Receptor-type tyrosine-protein phosphatase κ directly targets STAT3 activation for tumor suppression in nasal NK/T-cell lymphoma

Extranodal nasal-type natural killer/T-cell lymphoma (NKTCL) primarily develops in the nasal cavity but can also occur in extranasal sites, either as a primary extranasal or disseminated disease.^1,2  NKTCL histology is characterized by the angiocentric and angioinvasive infiltration of lymphoma cells and prominent necrosis.¹  NKTCLs can be widely disseminated in the leukemic phase^1,2  and are invariably associated with Epstein-Barr virus (EBV) infection in lymphoma cells.¹  Epidemiologically, NKTCL is prevalent in Asia and South America but is rare in the United States and Europe.¹ 

NKTCL is an aggressive disease¹  characterized by frequent deletions on chromosome 6q.^3-14  Through the use of array-based comparative genomic hybridization and gene expression profiling, oncogenic pathways operating in NKTCL have been delineated by various groups,^14-19  and the PRDM1, ATG5, AIM1, HACE1, and FOXO3 genes have been identified as tumor-suppressor gene candidates in the most frequently deleted region in NKTCL, 6q21, by functional analyses.^14-17 

Recent studies have demonstrated that constitutive activation of signal transducer and activator of transcription 3 (STAT3), a known pro-oncogenic transcription factor, is common in NKTCL and is an early feature in NKTCL pathogenesis.^15,20  STAT3 activation is accompanied by tyrosine phosphorylation at Tyr705, which induces dimerization and nuclear translocation, where STAT3 molecules activate the transcription of a wide array of genes that play critical roles in development, immunity, and proliferation.²¹  Recently, it was reported that STAT3 activation in NKTCL most often results from constitutive janus kinase 3 (JAK3) phosphorylation at Tyr980, and in a proportion of NKTCL primary tumors, constitutive JAK3 activation is associated with acquired mutations in the JAK3 pseudokinase domain.^22,23  Protein phosphorylation is a reversible process, and the degree of phosphorylation of a target protein reflects a net balance between the competing actions of protein kinases and phosphatases. Because recent large-scale genetic analyses of various human tumors have highlighted the relevance of protein tyrosine phosphatases (PTPases) as putative tumor suppressors^24,25 , in this study, we explored whether the loss of expression of the receptor-type tyrosine-protein phosphatase κ (PTPRK), the only PTPase among the 3 PTPases located in the commonly deleted 6q region (see supplemental Table 1 on the Blood Web site) that contains a STAT3-specifying YXXQ motif²⁶  at the first catalytic domain, leads to STAT3 activation and plays a role in NKTCL oncogenesis. The PTPRK gene, located at 6q22.2-q22.3,²⁷  belongs to the receptor PTPase (R-PTPase)-type R2B subfamily and is characterized by an extracellular adhesion molecule-like domain, a single transmembrane helix, and a cytoplasmic tyrosine phosphatase domain.^28-32 

In this study, we provide a novel molecular mechanism leading to STAT3 deregulation in NKTCL. We show that PTPRK binds to STAT3 and directly dephosphorylates phospho-STAT3 at Tyr705 and regulates the levels of phospho-STAT3^Tyr705 in NKTCL. PTPRK, located in the commonly deleted 6q region, is frequently underexpressed in NKTCL due to monoallelic loss and promoter hypermethylation, and PTPRK loss leads to STAT3 activation and contributes to NKTCL pathogenesis.

Archival paraffin and frozen tumor blocks from 27 Chinese patients diagnosed with NKTCL from 1996 to 2010 were retrieved for this study from the Department of Pathology at Queen Mary Hospital, Hong Kong. Twenty-two of the primary NKTCL cases included in this study have been previously described.^33-35  Clinical features were available for all 27 NKTCL patients. Treatment outcome data were available for the 17 NKTCL patients treated with the steroid-dexamethasone, methotrexate, ifosfamide, l-asparaginase, and etoposide (SMILE) regimen,^33-35  and the remaining 10 patients were treated with other multiagent chemotherapies, with or without radiotherapy. The SMILE protocol for the treatment of NKTCL patients was approved by the University of Hong Kong/Hospital Authority Hong Kong West Cluster Institutional Review Board.^33-35  Research was conducted in accordance with the Declaration of Helsinki.

