STAT3 and STAT5B Mutations in T/NK-Cell Chronic Lymphoproliferative Disorders of Large Granular Lymphocytes (LGL): Association with Disease Features

Simple Summary STAT3 and STAT5B mutations have been identified in a subset of T and NK large granular lymphocytic leukemia (T/NK-LGLL). The aim of our study was to evaluate the frequency and type of these mutations in all different subtypes of T/NK-LGL expansions (n = 100 patients), as well as to analyze its association with biological and clinical features of the disease. We show for the first time that STAT3/5B mutations were present in all different T/NK-cell LGLL categories here studied; further, STAT3 mutations were associated with overall reduced counts of almost all normal residual populations of immune cells in blood, together with a shorter time-to-therapy vs. wild type T/NK-LGLL. These findings contribute to support the utility of the STAT3 mutation analysis for diagnostic and prognostic purposes in LGLL. Abstract STAT3 and STAT5B (STAT3/STAT5B) mutations are the most common mutations in T-cell large granular lymphocytic leukemia (T-LGLL) and chronic lymphoproliferative disorders of NK cells (CLPD-NK), but their clinical impact remains unknown. We investigated the frequency and type of STAT3/STAT5B mutations in FACS-sorted populations of expanded T/NK-LGL from 100 (82 clonal; 6 oligoclonal; 12 polyclonal) patients, and its relationship with disease features. Seventeen non-LGL T-CLPD patients and 628 age-matched healthy donors were analyzed as controls. STAT3 (n = 30) and STAT5B (n = 1) mutations were detected in 28/82 clonal T/NK-LGLL patients (34%), while absent (0/18, 0%) among oligoclonal/polyclonal LGL-lymphocytosis. Mutations were found across all diagnostic subgroups: TCD8+-LGLL, 36%; CLPD-NK, 38%; TCD4+-LGLL, 7%; Tαβ+DP-LGLL, 100%; Tαβ+DN-LGLL, 50%; Tγδ+-LGLL, 44%. STAT3-mutated T-LGLL/CLPD-NK showed overall reduced (p < 0.05) blood counts of most normal leukocyte subsets, with a higher rate (vs. nonmutated LGLL) of neutropenia (p = 0.04), severe neutropenia (p = 0.02), and cases requiring treatment (p = 0.0001), together with a shorter time-to-therapy (p = 0.0001), particularly in non-Y640F STAT3-mutated patients. These findings confirm and extend on previous observations about the high prevalence of STAT3 mutations across different subtypes of LGLL, and its association with a more marked decrease of all major blood-cell subsets and a shortened time-to-therapy.

At present, diagnosis of LGLL remains a challenge. This is mostly due to the absence in a significant fraction of patients of tumor-associated phenotypic and/or molecular markers of clonality that would allow clear cut distinction among expansions of clonal vs. reactive/oligoclonal T cells [6,12,13], together with the lack of an universal marker of clonality for NK cells [14]. In addition, clonal/oligoclonal LGLs are also frequently detected in blood of asymptomatic subjects, either transiently (i.e., after allogeneic hematopoietic stem cell transplantation or solid organ transplantation) or persistently (i.e., in the elderly) [9,10,15]. So far, no reliable predictors of (indolent vs. aggressive) disease behavior have been established for T-LGLL and CLPD-NK.
Here we investigated the frequency and type of somatic mutations of the STAT3 and STAT5B genes in (purified) T-and NK-LGLs from a large series of T-LGLL and CLPD-NK patients. In addition, we analyzed the potential association between these gene mutations and both the lineage and biological features of clonal cells, the distribution of normal residual immune cells in blood, and other clinical features of the disease, including patient outcome.
