MYC is overexpressed in ENKTL which is associated with elevated cell proliferation
We retrospectively collected a cohort of 111 cases of ENKTL and examined the MYC expression by IHC. We found that the percentage of MYC-positive cells varied widely across the cases. Receiver operating characteristic (ROC) curve analysis identified 20% as the optimal cut-off value for predicting clinical outcomes, leading to the classification of 83 cases (74.9%) as exhibiting high MYC expression (Fig. 1A, B). However, there was no discernible correlation between MYC expression and key clinical features (Supplemental Table 3). We also probed MYC rearrangement and copy number variation by FISH in 60 selected cases. This revealed MYC gene locus gain in 3 cases (5%), but no evidence of MYC gene rearrangement or amplification (Supplemental Fig. 1). Considering that elevated MYC typically correlates with a higher proliferative rate, we evaluated the Ki-67 index in these cases. Defining a value of ≥60% as a high expression, based on the median value of this cohort, the Ki-67 index was found to be high in 63 of the 83 MYC-high cases (75.9%), compared to only one out of 28 MYC-low cases (3.6%, Fig. 1A, B). A strong correlation was demonstrated between the two markers (R=0.7989, Fig. 1C), suggesting that MYC upregulation is closely related to cell proliferation in ENKTL. Moreover, we analyzed 18 cases with both diagnostic and relapse biopsy samples and found that 13 cases (72.2%) displayed a higher percentage of MYC expression in the relapse sample (Fig. 1D, E). Taken together, this data indicates that MYC protein is frequently overexpressed in ENKTL, which is associated with increased proliferation of the lymphoma cells.
MYC overexpression is a marker of inferior clinical outcome in ENKTL
The median progression-free survival (PFS) and overall survival (OS) for this cohort of patients were 28.1 and 45.4 months respectively. Significantly, patients with high MYC expression had inferior outcomes compared to those with low expression, for both PFS (median 50.4 months, 95% CI 39.1-60.7 vs. median 78.3 months, 95% CI 50.1-86.1) and OS (median 60.6 months, 95% CI 48.9-70.5 vs. median 72.5 months, 95% CI 57.8-89.6, Fig. 2A, B). To further explore the significance of MYC expression in the clinical risk stratification, we assessed the commonly-used PINK-E model29 in this cohort of cases, by which 85 cases were categorized as low-risk, 13 as intermediate-risk, and 13 as the high-risk group. The survival analysis showed that, in general, this model was able to stratify cases with different clinical outcomes. However, one major problem of using this model is that more than 76% of cases were categorized into the low-risk group, in which the 3-year PFS and OS were 59.0% and 71.0%, respectively (Fig. 2C, D). Interestingly, when adding MYC expression to this model, we obtained a new stratification by using the score of 0-1 for low-risk, 2-3 for intermediate-risk, and ≥ 4 for high-risk, which exhibited improved efficacy, especially for distinguishing the low-risk group (Fig. 2E, F). Similarly, we examined the integration of Ki-67 and also obtained an improved stratification than PINK-E (Fig. 2G, H) when using the score of 0-1 for low-risk, 2 for intermediate-risk, and ≥3 for high-risk. We designated these two indexes as PINK-EM and PINK-EK, respectively, which have the potential to serve as useful tools in the clinical management of ENKTL.
MYC overexpression mediates proliferation and survival in ENKTL
To better understand the significance of MYC overexpression in ENKTL, we investigated MYC expression across seven NK lymphoma cell lines and normal NK cells. Both mRNA and protein levels were consistently low in normal NK cells but exhibited substantial variation in NK lymphoma cells. In particular, YT and NK-YS cells expressed MYC at levels similar to normal NK cells, whereas KHYG-1, NK-92, and IMC-1 cells demonstrated significantly elevated MYC expression (Fig. 3A, B). To unravel the functional role of MYC, we sought to deplete it in the cell lines demonstrating overexpression. Given the oncogenic nature of MYC and the inherent challenges of transfecting lymphoma cells, we opted for the siRNA approach and used a blend of two siRNAs to mitigate potential off-target effects. We evaluated three siRNA mixtures in NK-92 and IMC-1 cells, which displayed the highest MYC expression levels. The three mixtures exhibited varying KD efficacy with the first one (S1) to be the highest and thus being selected for subsequent experiments (Fig. 3C). Notably, all siRNA mixtures significantly inhibited cell viability, correlating closely with the degree of MYC depletion (Fig. 3C, D). Specifically, cell viability decreased by approximately two-thirds compared to control cells after 72 hours of transfection of S1. In addition, we noted a significant increase in cell apoptosis post MYC KD, by approximately 22% and 34% in NK-92 and IMC-1 cells, respectively (Fig. 3E). This indicates that MYC overexpression contributes to both the proliferation and survival of NK lymphoma cells.