The Moloney murine leukemia virus-based retroviral vector pBMN-GFP (Orbigen Inc., San Diego, CA), containing a downstream internal ribosomal entry site for green fluorescent protein (GFP) expression, and a RetroNectin-coated plate system were used for retroviral gene transduction. Full-length PTPRK complementary DNA (cDNA) (4341 bp) with a Myc-DDK tag sequence at the carboxyl-terminus was subcloned into the retroviral vector (DDK sequence: DYKDDDDK) (supplemental Figure 1Ai). NKYS and KHYG cells were successfully transduced with a retroviral vector containing PTPRK cDNA with transduction efficiencies of ∼12% and 7%, respectively (supplemental Figure 1Aii). The in vitro functional assays were performed on GFP-sorted cells following validation of PTPRK expression in PTPRK cDNA transduced NKYS cells (supplemental Figure 2).

PTPRK knockdown was accomplished using a CMV-puro (pLKO.1puro-CMV-TagRFP) lentiviral plasmid (Sigma-Aldrich, St. Louis, MO) for PTPRK short hairpin RNA (shRNA) transduction in SNK6 cells (supplemental Figure 1Bi), and the transduction efficiency was ∼20% (supplemental Figure 1Bii). In vitro functional assays were performed on phycoerythrin-sorted SNK6 cells, following validation that PTPRK expression was partially knocked down to ∼ 30% of the original level in sorted SNK6 cells using 2 independent shRNAs (supplemental Figure 3).

The methylation status of the PTPRK promoter was analyzed via methylation-specific polymerase chain reaction (PCR) (or MSP) and further confirmed with bisulfite genomic sequencing (BGS). The cytosine guanine dinucleotide (CpG) sites in the PTPRK promoter (−300 bp to +300 bp) and the position of the MSP and BGS primers (supplemental Table 2) used in this study are presented in supplemental Figure 4.

Detailed descriptions of the various methods employed in this study are provided in the supplemental “Materials and methods.”

Semiquantitative reverse transcription-polymerase chain reaction (RT-PCR) analysis showed that PTPRK mRNA was not expressed in 4 of 5 (80%) NKTCL cell lines but was highly expressed in 2 preparations of normal NK cells and an NKTCL cell line (SNK6) (Figure 1A). Significantly lower levels of PTPRK mRNA expression were detected in 15 of 27 (55.6%) NKTCL primary tumors, which is less than ∼ 30% of the levels in normal NK cells (Figure 1B). Immunohistochemical staining showed that PTPRK protein expression in tumor cells (Figure 2A) correlated with its mRNA expression in the 27 NKTCL primary tumors (P < .034; supplemental Table 3).

The immunohistochemical staining of phospho-STAT3^Tyr705 (brown nuclear staining) and PTPRK (brown cellular membranous staining between the cell-cell contact and faint cytoplasmic staining) in 5 NKTCL cell lines and 27 tumors showed high levels of phospho-STAT3^Tyr705 in the 4 PTPRK nonexpressing cell lines, whereas a very low level of phospho-STAT3^Tyr705 was detected in PTPRK-expressing SNK6 cells (supplemental Table 4). There was also a reciprocal correlation between nuclear phospho-STAT3^Tyr705 and PTPRK levels in tumor cells of the 27 NKTCL primary tumors examined (P < .040) (Figure 2A; supplemental Table 4).

Western blot and flow cytometry analysis using an anti–phospho-STAT3^Tyr705 antibody showed decreased phospho-STAT3^Tyr705 levels in PTPRK-transduced NKYS cells (Figure 2Bi-Ci), whereas partial knockdown of PTPRK (∼70%) significantly increased the levels of phospho-STAT3^Tyr705 in SNK6 cells transduced with 2 independent PTPRK-specific shRNAs (Figure 2Bii-Cii). Each of the shRNAs (#1 and #2) independently provided efficient knockdown of PTPRK expression.

