|Home | About | Journals | Submit | Contact Us | Français|
T-cell large granular lymphocytic leukemia is a rare lymphoproliferative disorder characterized by the expansion of clonal CD3+CD8+ cytotoxic T lymphocytes (CTLs) and often associated with autoimmune disorders and immune-mediated cytopenias.
We used next-generation exome sequencing to identify somatic mutations in CTLs from an index patient with large granular lymphocytic leukemia. Targeted resequencing was performed in a well-characterized cohort of 76 patients with this disorder, characterized by clonal T-cell–receptor rearrangements and increased numbers of large granular lymphocytes.
Mutations in the signal transducer and activator of transcription 3 gene (STAT3) were found in 31 of 77 patients (40%) with large granular lymphocytic leukemia. Among these 31 patients, recurrent mutational hot spots included Y640F in 13 (17%), D661V in 7 (9%), D661Y in 7 (9%), and N647I in 3 (4%). All mutations were located in exon 21, encoding the Src homology 2 (SH2) domain, which mediates the dimerization and activation of STAT protein. The amino acid changes resulted in a more hydrophobic protein surface and were associated with phosphorylation of STAT3 and its localization in the nucleus. In vitro functional studies showed that the Y640F and D661V mutations increased the transcriptional activity of STAT3. In the affected patients, downstream target genes of the STAT3 pathway (IFNGR2, BCL2L1, and JAK2) were up-regulated. Patients with STAT3 mutations presented more often with neutropenia and rheumatoid arthritis than did patients without these mutations.
The SH2 dimerization and activation domain of STAT3 is frequently mutated in patients with large granular lymphocytic leukemia; these findings suggest that aberrant STAT3 signaling underlies the pathogenesis of this disease. (Funded by the Academy of Finland and others.)
T-cell large granular lymphocytic leukemia was initially described as a clonal disorder of large granular lymphocytes involving blood, bone marrow, spleen, and liver.1 This disorder is characterized by the presence of abnormal CD3+CD8+CD57+ lymphocytes corresponding to activated effector cytotoxic T lymphocytes (CTLs).2,3 Large granular lymphocytic leukemia is frequently accompanied by autoimmune processes such as rheumatoid arthritis (often manifested as Felty’s syndrome) and immune-mediated cytopenias.4 Many cases are indolent, and distinguishing large granular lymphocytic leukemia from reactive processes involving large granular lymphocytosis may be difficult, since both conditions can be associated with a skewed CTL antigen-receptor repertoire (i.e., oligoclonal expansion rather than the normal polyclonal distribution) and can be linked to viral infections and autoimmune conditions.5–8 Furthermore, biclonal subtypes of large granular lymphocytic leukemia have been described, as well as clonal switching — both of which are suggestive of a pervasive antigenic drive.9 Thus, large granular lymphocytic leukemia may represent an extreme pole in the spectrum of CTL responses to antigen.
No recurrent chromosomal aberrations or mutations have been found to be associated with large granular lymphocytic leukemia. On the intracellular level, gene-expression studies have suggested dysregulation of apoptotic pathways (e.g., the FAS–FAS ligand and sphingolipid pathways) and activation of survival signaling pathways (e.g., the RAS and phosphatidylinositol 3-kinase [PI3K]–AKT pathways).10–14
Massively parallel second-generation sequencing technology has been used successfully to uncover the genetic background of some malignant hematologic disorders, such as acute myeloid leukemia,15,16 hairy-cell leukemia,17 the myelodysplastic syndrome,18 and chronic lymphocytic leukemia.19 The underlying genetic defects in large granular lymphocytic leukemia have not been elucidated. In hopes of illuminating the molecular pathogenesis of this disorder and providing insights into the aberrant regulation of T-cell activation in relation to the associated autoimmune disorders, we performed whole-exome and RNA sequencing of leukemic CTLs and matched healthy control cells obtained from a patient with large granular lymphocytic leukemia to identify somatic genetic changes.
The study population consisted of 77 patients with large granular lymphocytic leukemia. All patients had had CD3+CD8+ large granular lymphocytes in the peripheral blood for more than 6 months, and at the time of diagnosis, the clonality of T cells was confirmed by a polymerase-chain-reaction (PCR) assay for T-cell–receptor rearrangement. (The main characteristics of the patients are summarized in Tables S1 and S2 in the Supplementary Appendix, available with the full text of this article at NEJM.org.)
A commercial platform (Agilent) was used to prepare sample libraries of genomic DNA from the leukemic (CD8+) cells and nonleukemic (CD4+) cells obtained from the index patient (Patient 1 in Table S1 in the Supplementary Appendix), followed by in-solution exome capture. Massively parallel sequencing was performed with the Genome Analyzer II (Illumina). Candidate somatic mutations were identified according to a bioinformatic protocol (Fig. S1 in the Supplementary Appendix). Reads from RNA sequencing were aligned and analyzed to detect fusion genes. Gene expression was evaluated visually from RNA-sequence read alignments with the Integrative Genomics Viewer (Broad). (Whole-exome sequencing, bioinformatics, and other methods are fully described in the Supplementary Appendix.)
