Selective JAK2 Inhibition Specifically Decreases Hodgkin Lymphoma and Mediastinal Large B-cell Lymphoma Growth In Vitro and In Vivo

Classical Hodgkin lymphoma (cHL) is a tumor of crippled germinal center B cells that lack surface immunoglobulin expression and B-cell receptor-mediated signals and rely on alternative survival pathways (1). These tumors include small numbers of malignant Reed–Sternberg cells within an extensive immune/inflammatory cell infiltrate (1).

The most common subtype of cHL, nodular sclerosing HL (NSHL), shares certain clinical and molecular features with a non-Hodgkin lymphoma, primary mediastinal large B-cell lymphoma (MLBCL) (2). Both of these lymphoid malignancies frequently occur in young adults and often present as localized nodal/anterior mediastinal masses. In earlier studies, we and others defined shared molecular signatures of MLBCL and cHL including constitutive activation of NF-κB (2–4), frequent copy number gains of chromosome 9p24 and JAK2 overexpression (5, 6).

More recently, we integrated high-resolution copy number data with transcriptional profiles in cHL and MLBCL and defined chromosome 9p24.1 amplification as a recurrent alteration in 40% of primary nodular sclerosing Hodgkin lymphomas and 60% of primary MLBCLs (7). In these diseases, the recurrent 9p24.1 amplicon most often includes the immunoregulatory PD-1 ligand genes, PD-L1 (CD274) and PD-L2 (PDCD1LG2), and JAK2 (7). In cHL and MLBCL cell lines and primary tumors, JAK2 amplification increased JAK2 protein expression and activity and further induced PD-1 ligand expression via JAK2/STAT signaling (7). In these studies, we treated a panel of cHL and MLBCL cell lines with commercially available tool JAK2 inhibitors and found that these compounds decreased cell line proliferation and PD-L1 expression (7). These preclinical studies provided the rationale for evaluating the activity of clinical grade JAK2 inhibitors in lymphoid malignancies such as cHL and MLBCL with frequent JAK2 amplification.

Clinical grade JAK2 inhibitors including ruxolitinib and fedratinib (SAR302503, previously TG101348) have been extensively analyzed in preclinical models of myeloproliferative disorders with activating JAK2 mutations (JAK2V617F) and clinical trials of patients with these diseases (8–12). An additional less potent pan-JAK inhibitor, SB1518, has been assessed in myelofibrosis and additional hematologic malignancies (8, 13). Although ruxolitinib has equal efficacy against JAK1 and JAK2 (3 nmol/L IC₅₀) and additional TYK2 inhibition (20 nmol/L IC₅₀), fedratinib is a selective ATP-competitive JAK2 inhibitor (3 nmol/L IC₅₀) with less activity against the other JAK family members, JAK1, TYK2, and JAK3 (8, 10, 14). In initial preclinical murine models of the myeloproliferative disorder, polycythemia vera, fedratinib (TG101348) exhibited clear in vivo efficacy with reduction of JAK2V617F-driven disease (10, 11). In subsequent phase I and II clinical trials of fedratinib (TG101348) in myelofibrosis, the selective JAK2 inhibitor was well tolerated and associated with significant reduction in disease burden and durable clinical benefit (12, 15).

Given the importance of JAK2/STAT signaling in cHL and MLBCL, the shared recurrent amplification of 9p24.1/JAK2 and the utility of fedratinib in additional JAK2-dependent hematopoietic malignancies, we have assessed the in vitro and in vivo activity of this JAK2-selective inhibitor in cHL and MLBCL.

All cell lines were obtained from the DSMZ cell bank. The Karpas 1106P (K1106P) MLBCL cell line and HDLM2 cHL cell line were grown in RPMI-1640 medium (Mediates) supplemented with 20% FBS, L-glutamine, and penicillin/streptomycin. The KMH2, L428, and L1236 cHL cell lines were grown in RPMI-1640 medium (Mediates) supplemented with 10% FBS, L-glutamine, and penicillin/streptomycin and the SUPHD1 cHL cell line was maintained in McCoy's 5A medium (Invitrogen) supplemented with 20% FBS, L-glutamine, and penicillin/streptomycin.