Indirect immunofluorescence experiments demonstrated lower levels of nuclear phospho-STAT3^Tyr705 in NKYS cells with reconstituted PTPRK expression compared with the surrounding nontransduced cells (Figure 2Di), whereas SNK6 cells transduced with PTPRK-specific shRNA #2 showed substantially higher nuclear phospho-STAT3^Tyr705 levels than the surrounding nontransduced cells (Figure 2Dii).

To determine whether phospho-STAT3^Tyr705 dephosphorylation was directly mediated by PTPRK, a substrate-trapping assay was performed. Three glutathione S-transferase (GST) fusion proteins (wild-type [WT], mutant 1, and mutant 2) containing the first phosphatase domain of PTPRK were constructed. The mutant 1 and mutant 2 proteins contained mutations predicted to result in substrate trapping (D1057A in mutant 1 and C1089A in mutant 2). The fusion proteins were expressed in Escherichia coli, and equal amounts of each protein were attached to beads (Figure 3A). The beads were then incubated with cell lysates from NKYS and KHYG cells. The phosphorylated form of STAT3 was specifically bound by the 2 substrate-trapping mutants (the C1089A mutant coprecipitated more STAT3 and phospho-STAT3^Tyr705 than did the D1057A mutant), and did not bind to the WT or control GST proteins (Figure 3B).

Both the phosphorylated and nonphosphorylated forms of STAT3 were immunoprecipitated from HEK293T cells and incubated as phosphatase substrates with equal amounts of WT or mutant 2 PTPRK protein, or with GST alone, with or without the phosphatase inhibitor Na₃VO₄. STAT3 was dephosphorylated by WT phosphatase but not the phosphatase-dead mutant 2 or GST alone (Figure 3C). Phosphatase activity was also inhibited by Na₃VO₄, which indicated that this process was phosphatase dependent (Figure 3C). These results indicate that phospho-STAT3^Tyr705 is a direct substrate of PTPRK.

GFP-sorted PTPRK-transduced NKYS and KHYG cells were grown in culture medium and the restoration of PTPRK expression significantly suppressed the cellular proliferation rate (2-way analysis of variance; P < .05 for both cell lines) (Figure 4A). Moreover, when PTPRK-transduced cells were grown in methylcellulose, NKTCL cells re-expressing PTPRK formed smaller colonies and fewer colonies containing >20 cells compared with NKTCL cells transduced with the mock vector (χ², P < .001 for NKYS, P < .005 for KHYG) (Figure 4B). These results suggested that the clonogenicity of the NKTCL cells was significantly decreased after the restoration of PTPRK expression, and these data were consistent with the tumor-suppressive properties of PTPRK. Moreover, PTPRK re-expression significantly increased the population of cells undergoing late apoptosis, as shown by Annexin-V and 7-amino-actinomycin D (7-AAD) staining, with an average increase of 20% to 40% compared with cells transduced with the mock vector (χ², P < .001) (Figure 4C). Annexin-V staining precedes the loss of membrane integrity that later accompanies cell death, while 7-AAD can be used as a viability probe for nonviable cell exclusion methods.

We next investigated the effects of PTPRK expression on cell death and cell-cycle progression. The restoration of PTPRK expression increased the proportion of GFP-sorted PTPRK-transduced NKYS cells blocked in the G₀/G₁ phase (Figure 4D). Moreover, western blot analysis revealed cleavage of several members of the caspase pathway (caspase-3, caspase-9, and poly[ADP-ribose] polymerase) in PTPRK-positive NKYS cells (Figure 4E). These findings suggested that the restoration of PTPRK expression triggered caspase-mediated apoptosis in NKYS cells.