After exome sequencing, the candidate mutations were validated with the use of Sanger sequencing. For the STAT3 mutation screening, five primer pairs were designed to cover the six exons encoding the Src homology 2 (SH2) domain of STAT3 (exons 19 to 24). (The primers are described in Table S3 in the Supplementary Appendix.)
All statistical analyses are described in detail in the Supplementary Appendix.
The index patient was a 70-year-old man with untreated typical T-cell large granular lymphocytic leukemia. Figure 1A shows the results on flow cytometry. Whole-exome sequencing of purified CD8+ leukemic cells and CD4+ control cells from the index patient produced approximately 61.1 million and 60.2 million reads, respectively, that could be mapped to the reference genome (Fig. S1 in the Supplementary Appendix). The pairedend read length was 82 nucleotides. After somatic mutation calling and validation by capillary sequencing, the genes for signal transducer and activator of transcription 3 (STAT3), macrophage scavenger receptor 1 (MSR1), and homeobox C9 (HOXC9) were found to harbor somatic heterozygous mutations specific to the CD8+ cell population. The mutations in STAT3 and MSR1 introduced an amino acid substitution in the encoded proteins, whereas the mutation in HOXC9 was a deletion of an amino acid (Table S4 in the Supplementary Appendix). (The STAT3 mutation is shown in Fig. 1B, and in Fig. S2 in the Supplementary Appendix.) No fusion genes were identified by RNA sequencing.
The STAT3 mutation in chromosome 17 in the index patient was located in exon 21, encoding a portion of the STAT3 SH2 domain (Fig. 2A and 2B). A heterozygous missense mutation (T→A) resulted in a D661V variant protein. In the germline CD4+ control sample, only normal reads (63) were detected, whereas in the CD8+ tumor sample, 29 mutant reads and 39 germline reads were detected, resulting in a mutant allele frequency of 43% (Fig. S2 in the Supplementary Appendix). Also, sequencing of RNA derived from the CD8+ leukemic cells identified the D661V mutation as a prominent clone (Fig. S2A in the Supplementary Appendix), whereas it was absent from the CD8+ RNA from healthy control cells.
STAT3 is an oncogene, and its activation plays a key role in cell signaling in many types of cancer.20 Since the D661V mutation was located in the SH2 domain (Fig. 2), which mediates dimerization and activation of the STAT3 protein, we next designed primers for the secondary screening of the six exons encoding the STAT3 SH2 region in the remaining patients. Five other recurrent somatic missense mutations and one inframe insertion were identified in 30 additional patients (Fig. 2, and Fig. S3 and Tables S1 and S5 in the Supplementary Appendix), with a STAT3 mutation frequency of 40% (31 of all 77 patients in the cohort). The most common mutation hot spot was located at amino acid D661, resulting in three different variant proteins: D661V (mutation frequency, 9% [7 patients]), D661Y (mutation frequency, 9% [7 patients]), and D661H (mutation frequency, 1% [1 patient]). Y640F mutations were detected in 17% of patients (13 of 77), N647I mutations in 4% (3 of 77), and a K658N mutation in 1 patient, as well as an in-frame insertion mutation (Y657_K658insY) in an additional patient at the same site.
In the other 46 patients with large granular lymphocytic leukemia who had more than 20% clonal cells in the DNA sample (on the basis of CD8+ T-cell counts and Vβ expression), no mutations were detected in the STAT3 SH2 domain (Table S2 in the Supplementary Appendix). Also, no HOXC9 or MSR1 mutations were found in any other patient in the study cohort. In addition, two T-cell acute lymphoblastic leukemia cell lines (Jurkat and MOLT-4) and samples from 9 patients with acute myeloid leukemia and from 6 patients with acute lymphocytic leukemia were screened, and no STAT3 mutations were found. Similarly, CD8+ cells from 6 patients with reactive large granular lymphocytosis during tyrosine kinase inhibitor therapy,8 as well as from 2 patients with cytomegalovirus infection and from 1 patient with graft-versus-host disease, were studied and found to be negative for STAT3 mutations. Moreover, none of the identified mutations were present in two population-variation data sets: 1000 Genomes (www.1000genomes.org/data) and dbSNP (www.ncbi.nlm.nih.gov/projects/SNP).
Further sequencing analysis of sorted B-cell, T-cell, and natural killer (NK)–cell fractions was performed for Patients 1, 19, and 30 in the group of patients with STAT3 mutations. STAT3 mutations were found only in the T-cell (CD8+) fraction, whereas NK-cell and B-cell fractions showed wildtype STAT3 (Fig. S4 in the Supplementary Appendix).