The pSTAT1 and pSTAT5 monoclonal antibodies and pJAK2, pSTAT3, and JAK2 antisera were purchased from Cell Signaling. The pSTAT6 antiserum was obtained from ThermoFisher Scientific and the c-MYC and GAPDH antisera and the PIM1 monoclonal antibody were purchased from Santa Cruz Biotechnology. Secondary anti-mouse and anti-rabbit antibodies, conjugated with horseradish peroxidase, were obtained from GE Healthcare. Fedratinib was a gift from Sanofi Aventis.

cHL and MLBCL cell lines were resuspended at a concentration of 2 × 10⁵ cells/mL in culture medium and 50 μL of the cell suspension (1 × 10⁴ cells) was added to each well of a 96-well plate. Thereafter, 50 μL of medium and vehicle (dimethyl sulfoxide) or fedratinib was added to each well in 2-fold serial dilutions. Forty-eight hours later, cellular proliferation was evaluated with the AlamarBlue assay (Invitrogen) according to manufacturer's instructions.

Cells were treated with fedratinib or vehicle for 48 hours and cellular apoptosis was subsequently analyzed by flow cytometry with an Annexin V-APC/DAPI Apoptosis Detection Kit (BD Pharmingen) according to the manufacturer's instructions.

After extraction with radioimmunoprecipitation assay buffer, proteins from MLBCL and cHL cell lines were size fractionated with NuPAGE 4% to 12% Bis-Tris gel electrophoresis (Invitrogen), transferred to PVDF membranes (Millipore Corp.), blocked with 5% BSA in 0.1% TBST, incubated with primary antibodies diluted in 3% BSA in 0.1% TBST according to manufacture's instructions, and secondary antibodies conjugated to horseradish peroxidase and then detected with enhanced chemiluminescence (GE Healthcare).

After cells were treated with vehicle or fedratinib for 24 hours, RNA was extracted with TRizol (Invitrogen) and cDNA was synthesized with the SuperScript III First-Strand Synthesis System Kit (Invitrogen). qRT-PCR was performed using the Applied Biosystem 7300 real-time PCR system with inventoried TaqMan PD-L1 probes (Hs01125299) and internal reference huGAPDH (Applied Biosystem).

After treatment with vehicle or fedratinib for 48 hours, 1 × 10⁶ cells were resuspended in 100 μL of PBS, incubated with 2 μg of phycoerythrin-conjugated PD-L1 antibody (Clone 29E.2A3) or isotype control (immunoglobulin G2b κ, BioLegend) for 30 minutes at room temperature, washed with PBS, and then resuspended in 500 μL of PBS. After staining, 2 × 10⁴ cells were analyzed with a BD FACS Canto flow cytometer (BD Biosciences). FlowJo software (TreeStar) was used to select viable cells by forward and side scatter and generate histograms and median fluorescence intensities.

Intracellular phospho-flow cytometry was performed as previously described (7). In brief, single-cell suspensions were fixed, permeabilized and then stained with either isotype control or Alexa Fluor 647–conjugated phospho-STAT3 (pY705) antibody (BD Bioscience). Phospho staining was normalized to fixed, permeabilized, and isotype-stained cells and fold change was calculated by comparison of median fluorescence intensity values.