Partial PTPRK knockdown (∼70%) led to a significant increase in the proliferation rate of RFP-sorted, lentiviral shRNA-transduced SNK6 cells (2-way analysis of variance, P < .001 for both shRNA sequences effective against PTPRK) (Figure 5A) and the in vitro clonogenicity of SNK6 cells (χ², P < .001 for both shRNA sequences) (Figure 5B).

After GFP sorting, PTPRK-transduced and mock vector-transduced NKYS cells were pelleted and resuspended in fresh culture medium; PTPRK-transduced NKYS cells rapidly formed clumps of 5 to 10 cells, whereas NKYS cells transduced with the mock vector remained primarily in a single-cell suspension (Figure 6Ai) after 6 hours. These results suggested that PTPRK restoration enhanced cell-cell contact in NKYS cells.

PTPRK knockdown was accomplished by lentiviral shRNA in SNK6 cells, and following RFP-sorting, lentiviral shRNA-transduced SNK6 cells were centrifuged and resuspended in fresh culture medium; the cells transduced with negative control shRNA formed clumps of >20 cells, whereas SNK6 cells transduced with shRNA targeting PTPRK remained primarily in a single-cell suspension (Figure 6Aii) after 6 hours. These results indicated a loss of homophilic interactions triggered by the receptor PTPRK after PTPRK knockdown in SNK6 cells.

Matrigel cell migration and invasion assays demonstrated that PTPRK restoration significantly reduced the rate of migration and invasion of GFP-sorted PTPRK-transduced NKYS cells (χ², P < .001 for migration, P < .005 for invasion) (Figure 6Bii), whereas the partial knockdown of PTPRK in RFP-sorted, lentiviral shRNA-transduced SNK6 cells increased the migration and invasion rate (χ², P < .001 for both the migration and invasion assays) (Figure 6Bii).

To explore whether PTPRK is epigenetically silenced in NKTCL, we first treated 5 NKTCL cell lines with pharmacologic concentrations of the DNA demethylating agent 5-aza-dC for 3 or 6 days. Semiquantitative RT-PCR analysis revealed that PTPRK mRNA expression was restored in all 4 non–PTPRK-expressing cell lines following 5-aza-dC treatment (Figure 7A). Consistent with the RT-PCR results, immunofluorescence of a non–PTPRK-expressing NKYS cell line also revealed the re-expression of PTPRK at the protein level, following 5-aza-dC treatment (Figure 7B). This 5-aza-dC–induced re-expression indicated that PTPRK is downregulated by promoter hypermethylation in NKTCL.

The PTPRK promoter was unmethylated in normal NK cells and the PTPRK-expressing SNK6 cell line, whereas the PTPRK promoter was methylated in 4 non–PTPRK-expressing NKTCL cell lines (Figure 7C, top panel). In contrast to the normal NK cells and the PTPRK-expressing SNK6 cell line, the 4 non–PTPRK-expressing cell lines exhibited almost complete methylation of the entire CpG island region of PTPRK (Figure 7C, bottom panel). The CpG sites containing Sp1 (2 CpG sites; #8 and #9), heat shock factor (1 CpG site; #10), and CREB (3 CpG sites; #32, #33, and #34) were nearly completely methylated in the 4 cell lines, whereas these same sites were unmethylated or only partially methylated in normal samples and the SNK6 cell line. Moreover, reduced levels or complete loss of PTPRK mRNA expression significantly correlated with PTPRK promoter hypermethylation in the NKTCL cell lines (P = .025) (supplemental Table 5). Furthermore, treatment of the NKTCL cell lines with 5-aza-dC caused demethylation of the CpG sites in the PTPRK promoter and activated PTPRK mRNA expression (Figure 7C, bottom panel).

Next, MSP and BGS assays were performed on 27 NKTCL primary tumors and a methylated PTPRK promoter was detected in 16 of 27 tumors (59%) (Figure 7D; supplemental Table 6A). An unmethylated band was present in nearly all the tumors and was most likely due to the presence of tumor-infiltrating nontumor cells. Reduced levels or complete loss of PTPRK mRNA expression was also significantly correlated with PTPRK promoter hypermethylation in NKTCL primary tumors (P = .001) (supplemental Table 6B).