Of note, STAT3/STAT5B mutations were absent in all non-LGL T-CLPD patients, except for an adult T-cell leukemia/lymphoma case that showed the STAT3-S614R mutation. Similarly, none of the 26 purified polyclonal non-LGL lymphoid populations and the seven (FACS-sorted) myeloid-cell populations analyzed (collected from seven different patients with STAT3-mutated LGLs), tested positive for STAT3/STAT5B mutations.
Results expressed as number (percentage) of mutated LGL cases or cell populations analyzed (in italics) per disease type. Panel B: Empty cell means 0 cases. * Gene mutation not previously described. A total of 3/28 patients showed two different clonal LGL populations with distinct STAT3 gene mutations: (i) 1/3 cases had the Y640F and Y657dup gene mutations in two different clonal populations of TCD8 + cells; (ii) 1/3 cases showed the S614R and G618R STAT3 gene mutations in one clonal Tγδ + population and one clonal TCD8 + -cell population, respectively; (iii) 1/3 cases displayed the Y640F and G618R mutations in one clonal CD56 −/+lo NK-cell population and one clonal TCD8 + -cell population, respectively. Abbreviations (alphabetical order): CLPD-NK, chronic lymphoproliferative disorders of NK cells; dup, duplication; LGL, large granular lymphocyte; Mut, mutation; N., number; NA, not analyzed; Pop., populations; SH2, Src homology 2; T-LGLL, leukemia of T large granular lymphocytes.

Discussion
Although STAT3 and STAT5B mutations are present in a significant percentage of all LGLL patients, their clinical and biological significance remain to be (fully) established [18][19][20][21]24,26,29,31,33]. Here we confirm and extend on previous observations by showing a high frequency of STAT3 mutations in both TCD8 + -LGLL and CLPD-NK [16][17][18], but also among Tγδ + -LGLL and all other T/NK-lineage LGLL here studied. In addition, we also demonstrate for the first time that in most LGLL subtypes, STAT3 mutations are associated not only with a more marked neutropenia (as also has been reported by other authors in TCD8 + -LGLL and CLPD-NK) [7,9,20,30,34], but also with overall reduced counts of major normal residual populations of immune cells in blood, independently of the number of coexisting LGL clones. Such overall reduced blood counts of (most) normal residual immune-cell subsets may reflect a more profound immunodeficiency status in STAT3-mutated LGLL, which is not restricted to neutrophils, but also affects other myeloid cells-i.e., eosinophils, nonclassical monocytes, and dendritic cells-as well as lymphoid cells-i.e., the major T-cell subsets and NK cells. Although the mechanisms responsible for such decreased leukocyte-subset counts in blood of T/NK-LGLL patients remain unknown, the presence of cytopenias (i.e., neutropenia, anemia, and thrombocytopenia) in LGLL, has been associated with an altered production of hematopoietic cells in the BM-similarly to what occurs in children with primary immunodeficiencies (PID) carrying STAT3 germinal mutations [9,20,35]-and/or an accelerated cell turn-over in the periphery [36,37]. This is supported by the large frequency of BM failure syndromes, such as aplastic anemia, pure red-cell aplasia, hemolytic anemia, paroxysmal nocturnal hemoglobinuria and myelodysplastic syndromes, observed among T-LGLL patients, and to a lesser extend also among CLPD-NK cases [34]. The overall (more pronounced) reduction of blood-cell counts observed for several myeloid and lymphoid populations in STAT3-mutated vs. WT LGLL cases might be explained by the fact that common surface molecules shared by all the reduced hematopoietic cells could trigger cytotoxicity by clonal LGLs. In this sense, markers such as CD45, CD244, and HLA-I are present or expressed with higher intensity on the surface of neutrophils, eosinophils, dendritic cells, nonclassical monocytes, and NK cells [38][39][40][41][42] which could be potentially recognized by clonal LGLs, and trigger LGL-mediated cytotoxic mechanisms involved in the death/elimination of the target cells. Further studies are needed to confirm which of these or other markers shared by the different leukocyte-cell subsets decreased in T/NK-LGLL patients are potentially targeted by cytotoxic clonal T/NK-LGLs, particularly in STAT3-mutated cases, leading to LGLL-associated cytopenias.
