of CD62L were similar as in imatinibtreated patients.
CRACC receptor could act as activator or inhibitory trigger in immune cells depending on the microenvironment (47). Our results do not suggest a clear role for CRACC NKT-like cells in treated CML patients. We assumed CD69 and HLA-DR expression by NKT-like cells as the classical early/late activation markers on T cells, respectively. Thus, we demonstrated that treated patients have more HLA-DR+ NKT-like cells and more HLA-DR per cell and revealed a higher density of CD69. The expression of these activating markers was in accordance with our results that showed higher degranulation by NKT-like cells from treated CML patients.
Despite the very low expression of NCRs, we found higher frequencies of NKp30, NKp44, and NKp46 NKT-like cells and increased surface density of NKp46 in our group of treated CML patients. The previous data suggest a more activated and anti-tumor profile of NKT-like cells from treated CML patients. On the other hand, we found that these cells express less NKp44. This receptor was both diminished by 2nd generation TKIs and in CML patients achieving deep molecular response. Additionally, the most evident finding regarding NK receptors was the significant decrease in both frequency of NKp80 NKTlike cells and density of receptor expression in TKI treated CML patients. In our study, downregulation of NKp80 was also associated to 2nd generation TKIs. It has been speculated that leukemic cells can self-downregulate the expression of NKp80 ligand preventing NKp80-recognition (48).
The expression of immune checkpoints (ICP) on lymphocytes is associated with immune regulation and, in line with our results, some authors reported the upregulation of ICP associated with immune suppression on PB T cells from CML patients or patients with other hematologic malignancies (49)(50)(51)(52)(53). To our knowledge, there are no studies analyzing the expression of ICP on NKTlike cells in CML. In spite of the very low or absent expression of ICP on NKT-like cells, we observed an increased fraction of NKT-like cells expressing TIM-3 and/or LAG-3 in treated CML patients and decreased expression of LAG-3 associated to deep molecular response.
In summary, we studied TKI treated CML patients and found decreased frequency of NKT-like cells, higher cytotoxicity, and different patterns of membrane receptors (Figure 7). Major differences in receptor expression were associated with activation (HLA-DR, CD69), tumor recognition (NCRs and NKp80), and immune regulation (TIM-3 and LAG-3). Second generation TKIs were associated with remarkably activation status (CD69) and downregulation of NKp44 and NKp80. Deep molecular response correlates with downregulation of LAG-3. Those could represent potential markers for sustained treatment-free remission (TFR) after discontinuation.
Further studies are needed to clarify the importance of the NKT-like mechanisms cells involved in CML. Unmet needs include the analysis of the interactions between NKT-like cells and leukemic cells, the effect of TKIs on specific NKT-like cells pathways affecting function and phenotype, and the role of NKTlike cells on TKI discontinuation.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the manuscript/Supplementary Files.

ETHICS STATEMENT
All the volunteers agreed and signed informed consent to participate in the present study approved by the Ethical Committee of the Faculty of Medicine of the University of Coimbra and the Coimbra Hospital and University Centre (Portugal).  This work was also supported by grants PI13/02691 and PI16/01615 (to RS) from Spanish Ministry of Health, SAF2017-87538-R (to RT) from the Ministry of Economy and Competitiveness of Spain and IB16164 and R18085 from Junta de Extremadura (to RT) co-financed by European Regional Development Funds.