PTPRK allelic deletion analysis was performed on genomic DNA from NKTCL primary tumors by semiquantitative multiplex PCR using PTPRK and β-globin (as the normal allele control) gene-specific primers (Figure 7E). Overall, monoallelic PTPRK gene deletions were detected in 8 of 27 (30%) tumors, which included 2 of 11 (18%) tumors with unmethylated PTPRK promoters and 6 of 16 (38%) tumors with methylated PTPRK promoters (supplemental Table 6A). Monoallelic PTPRK gene deletions were also detected in 3 of 5 NKTCL cell lines studied, which is consistent with the deletion of 6q23.2 as previously reported in these cell lines.¹⁷ 

The mutational analysis of PTPRK via bidirectional DNA sequencing of the entire coding sequence of PTPRK transcripts only detected 1 silent mutation (coding sequence 583, GTT>GTC) in the SNK6 cell line. No inactivating deletions, insertions, or mutations were detected in the PTPRK mRNA obtained from NKTCL primary tumors.

A retrospective correlation of tumor cell PTPRK protein levels with the clinical features of 27 NKTCL patients showed that low PTPRK levels in tumor cells were significantly correlated with a higher number of extranodal sites (1 vs >1, Pearson’s χ²-test, P = .013), (supplemental Table 7) and were associated with a high international prognostic index (IPI) (0-1 vs 2, 3, or 4-5, Pearson’s χ²-test, P = .040) (supplemental Table 7).

Next, a retrospective correlation of the methylation status of tumor cell PTPRK promoter DNA was performed against the clinical features of the 27 NKTCL patients (Table 1). The univariate statistical correlation revealed that PTPRK promoter hypermethylation in tumors was highly significantly correlated with advanced-stage disease (P = .006) and with greater extranodal involvement in patients (P = .006). Moreover, patients harboring tumors with methylated PTPRK promoters exhibited a significantly higher IPI (P < .001).

Significantly, in 17 NKTCL patients treated with the SMILE protocol, an univariate Kaplan-Meier analysis revealed that patients with a tumor containing methylated PTPRK promoter exhibited a trend toward inferior disease-free survival (DFS) (log-rank P = .052, Breslow P = .024, Tarone-Ware P = .034; Figure 7F) and a significant correlation with overall survival (OS) (log-rank P = .049, Breslow P = .040, Tarone-Ware P = .044; Figure 7F).

One of the main characteristics of NKTCL is the constitutive activation of STAT3.^15,19,20  Tyrosine phosphorylation at Tyr705 is required for STAT3 to bind to specific DNA target sites, and it was recently reported that STAT3 activation results from constitutive JAK3 phosphorylation at Tyr980 in NKTCL.^22,23  JAK3 was found to be constitutively phosphorylated in 87% of NKTCL primary tumors, and in 21% of these cases, this was the result of activating mutations of JAK3.^22,23  It is likely that multiple mechanisms contribute to this constitutive STAT3 activation in NKTCL, as the net level of tyrosine phosphorylation of STAT3 at Tyr705 reflects a balance between the competing activities of protein kinases and phosphatases. In the present study, we describe a novel molecular mechanism leading to the observed STAT3 deregulation in NKTCL.

Importantly, we found a strong in vivo correlation between low PTPRK expression and high nuclear phospho-STAT3^Tyr705 levels in NKTCL primary tumors. Moreover, our findings from phosphatase substrate-trapping mutant assays, direct phosphatase assays, and in vitro functional studies show that PTPRK binds to STAT3 and directly dephosphorylates phospho-STAT3 at Tyr705, and regulates the levels of phospho-STAT3^Tyr705 in NKTCL. The loss of PTPRK expression (a negative regulator of tyrosine kinases), due to monoallelic deletions and transcriptional silencing through aberrant promoter hypermethylation, likely contributes to lymphomagenesis by activating the STAT3 oncoprotein in NKTCL cells. Previously, phospho-STAT3^Tyr705 was identified as a direct substrate of 2 other R-PTPases, protein tyrosine phosphatase, receptor type T³⁶  and protein tyrosine phosphatase, receptor type D.³⁷  Indeed, increased phospho-STAT3^TYR705 transcriptional activities due to the suppression/mutation of R-PTPases have been documented in several cancers.^24,25  Together with our finding showing that phospho-STAT3^Tyr705 is a true substrate of PTPRK, PTPases could be considered wardens of STAT3 signaling.³⁸ 