Next, we performed RNA sequencing in NK-92 and IMC-1 cells post MYC KD. This led to significant changes in gene expression in both cell lines. Specifically, 24 hours post KD, we identified 3995 significantly altered genes in NK-92 cells, with 2474 showing decreased expression and 1521 showing an increase. In IMC-1 cells, we observed significant alterations in 4856 genes, including a decrease in expression for 2931 genes and an increase for 1925 genes (Fig. 4A). Gene Set Enrichment Analysis (GSEA) showed that the differentially expressed genes (DEGs) were highly enriched in canonical MYC target genes in both cell lines (Fig. 4B), suggesting that MYC exerts similar oncogenic functions in ENKTL as it does in other types of cancers. A comparison identified 1746 downregulated and 742 upregulated genes commonly shared between the two cell lines. Pathway analysis showed the downregulated genes primarily involved in metabolic processes and cell cycle regulation, reinforcing the pro-proliferative function of MYC (Fig. 4C). Conversely, upregulated genes were highly in TNF-NF-kB and JAK-STAT signaling pathways, including both pathway activators/effectors and inhibitors, likely reflecting a feedback mechanism of oncogenic signaling (Fig. 4D). In addition, we also profiled the DEGs at 48 hours post MYC KD and obtained a similar result for function enrichment (Supplemental Fig. 2).
Identification of CDK4 as a potential therapeutic target in ENTKL with MYC overexpression
To confirm the identified MYC target genes, we analyzed gene expression profiling (GEP) data from 44 previously studied ENKTL cases.10 We divided cases into three equal-sized groups according to MYC expression levels and then examined the DEGs between the 15-case subsets of low and high MYC expression groups. On average, the high-MYC group exhibited approximately six times the MYC level of the low-MYC group, with DEG analysis revealing 176 upregulated and 58 downregulated genes in the MYC-high group (Fig. 5A, B). Then, we compared the DEGs between the primary cases and the cell lines with MYC KD and identify a list of 68 commonly shared genes, including 66 down-regulated and 2 upregulated (Fig. 5C, D). These genes likely represent bona fide target genes associated with MYC overexpression in ENKTL, and have potentials to serve as therapeutic targets to impair MYC function given that direct MYC inhibition is impractical in current clinical practice. Theoretically, an ideal target needs to meet two essential criteria: It should be intimately relevant to MYC function and be pharmacologically targetable. By a holistic evaluation of the MYC functions demonstrated in the cell experiments, we set our sights on two well-defined MYC targets, HK2 and CDK4.30, 31 We confirmed that, for both genes, the protein levels were significantly downregulated upon MYC KD (Fig. 5E). For pharmacological inhibition, benserazide, a drug to treat Parkinson's disease, was shown to be a selective HK2 inhibitor,32 whereas several inhibitors targeting CDK4, such as palbociclib, have been approved for the treatment of breast cancer. Therefore, both targets were subjected to further inhibition testing.