Immunohistochemistry (IHC) for pSTAT1 and pSTAT3 was performed using 4 μm-thick, formalin-fixed, paraffin-embedded tissue sections. Slides were baked, deparaffinized in xylene, passed through graded alcohols, and then antigen retrieved with 1 mmol/L EDTA, pH 8.0 (Invitrogen) in a steam pressure cooker (Decloaking Chamber; BioCare Medical) as per manufacturer's instruction. All further steps were carried out at room temperature in a hydrated chamber. Slides were pretreated with Peroxidase Block (Dako) for 5 minutes to quench endogenous peroxidase activity, and then washed in 50 mmol/L Tris-Cl, pH 7.4. Slides were blocked using Protein Block (Dako) as per manufacturer's instruction, and subsequently incubated with rabbit monoclonal pSTAT1 (clone 58D6, 1:100 dilution for cell lines; Cell Signaling) and rabbit monoclonal pSTAT3 (clone D3A7, 1:200 dilution for cell lines and primary Hodgkin lymphoma; 1:1500 for xenograft studies, Cell Signaling) in diluent (Dako) for 1 hour. Slides were then washed in Tris buffer and treated with anti-rabbit horseradish peroxidase–conjugated antibody (Envision Plus; Dako) for 30 minutes. After further washing, immunoperoxidase staining was developed using a 3,3′-diaminobenzidine (DAB) chromogen (Dako) for 5 minutes. Slides were counterstained with hematoxylin, dehydrated in graded alcohol and xylene, and mounted and coverslipped.

IHC for pJAK2 (clone D3A7, Catalog No. 3771, 1:12.5 dilution for cell lines; Cell Signaling) was performed using 4 μm-thick, formalin-fixed, paraffin-embedded tissue sections on Benchmark XT autostainer (Ventana Medical Systems). UltraView Universal DAB Detection Kit (Catalog No. 760–500; Ventana) was used per manufacturer's instruction. Slides were then washed in soap water and distilled water for post-IHC staining, dehydrated in graded alcohol and xylene, and mounted and coverslipped. Anti-CD20 IHC was performed according to standard protocols (Catalog No. N1502, clone L26; Dako).

The firefly luciferase and mCherry expressing Karpas 1106P and HDLM2 cell lines were engineered as previously described (16). Subsequently, 5 × 10⁶ viable Karpas 1106P Luc-mCherry cells in 250 μL PBS were injected via the lateral tail veins of 8-week-old male NOD SCID IL2Rγ^null mice (Charles River Laboratories). Disease burden was quantified with bioluminescence imaging (IVIS Spectrum, Caliper Life Sciences). Total body luminescence was quantified using the Living Image software package (Caliper Life Sciences) and presented as mean ± SEM with statistical significance determined by Student t test. Luc-mCherry HDLM2 cells (1 × 10⁷ cells) in 30% Matrigel were inoculated subcutaneously into mice and tumor size was measured by calipers.

Two weeks following tumor inoculation, animals with established disease documented by imaging (Karpas 1106P) or caliper measurement (HDLM2) were divided into 2 cohorts with equal mean bioluminescence or tumor size and treated twice daily with 120 mg/kg fedratinib or vehicle (0.5% methylcellulose, 0.05% Tween 80) via gavage.

An independent cohort of mice was treated for 5 days with either vehicle or 120 mg/kg fedratinib to assess the pharmacodynamic efficacy of the JAK2 inhibitor. Two hours after the last dose, mice were euthanized and tissues fixed by intracardiac perfusion with 10% formalin. Tumor tissues were harvested and subjected to further fixation overnight in 10% neutral-buffered formalin in preparation for immunohistochemical analyses. All animal studies were performed according to Dana-Farber Cancer Institute Institutional Animal Care and Use Committee approved protocols.

We first treated a panel of cHL and MLBCL cell lines of known 9p24.1/JAK2 copy number with increasing doses of fedratinib (0.313–5 μmol/L) or vehicle alone. The proliferation of cHL and MLBCL cell lines was significantly inhibited by fedratinib and there was an inverse correlation between 9p24.1/JAK2 copy number and the compound EC₅₀. At a given dose of the JAK2 inhibitor (1.25 μmol/L), there was a significant association between the ranked values of inhibition and copy number gain (P = 0.009, all cell lines; P = 0.019, cHL cell lines, Kruskal–Wallis test; Fig. 1A and Supplementary Table S1). In addition, fedratinib induced the apoptosis of cHL and MLBCL cell lines and had a more pronounced effect on lines with higher 9p24.1/JAK2 copy numbers (Fig. 1B). These results indicate that the selective JAK2 inhibitor inhibits the proliferation and induces the apoptosis of cHL and MLBCL cell lines in a JAK2 copy number–dependent manner.