Of note, here we demonstrated absence of STAT3 mutations in paired (purified) neutrophils from 7/7 STAT3-mutated LGLL patients. These results suggest that, in contrast with PID cases that carry STAT3 mutations, involvement of hematopoietic progenitor cells by STAT3 mutations is not a frequent finding in LGLL and, thereby, it does not explain the overall decreased production of hematopoietic cells observed in common in both diseases; alternatively, the presence of (over)proliferating STAT3 mutated LGLs in BM might indirectly impact on normal hematopoiesis, because of resource competition. In contrast, this specific group of PID patients classified as phenocopies of inborn errors of immunity [43][44][45] and characterized by germline STAT3 gain-of-function (GOF) mutations, also shows systemic autoimmune diseases associated with lymphoproliferation at the expense of clonal T-LGLs.
Altogether, these findings would support the existence of underlying autoimmunity mechanisms that lead to the abnormally decreased numbers of many different myeloid-and lymphoid-cell populations reported here and in previous studies in LGLL [1,9,10,46], suggesting the potential existence of shared pathogenic mechanisms for the cytopenias observed in both STAT3-mutated PID and LGLL. Thus, most STAT3 (and also STAT5B) mutations detected in both diseases have been identified as GOF mutations that involve the SH2 binding domains of both STAT proteins either at dimerization/Sheinermen residues or at the phosphate-binding (pY) and pY+3 peptide binding pocket of the α-helix, BD loop [47]. If this hypothesis holds true, these mutations would lead to upregulation of Th17 cells in parallel to the inhibition of Tregs; this could translate into an increased production of IL-17 and IL-22 that might contribute to the greater rate of autoimmunity and cytopenias reported here for STAT3-mutated vs. WT LGLL patients. The presence of (much less pronounced) cytopenias also among the latter LGLL patients in our study could be due to the potential presence in a subset of these patients of STAT3 mutations outside the SH2 domain, which were not investigated in our cohort. In this regard, recent studies suggest that LGLL patients carrying STAT3 mutations in the DNA binding domain and the coiled-coil domain of this gene also show a high frequency of neutropenia [25,28,48,49].
In this regard, it should be noted that our patient cohort was not biased by either the clinical condition of the patient or the presence of large expansions of PB LGLs, as all correlative T-LGLL cases referred during the recruitment period were investigated. This might explain the lower frequency of STAT5B-mutated cases vs. previous reports, in which more aggressive LGLL cases and/or with larger LGL expansions in PB have been studied. The presence of the same STAT3 mutation(s) in different subtypes of T-LGLL and CLPD-NK (e.g., Y640F) further supports the occurrence of similarly (or identical) altered activation pathways across the different subtypes of LGLs, regardless of the specific T/ NK-cell lineage involved. Most intriguing is the presence of different STAT3 gene mutations in distinct clonal LGL populations from the same patient, which may suggest the existence of an underlying (e.g., genetic or environmental) predisposition for acquisition of STAT3 mutations. At the same time, it further confirms that STAT3 mutations in LGL do not occur at early stages of hematopoiesis. These results, together with the coexistence of ≥2 distinct LGL clones in up to one third of TCD8 + -LGLL cases, support previous studies suggesting that chronic (viral or autologous) antigen stimulation might be involved in the pathogenesis of oligoclonal/clonal LGL expansions and the sequential selection and expansion of a progressively smaller number of LGL clones [10,[50][51][52][53]. If this hypothesis holds true, survival of the antigen-selected LGL population(s) might then be maintained via upregulation of specific activation pathways, particularly those involving JAK/STAT signaling [53]. In line with this hypothesis, mutations in several genes other than STAT3 and STAT5B involved in this JAK/STAT signaling pathway, such as the IGSF3, JAK3, PTPRT, TTN, and USH2A genes, have also been sporadically reported in LGLL [25,27,28,54]. Independently of its biological significance, the demonstration of somatic STAT3/STAT5B