All identified STAT3 mutations were located only a few codons apart in exon 21 (Fig. 2A). Furthermore, they all induced an amino acid change that made the protein surface more hydrophobic. The protein structure shows that they are positioned very close to each other in the dimerization interface (Fig. 2B). Only one of the six STAT3 mutations (Y640F) has been described previously, in inflammatory hepatocellular adenoma.21 This mutation alters a conserved tyrosine residue, leading to cytokine-independent tyrosine phosphorylation and STAT activation, as well as to cytokine-dependent hyperactivation.21,22
A comparison of the clinical characteristics of patients with large granular lymphocytic leukemia showed that neutropenia was more common among patients with a STAT3 mutation (24 of 31 patients [77%]) than among those without a STAT3 mutation (23 of 46 patients [50%]) (P=0.02) (Tables S1 and S2, respectively, in the Supplementary Appendix). Similarly, patients with STAT3 mutations had rheumatoid arthritis more often than did those without STAT3 mutations (26% vs. 6%, P=0.02). No differences were found between the groups with and those without STAT3 mutations in terms of age, sex, hemoglobin level, and leukocyte or lymphocyte values, and the frequency of signs of B-cell dyscrasia, anemia, or thrombocytopenia was similar in the two groups. Basic laboratory values and the presence or absence of related autoimmune disorders did not differ significantly according to the specific STAT3 mutation (D661V, D661Y, or Y640F), but the number of patients with each mutation was small (Fig. S5 in the Supplementary Appendix).
In one patient with the Y640F mutation (Patient 31), SNP-A karyotyping identified unisomal parental disomy in 17q, where STAT3 is located, resulting in a homozygous mutation (Fig. S6 in the Supplementary Appendix). This patient had both anemia and neutropenia and had been treated with cyclosporine and alemtuzumab.
Bone marrow–biopsy samples were obtained from one healthy control and from three patients with large granular lymphocytic leukemia, one with the D661V mutation and two with the Y640F mutation (Fig. 3). Immunohistochemical staining showed that the lymphocyte infiltration in the patients with leukemia expressed CD57, which is typical of large granular lymphocytes (Fig. 3B, 3C, and 3D). Unlike the healthy control (Fig. 3A), the patients with leukemia had positive staining of the nuclei of infiltrated lymphocytes, with phosphorylated STAT3 antibody indicating STAT3 activation. These findings were confirmed by Western blot analysis of nuclear extract samples from two of the patients (one with the Y640F mutation and one with the D661V mutation) and from the healthy control (Fig. 3E).
With the use of the HumanHT-12 v4 Expression BeadChip array, RNA expression analysis was performed on CD8+ sorted cells from 8 patients with STAT3 mutations (D661H, D661Y, D661V, and Y640F) and from 10 patients without STAT3 mutations. CD8+ cells from 5 healthy volunteers were obtained for comparison. INFGR2, IRF7, IRF9, PSMB9, BCL2L1, STAT1, STAT2, and STAT3 were significantly up-regulated and SOCS1 was down-regulated in the samples from the patients with and those without STAT3 mutations (Fig. 4A). The expression of JAK2 was significantly higher only in the patients with a STAT3 mutation, whereas expression of interleukin-6, one of the main up-stream regulators of STAT3, was not significantly changed in these patients, although the values ranged widely. No significant differences were found in BCL2 or MCL1 expression in between-group comparisons (Fig. 4A). Analysis of the RNA sequencing data from the index patient with the D661V mutation showed similar up-regulation of STAT3 target genes, including the FAS ligand (Fig. S7 in the Supplementary Appendix).
To explore the functional implications of the Y640F and D661V mutations, we generated expression constructs for wild-type, Y640F, and D661V variants of the STAT3 protein. When expressed in HEK293 human embryonic kidney cells carrying a luciferase reporter gene under the control of a STAT3-responsive sis-inducible element (SIE), the Y640F mutant exhibited strikingly increased activity under both basal (unstimulated) and interleukin-6–stimulated conditions (Fig. 4B). The D661V mutant caused more modest (but still significant) basal and interleukin-6–stimulated transcriptional activation (Fig. 4B).
