MGA deletion leads to Richter’s transformation by modulating mitochondrial OXPHOS

Abstract

Richter’s transformation (RT) is a progression of chronic lymphocytic leukemia (CLL) to aggressive lymphoma. MGA (Max gene associated), a functional MYC suppressor, is mutated at 3% in CLL and 36% in RT. However, genetic models and molecular mechanisms of MGA deletion that drive CLL to RT remain elusive. We established an RT mouse model by knockout of Mga in the Sf3b1/Mdr CLL model using CRISPR-Cas9 to determine the role of Mga in RT. Murine RT cells exhibited mitochondrial aberrations with elevated oxidative phosphorylation (OXPHOS). Through RNA sequencing and functional characterization, we identified Nme1 (nucleoside diphosphate kinase) as an Mga target, which drives RT by modulating OXPHOS. Given that NME1 is also a known MYC target without targetable compounds, we found that concurrent inhibition of MYC and electron transport chain complex II substantially prolongs the survival of RT mice in vivo. Our results suggest that the Mga-Nme1 axis drives murine CLL-to-RT transition via modulating OXPHOS, highlighting a potential therapeutic avenue for RT.

INTRODUCTION

Chronic lymphocytic leukemia (CLL) is the most prevalent adult hematologic malignancy in North America (1, 2). Richter’s transformation (RT) affects 10 to 15% of patients with CLL (3), turning indolent CLL to more aggressive lymphoma, mainly diffuse large B cell lymphoma (DLBCL), and leading to a dismal prognosis. Median overall survival is less than 1 year, even with current moderate or high-intensity chemoimmunotherapy treatments (4). Moreover, patients with RT after targeted therapy such as ibrutinib have a survival of about 4 months (5). Thus, understanding RT biology and designing better therapeutic approaches are urgently needed to treat this deadly disease.

Despite the pressing unmet clinical need, RT research faces considerable roadblocks. Difficulties in RT diagnosis arise from clinical symptoms and pathological changes in the morphology of transformed cells. Lack of a clear-cut diagnostic definition, the rarity of RT samples, and similarity with aggressive CLL make identifying RT challenging (6). As a result, even though large-scale sequencing data are available in CLL (7, 8), molecular mechanisms underlying the CLL-to-RT transition are still limited. Second, the lack of human RT cell lines and limited animal models that faithfully recapitulate the human CLL-to-RT transition hamper mechanistic studies of RT and the testing of therapeutic agents in vivo (9–14). Hence, using mouse models to examine the CLL-to-RT transition may pave the way for breakthroughs in understanding and treating RT.

Genetic studies using matched CLL and RT samples derived from 19 patients revealed that genetic lesions involved in the cell cycle (89%, TP53 and CDKN2A/B), MYC activation (74%, MYC and MGA), NOTCH pathway (32%, NOTCH1 and SPEN), nuclear factor κB signaling (74%, BIRC3 and EGR2), and chromatin remodeling (79%, SETD2, SETD1A/B, and ARID1A/B) are enriched during CLL-to-RT transformation (15). The functional roles of TP53/CDKN2A/B deletions and NOTCH activation have been demonstrated to promote the transition of CLL to DLBCL (10, 11). However, the roles of other genetic lesions during CLL-to-RT remain uncharacterized. In particular, loss-of-function mutations or deletions in MGA (Max gene associated), a MYC transcriptional repressor, are found in CLL at 3% but at 36% in RT, indicating a critical yet unknown role during CLL-to-RT (7, 15, 16). Here, we start with a murine CLL model based on the two most common genetic alterations found in human CLL samples, Sf3b1 mutation and Mdr deletion [mimicking del(13q)], and determine the impact of MGA deletion in the pathogenesis of RT. Our results highlight a MGA-driven regulatory axis driving the CLL-to-RT transition and emphasize targeting this pathway as an effective treatment.