We treated the seven NK lymphoma cell lines with escalating doses of benserazide and Palbociclib, and found that cells displayed a sensitivity profile strongly correlated with MYC expression level. Specially, cells with MYC overexpression were more susceptible to inhibition (Fig. 6A; Supplemental Fig. 3A). However, the efficient inhibition of benserazide required doses (>10 µM) that would be prohibitive for potential in vivo application. In contrast, palbociclib demonstrated superior potency with effective doses in the nanomolar range and displayed better differentiation between MYC-high and MYC-low cells. Therefore, it was subjected to further investigation, which showed that the treatment not only resulted in substantial cell cycle G1 arrest but also induced significant apoptosis, especially in IMC-1 cells (Fig. 6B). Because CDK4 promotes cell cycle progression through phosphorylating the tumor suppressor protein Rb, thereby releasing E2F, we examined this signaling pathway with palbociclib treatment and observed time-dependent de-phosphorylation of Rb at multiple sites (Fig. 6C). Interestingly, we found that the MYC expression level was significantly decreased, especially after 48 hours of treatment (Fig. 6C). Moreover, RT-PCR assay revealed a marked depletion of MYC mRNA levels (Supplemental Fig. 3B), suggesting that MYC repression likely resulted from transcription reprogramming upon Rb activation. Supportively, the palbociclib-mediated MYC depletion was significantly rescued upon Rb KD in NK-92 and IMC-1 cells (Fig. 6D). Moreover, we applied palbociclib treatment in the Burkitt lymphoma cell lines, Namalwa and Raji, which have MYC/IgH rearrangements, and found that the MYC level was barely affected (Supplemental Fig. 3C). In addition, to determine whether MYC repression was a simple consequence of cell cycle arrest, we performed the double thymidine block but did not observe the depletion of MYC as in the palbociclib treatment (Supplemental Fig. 3D). Collectively, our data indicate that active cell cycle progression mediated by the E2F transcription program is essential for MYC overexpression in NK lymphoma cells, whereby a regulatory feedback loop between MYC and CDK4 is thus formed (Fig. 6E).
Palbociclib suppressed tumor growth in xenograft mouse models
For in vivo testing, we first established the IMC-1 CDX model, in which the tumor cells mainly resided in the viscera, especially the liver (Supplemental Fig. S4). Compared to the vehicle control, palbociclib treatment at 50 mg/kg significantly prolonged the survival of animals, with a 50% increase in median survival (60 days in the treatment group vs. 40 days in the control group) (Fig. 7A). To assess treatment effectiveness under low MYC expression conditions, we conducted an additional experiment using the xenograft model of YT cells which are characterized by minimal MYC expression. Consistent with in vitro data, the YT cell xenograft, which readily forms subcutaneous tumors, showed no response to the treatment (Supplementary Fig. 5). To further evaluate this therapeutic effect, we employed a PDX model of ENKTL with MYC overexpression. Because in the IMC-1 CDX experiment we observed that male mice generally had longer survival, likely due to higher body weight in males at the comparable age, we performed the PDX studies separately for female and male mice. The growth pattern of the PDX model was similar to that of the IMC-1 CDX model, with the viscera organs, especially the liver predominantly involved. In addition to using palbociclib as a single agent at either 50 mg/kg or 100 mg/kg, we also investigated whether a combination with gemcitabine, a core chemotherapeutic drug used in ENKTL treatment, would improve therapeutic efficacy. By preliminary testing, we established a well-tolerated treatment schedule in which gemcitabine was administered as 100mg/mg on day 1, followed by palbociclib 100 mg/kg on days 4-6, continuously for 3 weeks. We found that either palbociclib or gemcitabine monotherapy, moderately prolonged the animal survival, (median survival increases: 31%-50% in female mice and 21%-44% in male mice), whereas the combined treatment substantially improved the survival (median survival increases: 93% in female mice, p=0.0021, and 67% in male mice, p=0.0018) (Fig. 7B). Besides the survival assessment, we also employed a cohort of mice (three per treatment group) to examine the tumor growth in major organs at the end of treatment. Compared to control groups, palbociclib treatment at 100 mg/kg as a single agent, significantly decreased tumor burden in visceral organs, especially in the liver, along with a marked reduction in MYC expression. However, residual tumor cells, especially along blood vessels, are easy to be identified. While gemcitabine monotherapy induced less significant tumor reduction compared to palbociclib, the combination of both marked improved the therapeutic efficacy, leaving minimal residual tumor cells in various visceral organs (Fig. 7C).