To more specifically define the downstream consequences of fedratinib treatment in cHL and MLBCL cell lines with defined JAK2 copy number, we assessed the effect of the compound on the phosphorylation of JAK2 and multiple STAT family members. As anticipated, baseline levels of pJAK2 were most abundant in the cHL cell lines with highest 9p24.1/JAK2 copy gain, SUPHD1 and HDLM2 (Fig. 2A). At 24 hours, treatment with fedratinib markedly decreased JAK2 phosphorylation in a dose-dependent manner and cHL and MLBCL cell lines with high 9p24.1/JAK2 copy number were most sensitive to chemical JAK2 inhibition (Fig. 2A).

Four STAT family members have been reported to be highly active in cHL and MLBCL, STAT1, 3, 5, and 6 (1, 7, 17, 18), prompting us to assess their phosphorylation following fedratinib treatment. The phosphorylation of STAT1, 3, and 6 was dramatically inhibited by chemical JAK2 blockade in a 9p24.1 copy number–dependent manner in the cHL and MLBCL cell lines (Fig. 2A). In contrast, basal pSTAT5 levels were less closely correlated with 9p24.1/JAK2 copy number and variably responsive to fedratinib treatment (Fig. 2A), potentially reflecting the regulation of STAT5 expression and activity by additional pathways and other JAK family members (19, 20). After demonstrating the 9p24.1/JAK2 copy number–associated decrease in JAK2 and STAT1, STAT3, and STAT6 phosphorylation following 24 hours of fedratinib treatment, we evaluated the time course of chemical JAK2 inhibition. In cHL and MLBCL cell lines with high 9p24.1/JAK2 copy number, phosphorylation of JAK2, STAT1, 3, and 6 decreased after only 2 to 4 hours of fedratinib treatment (Fig. 2B). These findings reveal a copy number–dependent, rapid and specific sensitivity of cHLs and MLBCL to chemical JAK2 inhibition.

Given the close association between sensitivity to fedratinib treatment and decreased phosphorylation of JAK2 and downstream STATs, we sought to develop an immunohistochemical signature of baseline JAK2 activity and response to chemical JAK2 inhibition. In cHL and MLBCL cell lines with high 9p24.1/JAK2 copy numbers, basal JAK2, STAT1, and STAT3 phosphorylation were readily detectable by IHC and dramatically reduced following 24 hours of fedratinib treatment (Fig. 3A and B). Similar results were obtained after only 4 hours of fedratinib exposure (Supplementary Fig. S1).

In a representative primary cHL with known 9p24.1/JAK2 amplification, STAT3 phosphorylation was also apparent by IHC (Fig. 3C). These data suggest that pSTAT3 may be a useful marker to assess baseline JAK2/STAT pathway activity and sensitivity to targeted chemical inhibition in vivo.

The previously described dependence of PD-L1 expression on JAK2/STAT signaling (7) prompted us to assess PD-L1 transcript abundance and cell surface expression following fedratinib treatment. Chemical JAK2 inhibition reduced PD-L1 transcript levels and cell surface expression in 3 of the 4 cHL cell lines and the MLBCL cell line with high 9p24.1 copy numbers (Fig. 4A and B), underscoring the potential role of fedratinib in modulating PD1 signaling.

We next assessed the consequences of chemical JAK2 inhibition on an additional downstream target of JAK2/STAT signaling, PIM1, and an indirect target of JAK2, c-MYC. In previous studies, JAK2 was reported to phosphorylate the histone H3 tail and block heterochromatin formation and epigenetically modify c-MYC expression (21, 22). We found that fedratinib treatment primarily decreased c-MYC expression in cHL cell lines with higher 9p24.1/JAK2 copy numbers and modestly reduced c-MYC levels in the MLBCL cell line (Fig. 5).