mutations in LGLs (including coexistence of ≥2-small-LGL populations with different STAT3 mutations in the same patient) strongly support their clonal nature. Therefore, the presence of STAT3 mutations would contribute to a more robust demonstration of clonality and diagnosis of T/NK-LGLL, particularly in cases where oligoclonal LGL expansions are observed on phenotypic grounds, and in CLPD-NK patients in whom there are no alternative techniques for confirmation of NK-cell clonality, as shown here. In line with this, it should be noted that during the period of patient recruitment, a total of 10 male patients suspected to have CLPD-NK (by phenotype) were studied; from them, five could not be included in our study, as confirmation of NK-cell clonality could not be done, while the presence of STAT3 mutations was definitive for NK-cell clonality confirmation in four male (out of 10) patients from our cohort. Of note, in our study STAT3 and STAT5B mutations were investigated using conventional sequencing techniques (i.e., Sanger) on highly-purified clonal T-and NK-LGL fractions, such approach allowing accurate identification of mutations even in cell populations which were present at very low frequency in whole PB and/or corresponded to bi(multi)clonal cases. At present, the impact of STAT3 mutations in the clinical behavior and outcome of LGLL remains controversial. Thus, preliminary studies reported an association of STAT3/STAT5B mutations with more aggressive disease [18,26,[30][31][32]. Overall, our results showed a deleterious impact of STAT3 mutations in T-LGLL, as reflected by significantly lower platelet counts and a higher frequency of neutropenia (and severe neutropenia), together with a tendency toward a greater prevalence of autoimmune disease conditions, including cytopenias. Altogether, this translated into a significantly higher frequency of STAT3-mutated T-LGLL cases requiring therapy (due to the associated-autoimmune disorders).
Both T-LGLL and CLPD-NK showed a significantly shorter time-to-therapy in STAT3-mutated vs. nonmutated patients, which was more evident in non-Y640F STAT3-mutated cases (i.e., G618R, N647I, 656ins, D661V, and D661Y) compared to Y640F-mutated ones; these preliminary results confirm and extend recent reports on the potential impact of the type of STAT3 mutation on different clinical manifestations of T/NK-LGLL [33]. Despite the low number of cases per LGLL subtype, such more adverse clinical behavior appeared to be shared by distinct subtypes of LGLL, including TCD8 + -LGLL and Tγδ + -LGLL. Additional multicentric studies in larger patient cohorts with longer follow-up are necessary to definitively confirm the prognostic impact of STAT mutations in other less frequent LGLL subtypes.

Patients and Samples
One-hundred consecutive patients referred to the Cytometry Service of the University of Salamanca (NUCLEUS) between February 2013 and March 2018 were studied-either at diagnosis (n = 79) or during follow-up (n = 21)-to confirm/rule out the diagnosis of LGLL. Presence of increased (absolute and/or relative) numbers of LGL cells [10] and/or phenotypically abnormal LGLs in blood (n = 87) or bone marrow (BM, n = 13) was confirmed in all cases (Table S9) (v) the remaining three cases corresponded to one Tαβ + DP and two Tαβ + DN clonal cases (Table S9). In-depth analysis of all LGL populations was carried out in 59/82 clonal patients (72%) in whom detailed disease features were available. Of these 59 patients, 45 (76%) showed one-single expanded clone, while expansions of ≥2 different clonal LGL populations were detected in the other 14/59 (24%) cases (Table S9). Median follow-up (from diagnosis) for these patients at the moment of closing the study was of 44 months (range: 1-230 months); 16/68 patients (24%) were treated (with immunosuppressive drugs such as cyclophosphamide, methotrexate, cyclosporine A, and corticoids) either before (n = 8) or after (n = 8) they were investigated for the presence of STAT3/STAT5B mutations (Table 3). In parallel, 17 patients with T-CLPD other than LGLL (Table S9) and 628 age-matched healthy donors (HD), were analyzed as controls. In addition, polyclonal (normal residual) LGL populations (n = 26) and myeloid populations (n = 7) from patients with LGL lymphocytosis were also purified and submitted to further mutational analyses (Table S9).