The loss of R-PTPases was previously implicated in tumor development and progression.^39,40  In our study, the restoration of PTPRK in NKTCL cells in vitro suppressed tumor cell growth and triggered caspase-mediated apoptosis, whereas the partial knockdown of PTPRK increased tumor cell growth in NKTCL cells. Furthermore, because the nonenzymatic N-terminal extracellular domain of R-PTPases provides a link between cell-cell contact and cellular signaling events involving tyrosine phosphorylation,^31,32  the loss of PTPRK expression in NKTCL cells also resulted in tumor growth and migration. In NKTCL patients, we found a clinical correlation between low tumor cell PTPRK levels and high IPI, suggesting that low PTPRK expression may lead to the dissemination of lymphoma cells to extranasal sites and a possible contribution of PTPRK loss to the NKTCL aggressiveness that is associated with PTPRK non-enzymatic cell-cell contact functions. Similarly, knockdown of PTPRK was recently shown to result in increased proliferation, adhesion, invasion, and migration of breast cancer cells in vitro.⁴¹  A decrease in cell proliferation and colony formation rate was also reported in Raji (Burkitt lymphoma B-cell line) and Jurkat (T-cell leukemia cell line) cells transduced with PTPRK.⁴²  Further evidence for the role of PTPRK in lymphomagenesis comes from previous studies of primary central nervous system lymphomas (CNSLs) with a 140-kb deletion at 6q22-q23 (which contains the PTPRK gene), as well as the observation that a loss of PTPRK expression is common in CNSLs.⁴³  Moreover, there is a tendency for a shorter survival time in CNSL patients with tumors that lack PTPRK expression compared with patients whose tumors maintain PTPRK expression.⁴³  Recently, decreased levels of PTPRK transcript have also been associated with a poor prognosis in breast cancer.⁴¹ 

PTPRK loss has also been described in other lymphoid and nonlymphoid tumors. For example, PTPRK was reported to be downregulated by EBV nuclear antigen 1 in EBV-associated Hodgkin lymphoma,⁴⁴  and PTPRK loss secondary to inactivating mutations was found to contribute to glioma pathogenesis.⁴⁵  However, our results showed that PTPRK inactivation by somatic mutations is not a major mechanism in NKTCL. Recently, aberrant hypermethylation of the PTPRK gene promoter was described in 47% of primary acute lymphoblastic leukemia cases,⁴²  and our results suggest that transcriptional silencing through aberrant promoter hypermethylation exacerbates the consequences of allelic loss of the PTPRK gene in NKTCL.

A significant impact of PTPRK promoter hypermethylation on OS was described in acute lymphoblastic leukemia patients treated with the hyper-CVAD protocol.⁴²  Traditional chemotherapeutic regimens typically result in overall response rates of only 40% to 50% for newly diagnosed NKTCL patients and are typically ineffective for relapsed cases.³³  A novel protocol, SMILE, was recently developed and has resulted in overall response rates of ∼ 80%.^33-35  Because prognostic indicators change when novel treatment strategies are used, our results showed that methylated PTPRK significantly correlated with advanced-stage disease, as well as unfavorable prognosis in NKTCL patients treated with the SMILE regimen. However, the clinical impact of these findings should be considered as very preliminary, as the analysis included only a small sample of 17 NKTCL patients treated with SMILE; these results therefore require future validation in a separate, larger cohort of NKTCL patients treated with SMILE to further verify the prognostic value of our findings.