The PIM1 oncogene encodes a serine/threonine kinase, which strongly cooperates with c-MYC in murine lymphoma formation (23–25). PIM1 and c-MYC are frequently coexpressed in human hematologic malignancies and PIM1 is a known target of JAK2/STAT signaling (26–28). Consistent with these observations, baseline PIM1 transcript abundance was >3-fold higher in the cHL and MLBCL cell lines with high 9p24.1/JAK2 copy numbers (Supplementary Fig. S2). Treatment with fedratinib significantly reduced the expression of PIM1 in the cHL lines with the highest 9p24.1/JAK2 copy number but did not decrease PIM1 expression in the Karpas 1106P MLBCL cell line. These data point to potential differences in the downstream targets of chemical JAK2 inhibition in cHLs and MLBCLs with JAK2 amplification.

After demonstrating the activity and specificity of fedratinib in in vitro assays and determining the optimal dose for in vivo studies (Supplementary Fig. S3), we further investigated the antitumor efficacy of the chemical JAK2 inhibitor in murine xenograft models of cHL and MLBCL with 9p24.1/JAK2 amplification [HDLM2 (cHL) and Karpas 1106P (MLBCL)]. Karpas 1106P cells, which ectopically expressed the firefly luciferase gene, were injected via tail vein and tumor growth was serially monitored via bioluminescence imaging. After documenting established tumors, the MLBCL-bearing mice were treated with fedratinib 120 mg/kg, p.o. twice a day. In an initial cohort of animals, we analyzed pSTAT3 expression as a pharmacodynamic marker following 5 days of treatment. Fedratinib-treated animals had markedly decreased immunostaining of pSTAT3 in lymphomatous Karpas 1106P bone marrow infiltrates (Fig. 6A). Consistent with these initial findings, chemical JAK2 blockade decreased the in vivo growth of Karpas 1106P MLBCL (Fig. 6B) and prolonged the survival of tumor-bearing animals (Fig. 6C, P = 0.0002).

In an additional xenograft model, HDLM2 cHL cells were inoculated subcutaneously into the flanks of NSG mice and tumor growth was monitored by external caliper measurements. Treatment was initiated following the establishment of 100 mm³ tumors. As in the Karpas 1106P MLBCL xenografts (Fig. 6A), 5 days of fedratinib treatment markedly reduced pSTAT3 expression in the HDLM2 cHL xenografts (Fig. 6D). In addition, HDLM2 tumor growth was significantly decreased from 5 days to 25 days of treatment, the last measurable timepoint (Fig. 6E). Thereafter, animals were sacrificed, subcutaneous tumor masses were harvested, and suspended residual tumor cells were analyzed for pSTAT3 expression by intracellular phosphoflow cytometry. In the mCherry-positive residual viable HDLM2 cHL cells, pSTAT3 expression was significantly reduced following fedratinib treatment in comparison to vehicle alone (Fig. 6F). Taken together, these data confirm the prolonged in vivo activity of the chemical JAK2 inhibitor and the associated reduction of JAK2/STAT signaling in the MLBCL and cHL xenograft models.

Our combined in vitro and in vivo studies demonstrate that chemical JAK2 inhibition specifically decreases cHL and MLBCL growth in a 9p24.1/JAK2 copy number–dependent manner. Specifically, we find a copy number–dependent inhibitory effect of the clinical JAK2 inhibitor, fedratinib, on cellular proliferation and viability of cHLs and MLBCLs, JAK2/STAT phosphorylation, and regulation of downstream targets. For the first time, we also document the highly significant in vivo efficacy of the JAK2 inhibitor in xenograft models of cHL and MLBCL with 9p24.1/JAK2 amplification and utilize pSTAT3 as a robust biomarker of JAK2/STAT pathway activity and treatment response. Although clinical development of fedratinib was recently halted because of unanticipated neurotoxicity, additional JAK2 inhibitors are approved for use in other hematologic malignancies and available for analysis in cHL and MLBCL.