RESULTS

Mga deletion leads to rapid CLL-to-RT transition upon engraftment into recipient mice

To determine the functional role of Mga in CLL and RT, we used a method (9, 17) to introduce genetic lesions into murine B cells, in which murine hematopoietic stem cells (Lin⁻cKit⁺Sca1⁺; hereafter LSK) were genetically edited in vitro and then adoptively transferred into sublethally irradiated CD45.1 mice. We first crossed a murine CLL line (CD19^cre/+Mdr^fl/+Sf3b1 K700E^fl/+) (18) with a mouse strain that conditionally expresses Cas9 (19) to obtain a donor mouse line (Cd19-Cre^fl/+Sf3b1^fl/+ Mdr ^fl/+Cas9^fl/+). We isolated LSK cells (CD45.2⁺) from these mice and transduced them with lentivirus expressing nontargeting single guide RNA (sgRNA) or sgRNA targeting Mga in vitro. These in vitro–edited LSK cells were then engrafted into sublethally irradiated recipient mice (CD45.1, n = 15 per group) by tail-vein injection. CLL onset (B220⁺CD5⁺ CLL-like cells) was monitored by flow cytometric analysis of peripheral bleeds bimonthly, starting at age 6 months and ending at 24 months (Fig. 1A). All transferred LSK cells achieved successful engraftment along with high efficiency of Mga KO (fig. S1, A and B). Our disease monitoring (at 14, 17, and 19 months) revealed a mixture of small and large cells on the basis of forward and side scatter using peripheral blood–based flow cytometry (Fig. 1B and fig. S1, C and D). B220 and CD5 were expressed by both small and large cells. Three of 15 mice carrying Mga deletion showed clonal expansion of B220⁺CD5⁺ cells and leukemic infiltration of tissues by 21 months, whereas no mice in the control group developed CLL-like disease (Fig. 1B and fig. S1, C to E). The observation of large cells in the peripheral bleeds (fig. S1F) led us to consider whether the large B220⁺CD5⁺ cells were RT cells.

Fig. 1. Knockout of Mga leads to RT in mice.

(A) Schema for generation of CLL-to-RT murine model. Genetically manipulated LSK (lin⁻c-kit⁺Sca-1⁺) hematopoietic progenitor cells derived from Cd19-Cre^fl/+Cas9 ^fl/+Sf3b1^fl/+Mdr ^fl/+mice (CD45.2) were transduced with lentivirus expressing sgRNAs against Mga or control were engrafted into sublethally irradiated CD45.1 recipient mice. The disease onset was monitored by examining the presence of B220⁺CD5⁺ cells from peripheral blood by flow cytometer. Transduced cells were identified based on mCherry expression. (B) CLL disease in sgMga group monitoring in the peripheral blood 14, 17, and 19 months posttransplant by flow cytometry. Orange indicates the small lymphocytes, and green indicates the large-sized lymphocytes. Accumulation of B220⁺CD5⁺ (%) CLL-like cells is shown in the small and large lymphocytes. sgControl is also shown (blue box). sgControl no CLL is shown in the orange box. (C) Deletion of Mga leads to CLL. Accumulation of B220⁺CD5⁺(%) CLL-like cells in the spleens, bone marrow, and lymph nodes from primary engrafted mice (CD45.2) was detected by flow cytometer 21 months postengraftment. (D) Schema of engraftment of splenic B220⁺CD5⁺ CLL-like cells (derived from CD45.2) into either NSG or sublethally irradiated CD45.1 mice by tail-vein injection. (E and F) Example of accumulation of B220⁺(%) cells in the spleens, bone marrow, and blood from the secondary NSG (E) and CD45.1 (F) mice. (G) Immunohistochemistry staining of CD5, B220, MYC, and Ki67 with H&E staining on spleen sections from primary (CLL) and secondary (RT) mice. 40× (scale bars, 20 μm) and 100× (scale bars, 10 μm) magnification images are shown. (H) Representative H&E staining images from mouse bone marrow and liver in CLL and RT mice. 40× (scale bars, 20 μm) and 100× (10 μm) magnification images. (I) Flow cytometric analysis of CD71, CD38, and CD21 expression in two mouse RT cases and one CLL spleen case.