In addition to the direct dephosphorylation of STAT3 at Tyr705, PTPRK may target STAT3 activation at several levels to achieve tumor suppression in NKTCL. For instance, based on our preliminary in vitro PTPRK restoration and knockdown experiments (supplemental Figure 5), we speculate that PTPRK may also dephosphorylate phospho-JAK3 at Tyr980 and that loss of PTPRK expression in NKTCL potentially provides an additional mechanism leading to constitutive JAK3 phosphorylation at Tyr980 in NKTCL, particularly in cases without JAK3 activating mutations.^22,23 

PTPRK has also been linked to the downregulation of other signaling pathways. It was recently reported that Raji cells transduced with PTPRK showed decreased levels of phospho-Erk1/2 (T202/Y204), phospho-Akt (S473), phospho-STAT3 (S727), and phospho-STAT5 (Y694),⁴²  whereas increased levels of tyrosine-phosphorylated JNK were reported in PTPRK-knockdown prostate cancer cells.⁴⁶  Although these studies implied that PTPRK regulates the phosphorylation of these proteins, the authors did not investigate which one of these dephosphorylation of proteins was directly mediated by PTPRK. Further studies are required to identify and characterize other substrates whose dephosphorylation is directly mediated by PTPRK, rather than through the phosphatases downstream of PTPRK, to assess the potential involvement of additional signaling pathways that may act in conjunction with STAT3 in the molecular pathogenesis of NKTCL. Indeed, a recent study identified PTPRK as a novel binding partner of CD133, which has the ability to directly dephosphorylate tyrosine phosphorylation of CD133 in colon cancer cells.⁴⁷ 

NKTCL is an aggressive disease and the treatment outcome is suboptimal, especially for advanced stage and relapsed/refractory cases.³³  As results from this study demonstrate that methylation of the PTPRK promoter can be reversed by demethylating agent 5-aza-dC, PTPRK-mediated cell signaling may be potentially targeted with epigenetic therapy in NKTCL to develop more effective therapeutic strategies. The re-expression of PTPRK and its correlations with phospho-STAT3^Tyr705 levels as a biomarker could be assessed and correlated with treatment outcome in NKTCL patients treated with DNA hypomethylating agents. Hypomethylating agents are currently used in the treatment of patients with advanced myelodysplastic syndromes.⁴⁸ 

The authors thank Dr Junjiro Tsuchiyama for providing the NKYS cell line⁴⁹  for this study. The authors also thank Dr Wing-Yan Au, Dr Chor-Sang Chim, Dr Anskar Y.H. Leung, and other members of the Division of Hematology, Department of Medicine, Queen Mary Hospital, Hong Kong, who participated in diagnosing and managing the NKTCL patients included in this study, and Silas S. Brown for his careful reading of the manuscript.

This study was supported by grants from the Research Grants Council of Hong Kong (GRF; HKU/775607M) and the University of Hong Kong (SFPBR; 200511159119).

Contribution: Y.-W.C., T.G., and L.S. designed and performed the experiments and analyzed data; K.-Y.W., M.L.Y.W., Y.-P.C., and J.C.O.T. performed the experiments; E.T. and Y.-L.K. selected the NKTCL cases and provided data on the patients’ clinical features and treatment outcomes; F.L. selected the tumor blocks; W.P.L. and G.-D.L. selected the tumor blocks and contributed vital reagents; W.W.L.C. and R.K.H.A.-Y. analyzed the immunohistochemistry slides; N.S. contributed vital reagents; Q.T. offered valuable advice throughout the project; Y.-W.C. and T.G. wrote the manuscript; G.S. conceived the idea, designed and supervised the study, and finalized the manuscript; and all authors commented on and approved the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Gopesh Srivastava, Department of Pathology, The University of Hong Kong, Queen Mary Hospital Compound, 102 Pokfulam Rd, Hong Kong; e-mail: gopesh@pathology.hku.hk.