All patients and controls gave their written informed consent to participate in the study, and the study was approved by the Ethics Committee of the University Hospital of Salamanca/IBSAL (Salamanca, Spain).

Immunophenotypic Studies
EDTA-anticoagulated whole blood or BM samples were immunophenotyped using a direct immunofluorescence stain-and-then-lyse technique, based on the EuroFlow T-cell and NK-cell CLPD panels and the EuroFlow standard operating procedures [12,55,56]. Briefly, samples were sequentially stained with the LST tube ("Lymphocyte Screening Tube") for analysis of the distribution of the major leukocyte subsets. Depending on the type of expanded/aberrant cells identified with LST, either the T-cell or the NK-cell CLPD panels were then applied for their further characterization and classification. In T-CLPD, T-cell clonality was assessed prior to staining with the T-CLPD panel using either the IOTest ® Beta Mark TCR-Vβ Repertoire Kit (Beckman-Coulter, Brea, CA, USA) in case of Tαβ + T-CLPD or the anti-TCRVγ9 and anti-TCRVδ2 antibody reagents (Beckman-Coulter) in case of Tγδ + T-CLPD (Table S10). Immediately after sample preparation, stained cells were measured in a FACSCanto-II flow-cytometer (Becton/Dickinson Biosciences, San Jose, CA, USA), using the FACSDiva TM software (Becton/Dickinson Biosciences). Instrument setup, calibration, and daily quality control and monitoring were performed according to well-established EuroFlow protocols [55,56]. For data analysis, the INFINICYT TM software (Cytognos, Salamanca, Spain) was used. The gating strategy employed for the identification of the different T/NK-cell populations, to calculate their relative distribution in blood or BM and to further assess their phenotype, is illustrated in Figures S5 and S6. Absolute cell counts/µL of blood were calculated using a dual platform procedure [57].

Assessment of T-and NK-Cell Clonality on FACS-Sorted Cell Populations
The clonal nature of the expanded LGL populations was assessed in highly-purified FACS-sorted cells (purified from 3-5 mL of whole PB or BM using a FACSAria-III cytometry-Becton/Dickinson Biosciences, to collect at least 10,000 cells/population with a purity ≥95%), based on the presence of single or a few dominant TCRβ and/or TCRγ VDJ gene rearrangements for T-LGL; for CLPD-NK, the polymerase chain reaction (PCR)-based HUMARA assay for analysis of the pattern of inactivation of the human androgen receptor gene coded in chromosome X of heterozygous female patients was used, following well-established protocols and criteria [14,50,[58][59][60]. Those cases in which the sorted LGL population(s) showed one reproducible clonal peak/band by PCR, were considered as "clonal" cases; in turn, those cases in which sorted LGL population(s) with a particular (homogeneous) phenotypic profile showed three or more peaks/bands and a Gaussian curve/smear (with or without minor reproducible peaks/bands), were considered as "oligoclonal" and "polyclonal" cases, respectively [61].
In four CLPD-NK male cases, NK-cell clonality was established on purified cells through confirmation of the presence of STAT3 (somatic) mutations.