Genomic analyses from our group and others highlight the frequency of 9p24.1/JAK2 copy gain in 40% of primary NSHLs and over 60% of primary MLBCLs (6, 7). The current studies suggest that cHLs and MLBCLs with genetic alterations of JAK2 will be most sensitive to fedratinib treatment. Although 9p24.1/JAK2 amplification seems to be the predominant mechanism of deregulating JAK2 activity in cHL and MLBCL, additional alterations have been described. In a series of 131 cHLs, rare JAK2 rearrangements were identified, including 2 SEC31A-JAK2 translocations and 2 JAK2 translocations with unknown partners (29). To date, there is no evidence for JAK2V617F or exon 12 mutations in primary cHL or MLBCL (30, 31). However, additional alterations of negative regulators of JAK2/STAT signaling, such as PTPN2 and SOCS1, have been described. Inactivating PTPN2 mutations were identified in a single cHL cell line but were not seen in 27 additional primary NSHLs (32). Deletions or inactivating mutations of SOCS1 are more common in cHL and MLBCL cell lines and primary tumors (33, 34). Given the increased activity of fedratinib in cHLs and MLBCLs with JAK2 amplification and the additional potential mechanisms of perturbing JAK2/STAT signaling (29, 32–34), it will be important to comprehensively analyze the pathway and its regulators in future clinical trials.

In our current studies, we find pSTAT3 to be a robust biomarker of copy number–dependent JAK2 activity. Our findings are consistent with earlier studies linking numerical aberrations of JAK2 with increased percentages of STAT3⁺ tumor cells in cHL and MLBCL (6, 35).

In addition to inhibiting phosphorylation of JAK2-dependent STAT family members, fedratinib decreased the expression of candidate downstream targets, c-MYC and PIM1, in cHL cell lines with high 9p24.1/JAK2 copy number. However, in the fedratinib-treated MLBCL cell line, c-MYC modulation was less striking and PIM1 was not downregulated. These observations point to potential differences in the signaling and resistance pathways in cHLs and MLBCLs.

In earlier studies, we demonstrated that JAK2 was co-amplified with PD-L1 (CD274) and PD-L2 (PDCD1LG2) loci as part of the 9p24.1 amplicon, of note because JAK2/STAT signaling further induces the expression of PD-1 ligands (7). The 5′ regulatory region of PD-L1 includes a classic ISRE/IRF1 module and several degenerate STAT binding sites (7). Previous functional studies confirm that chemical JAK2 inhibition modulates PD-L1 expression via the documented ISRE/IRF1 control element (7). In this study, we extend these findings to demonstrate that the clinical JAK2 inhibitor, fedratinib, decreases PD-L1 transcript abundance and cell surface expression in 3 of 4 cHL lines and the MLBCL cell line with high 9p24.1/JAK2 copy numbers.

In this study, we also document the in vivo efficacy of chemical JAK2 inhibition in 2 xenograft models of tumors with 9p24.1/JAK2 amplification, Karpas 1106P MLBCL and HDLM2 cHL. In both xenograft models, pSTAT3 was a robust pharmacodynamic biomarker of the efficacy of chemical JAK2 inhibition. Furthermore, chemical JAK2 inhibition significantly decreased tumor growth in both xenograft models and prolonged survival in the model evaluable for this endpoint (Karpas 1106P).

In summary, the current studies support the clinical evaluation of chemical JAK2 blockade in cHL and MLBCL with known 9p24.1/JAK2 copy gain and further consideration in additional lymphoid tumors with 9p24.1/JAK2 alterations (6, 36).

M.A. Shipp reports receiving a commercial research grant from Sanofi Aventis. No potential conflicts of interest were disclosed by the other authors.

Conception and design: Y. Hao, B. Chapuy, M.A. Shipp

Development of methodology: Y. Hao, B. Chapuy, M.A. Shipp

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Hao, B. Chapuy, S.J. Rodig, M.A. Shipp

Analysis and interpretation of data (e.g., statistical analysis, biostatist\ics, computational analysis): Y. Hao, B. Chapuy, S. Monti, H.H. Sun, S.J. Rodig, M.A. Shipp

Writing, review, and/or revision of the manuscript: Y. Hao, B. Chapuy, S.J. Rodig, M.A. Shipp

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H.H. Sun

Study supervision: M.A. Shipp

This work was supported by NIH CA161026, the Miller Family Research Fund, and the Sanofi Aventis Research Agreement.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.