We transplanted total splenic cells (Fig. 1C, mainly B220⁺CD5⁺) from the primary engrafted mice into either immunocompromised NSG or immunocompetent sublethally irradiated CD45.1 mice (secondary mice) (Fig. 1D). Transplantation of the total splenic cells resulted in a rapid expansion of B220⁺ B cells along with CD5 loss at 3 weeks postengraftment and leukemic infiltration in different lymphoid tissues (spleen, bone marrow, and lymph node) (Fig. 1, E and F). All of the mice had splenomegaly and infiltration in bone marrow and livers (Fig. 1, G and H). Moreover, hematoxylin and eosin (H&E) staining of spleen sections confirmed the presence of CLL-like cells with morphology highly similar to leukemia in the primary engrafted mice; however, upon secondary engraftment, these cells became larger with a morphology similar to aggressive lymphoma cells (Fig. 1G and fig. S2). Immunohistochemical staining with CD5, B220, Ki67, and MYC confirmed CLL identity in the primary mice and supported the presence of hyperproliferative and aggressive lymphoma in the secondary mice. We observed increased Ki67 and MYC in the aggressive secondary mice splenic B cells (figs. S2 and S3). Of note, murine RT cells showed a significantly (P < 0.001) higher Ki67 proliferation index (>20%) than did CLL cells (5 to 10%), indicating a high rate of RT cell proliferation. Consistent with our murine model findings, we observed high MYC and Ki67 (P < 0.001) in human RT samples compared with those of healthy controls by immunohistochemistry (figs. S4 and S5). Flow cytometric analysis of murine RT cells revealed increased CD71 and CD21 expression along with decreased CD38 expression, consistent with previously reported RT cell characteristics (Fig. 1I) (20, 21). The CLL-like and aggressive lymphoma–like cells were clonal based on immunoglobulin k expression and shared the same immunoglobulin heavy-chain gene variable region (IGHV) segment usage (table S1). These observations indicated that Mga deletion coupled with Mdr deletion and Sf3b1 mutation results in CLL transformation to RT, establishing a murine CLL-to-RT transition murine model.

Mga deletion leads to activation of MYC and NME1 in CLL and RT cells

To dissect the molecular mechanism underlying the CLL-to-RT transition, we performed RNA sequencing (RNA-seq) using RNA derived from splenic B cells from four mice without CLL, two mice with CLL, and their subsequent transformed RT cells from the secondary engraftment. Through differential gene expression analysis, we identified 424 CLL- and 1508 RT-associated up-regulated genes, as well as 200 CLL-to-RT–associated up-regulated genes (Fig. 2A). CLL dysregulated genes were highly enriched for MYC pathway targets and DNA repair, whereas RT dysregulated genes were mainly enriched for MYC targets and inflammation pathways (Fig. 2B). Of note, CLL-to-RT transition was coupled with significant (q < 0.001) enrichment of MYC pathway targets and oxidative phosphorylation (OXPHOS) (Fig. 2, C and D). When overlapping the up-regulated (log₂fc > 1.5 and P < 0.05) gene list with murine CLL and RT datasets, we found 75 up-regulated genes shared between CLL and RT groups (Fig. 2E), of which three genes (MYBBP1A, NME1, and RPL22) appear to be MYC targets directly regulated by MGA on the basis of MGA binding protein and MYC target overlap analysis (22). Among the three genes, NME1 has been previously identified as transcriptionally regulated by MGA via promoter binding in lung adenocarcinoma (22). We confirmed the up-regulation of NME1 and MYC at the mRNA and protein levels in murine RT and CLL cells (Fig. 2F and fig. S6, A and B). We also validated NME1 up-regulation (~5- to 20-fold) in human CLL and RT samples by reverse transcription polymerase chain reaction (PCR; P < 0.01) and immunohistochemistry (P < 0.0001) (Fig. 2G and fig. S6, C and D), respectively. High NME1 expression was significantly (P = 0.004) associated with inferior survival in patients with CLL and DLBCL (Fig. 2, H and I), indicating a possible critical yet unknown role of this gene in lymphomagenesis.

Fig. 2. CLL-to-RT transition is coupled with MYC activation and increased NME1 expression.