Analysis of STAT3 and STAT5B Gene Mutations
STAT3 and STAT5B gene mutations were analyzed on genomic DNA extracted from all but six highly-purified T/NK-cell populations (159/165 clonal T/NK-LGL populations) from 117 subjects, including (Table S9 and Figure S7 (Table S11) [11,16,23]. All (forward and reverse) primer pairs produced a single discrete PCR amplicon of the expected length. Amplified products were purified and then sequenced by conventional Sanger techniques at the Genomic Unit of the Cancer Research Center (IBMCC, USAL-CSIC, Salamanca, Spain). All mutations were screened by bidirectional (forward and reverse) sequencing. Sequencing data was analyzed using the Chromas Lite Sequencing Software 2.1.1 (Technelysium, South Brisbane, Australia) and scored as somatic mutations based on their absence in paired non-LGL populations (i.e., normal residual non-LGL TCD4 + cells).

Statistical Methods
The

Conclusions
Our results show that STAT3/STAT5B mutations might be of diagnostic value in LGLL for demonstration of clonality in T-LGLL and CLPD-NK. In addition, we show for the first time that the presence of STAT3/STAT5B mutations is associated in LGLL with reduced numbers of most normal residual blood-leukocyte subsets, independently of the specific LGLL cell-lineage involved, with important clinical consequences including a shorter time-to-therapy. Further multicentric studies in large cohorts of LGLL patients with long-term follow-up are needed to confirm these observations.
Supplementary Materials: The following are available online at http://www.mdpi.com/2072-6694/12/12/3508/s1, Figure S1: Lollipop diagrams representing the specific STAT3 (A) and STAT5B (B) mutations identified in our T/NK-LGLL cohort and their relative position in the SH2 domain of both genes, Figure S2: Multidimensional phenotypic comparisons between wild-type and STAT3 or STAT5B mutated populations of clonal LGLL cells, Figure S3: Distribution of clonal and normal residual peripheral blood cells in TCD8 + -LGLL patients classified according to their STAT3 mutational status and the presence of biclonal vs. monoclonal LGL, Figure S4: Distribution of distinct maturation-associated cell subsets for the major normal residual T cells in WT vs. STAT3-mutated vs. age-matched HD in TCD8 + -and Tγδ + -LGLL, Figure S5: Phenotypic identification of distinct subtypes of aberrant clonal LGL of different T-cell lineages in T-LGLL patients, Figure S6: Phenotypic identification of distinct subtypes of aberrant clonal NK-cells in CLPD-NK patients, Figure S7: Flow-chart illustrating how the cases and cell populations were selected for the analysis of STAT3 and STAT5B mutations, Table S1: Distribution of STAT3 or STAT5B-mutated T-LGL cell populations according to the T-cell receptor beta chain variable region (TCR-Vβ) or T-cell receptor gamma 9 and delta 2 chain region (TCR-Vγ9 and TCR-Vδ2) expressed, Table S2: Distribution of normal PB leukocyte subsets in T-LGLL according to their STAT3 or STAT5B mutational status, Table S3: Distribution of normal PB leukocyte subsets in TCD8 + -LGLL according to their STAT3 mutational status, Table S4: Distribution of normal PB leukocyte subsets in Tγδ + -LGLL according to their STAT3 mutational status, Table  S5: Distribution of normal PB leukocyte subsets in CLPD-NK according to their STAT3 mutational status, Table  S6: Clinical and biological features of clonal T-LGLL cases with wild-type (WT) vs. mutated STAT3/5B, Table S7: Clinical and biological features of TCD8 + -LGLL according to their STAT3 mutational status and the presence of bi(multi) vs. monoclonal LGL populations, Table S8: Clinical and biological features of clonal CLPD-NK cases with wild-type (WT) vs. mutated STAT3, Table S9: Distribution of LGLL cases and non-LGL CLPD patients (n = 117) included in this study and the corresponding populations of LGL identified (n = 165) classified according to their phenotypic profile and (mono vs. oligo/poly)clonal nature, Table S10: Panels of fluorochrome-conjugated antibodies (specificities and sources) used for sequential immunophenotypic analyses of LGLL by flow cytometry, Table S11: STAT3 and STAT5B primers and temperature conditions used for PCR-based gene mutation analysis.