By means of whole-exome and targeted sequencing, we found that 40% of patients with large granular lymphocytic leukemia had activating mutations in STAT3. The STAT3 missense mutations (D661V, D661Y, D661H, Y640F, N647I, and K658N), as well as the insertion mutation (Y657_K658insY), were located in the SH2 domain on the dimerization interface that mediates STAT3 activation.23 Structural modeling indicated that all the observed mutations increased the hydrophobicity of the STAT3 SH2 dimerization surface. The mutational hot spot in large granular lymphocytic leukemia resides close to the transcriptional activation domain, and constitutive activation may result from increased stabilization of STAT3 homodimers or heterodimers through enhanced hydrophobic attraction between STAT monomers.23 Leukemic large granular lymphocytes have been shown to express high levels of activated STAT3,24 supporting the functional role of the STAT3 SH2 domain mutations. In accordance with this finding, an in vitro assay showed up-regulation of STAT3 target gene expression in patients with large granular lymphocytic leukemia who had the mutation and increased transcriptional activity of the mutated STAT3 isoforms. Furthermore, STAT3 was phosphorylated in the patients with the mutation.
Studies done in T-cell–specific Stat3-deficient mice show that Stat3 activation is responsible for interleukin-6–dependent T-cell proliferation by preventing apoptosis independently of Bcl-2.25 Furthermore, in a liver-specific mouse model, constitutively active Stat3 was shown to provide protection against Fas-mediated liver injury, whereas liver-specific Stat3-knockout mice failed to survive a typically nonlethal dose of FAS agonist.26 Similarly, leukemic large granular lymphocytes appear to be resistant to FAS-induced apoptosis, despite high levels of FAS and FAS ligand expression27; when STAT3 expression is reduced with an antisense oligonucleotide approach, large granular lymphocytes appear to regain their FAS sensitivity.24,28 Also, other antiapoptotic and survival-promoting signaling pathways, including some regulated by STAT3, have been shown to be upregulated in leukemic large granular lymphocytes.10,13,24 These findings support the hypothesis that constitutive STAT3 activation in large granular lymphocytic leukemia cells enhances antiapoptotic pathways. New, specific STAT3 inhibitors (e.g., OPB-3112)29 have recently been evaluated in early clinical trials for the treatment of other hematologic cancers, and our findings suggest that preclinical and clinical studies of these drugs are also warranted in patients with large granular lymphocytic leukemia.
The diagnosis of large granular lymphocytic leukemia can be difficult to establish, since benign large granular lymphocytosis with clonal T cells may occur in conjunction with viral infections (e.g., cytomegalovirus infection) or autoimmune disorders. Somatic STAT3 mutations may provide a diagnostic tool for classifying some cases of large granular lymphocytosis as true T-cell leukemias that are distinct from clonally skewed reactive processes and from hyperreactive or persistent T-cell responses. We included in our study only patients with one major Vβ clone in order to ensure that the study cohort would be clinically and genetically uniform. More sensitive techniques (e.g., Vβ clone sorting, mutation-specific real-time quantitative PCR assay) would allow for the detection of mutations in patients with smaller or multiple clones as well; thus, the true mutational frequency in large granular lymphocytic leukemia may be higher or lower. Since expression analysis showed that patients with and those without STAT3 mutations had many common overexpressed STAT3 target genes, further studies are warranted to determine whether STAT3 mutation–negative patients have mutations in other STAT-related pathways.
Patients with STAT3 mutations had neutropenia and rheumatoid disorders more often than did patients with wild-type STAT3. Large granular lymphocytes can secrete several cytokines that may play a role in immune-mediated cytopenias and autoimmune disorders.30,31 The associated clinical findings may also be due to antihost activity of expanded CTLs, since it has been shown that leukemic large granular lymphocytes can directly lyse pulmonary-artery endothelial cells and human synovial cells.32 Constitutive activation of STAT3 has been observed in patients with autoimmune disorders such as rheumatoid arthritis and Crohn’s disease.33–35 The strong link between large granular lymphocytic leukemia and these disorders warrants screening for STAT3 mutations in T lymphocytes from patients who have autoimmune disease in the absence of overt large granular lymphocytic leukemia.
In conclusion, STAT3 is frequently mutated in large granular lymphocytic leukemia, highlighting key steps in the molecular pathogenesis of this rare disease. Structural and functional data suggest that the observed mutations result in STAT3 hyperactivation and may confer ligand-independent signaling.
Supported by a special subsidy for health sciences, research, and training provided by the Finnish government and by grants from the Academy of Finland, the Finnish Cancer Societies, the Sigrid Juselius Foundation, the Finnish Association of Hematology, the National Clinical Graduate School, the Finnish Funding Agency for Technology and Innovation, the European Regional Development Fund, the Jane and Aatos Erkko Foundation, the K.A. Johansson Foundation, and the Signe and Ane Gyllenberg Foundation, and by grants from the National Institutes of Health (R01 CA127264A, to Mr. Clemente; R01AI085578, to Ms. Guinta; CA98472, to Dr. Loughran; and 2K24HL077522, to Dr. Maciejewski).
We thank the personnel at the Hematology Research Unit Helsinki and Institute for Molecular Medicine Finland for their expert clinical and technical assistance.
Disclosure forms provided by the authors are available with the full text of this article at NEJM.org.