(A) RNA-seq was performed using RNA derived from splenic B cells either from mice without or with CLL (Cd19-Cre^+/− Sf3b1 ^fl/+Mdr^fl/+ Cas9 ^fl/+ Mga^−/−), as well as mice transformed to RT (Cd19-Cre^+/− Sf3b1 ^fl/+Mdr^fl/+ Cas9 ^fl/+ Mga^−/−). Heatmap illustrates differential gene expression among splenic B cells without CLL (n = 4), with CLL (n = 2), and with matched RT (n = 2). Significantly up-regulated (log₂fc > 1.5, P < 0.05) and down-regulated genes (log₂fc < −1.5, P < 0.05) are shown in red and blue, respectively. (B) Gene set enrichment analysis (GSEA) of differentially expressed genes in murine CLL and RT with respect to no disease showing top dysregulated pathways ranked by normalized enrichment score (NES). (C and D) GSEA analysis of ranked differentially expressed genes between murine CLL and RT. (E) Venn diagram showing the overlap of significantly dysregulated genes associated with CLL and RT with publicly available datasets, including Hallmark MYC targets V1/2 and MGA ChIP-seq data A549 (ChIP-Atlas). (F) Immunoblotting of MYC, NME1, and GAPDH using splenic B cells derived from wild type (WT); DM, no disease double-mutant (Cd19Cre^fl/+Mdr ^fl/+ Sf3b1 ^fl/+ Mga^−/−) CLL and RT mice. (G) MYC, NME1, and actin expression immunoblotting using human normal B and CLL B cells. (H and I) Kaplan-Meier survival curves for patients with CLL (H) or DLBCL (I) with high or low NME1 expression. CLL (GSE22762) and DLBCL (GSE10846) datasets are from the *PRECOG (prediction of clinical outcomes from genomics) database. (J) Venn diagram shows the overlap of murine-RT dysregulated genes with human CLL-RT dysregulated genes (15). Scatter plot showing the correlation of common dysregulated pathways in murine and human RT. (K) Venn diagram showing overlap of murine-RT dysregulated genes with human RT dysregulated genes (14). Pathway enrichment analysis from the common genes (n = 109) between murine and human RT (n = 3). (L) Heatmap representing hierarchical clustering of murine CLL-RT with human CLL and DLBCL cell lines.

To determine whether murine RT cells have a similar gene expression pattern to human RT cells, we overlapped murine RT-associated genes with three recently reported human RT cell datasets (14, 15, 23). From all three datasets, overlapped RT-associated genes revealed not only the previously identified up-regulated genes but also the enrichment of pathways including MYC, enrichment of MYC pathway targets, cell cycle, mTORC1 signaling, and glycolysis, indicating that our murine RT model resembles human RT (Fig. 2J). In particular, upon overlapping our murine RT RNA-seq with a recently published human RT RNA-seq dataset (n = 6), we found NME1 and NME4 up-regulated in the human RT along with enrichment of MYC, OXPHOS, and mTORC1 pathways (Fig. 2J), indicating that our murine RT cells shared a gene expression pattern similar to human RT cells. Similarly, on overlapping the murine RNA-seq dataset with two human RNA-seq datasets (14, 23), MYC targets V1 and mTORC1 were among the significantly (P = 0.02) enriched pathways (Fig. 2K and fig. S6E). It is known that about ~90% of human RT cells are DLBCLs of the activated B cell–like (ABC) type (24, 25). To determine what type of human lymphoma our murine RT cells resembled, we compared the gene expression of murine RT cells to 12 different types of human hematologic malignancy cell lines, including MYC-driven DLBCL, double-hit or triple-hit lymphoma (DTHL), and Burkitt’s lymphoma. Hierarchical clustering of the top 370 highly variable genes indicated that murine RT cells were similar to human ABC DLBCLs (Fig. 2L). In addition, murine CLL cells were also found to be in distinct clusters and clustered away from human CLL cells, with a gene expression signature distinct from nontumor cells (Fig. 2L). Together, our findings show that murine RT cells resemble human RT cells with DLBCL gene expression signature, highlighting that our murine RT model is faithful to human disease.

RT cells have aberrant mitochondrial structural changes, increased OXPHOS, and metabolic reprogramming

Our gene expression analysis suggests that MYC, OXPHOS, and glycolysis pathways are enriched in RT cells. MYC is known to promote cell growth by regulating OXPHOS (24, 25). The increased MYC expression in RT cells and their large size and aggressive growth led to the hypothesis that MGA deletion drives CLL-to-RT transition through mitochondrial deregulation. Mitochondria are dynamic organelles, and alteration of respiration is usually coupled with aberrant ultrastructure, mitochondrial DNA (mtDNA), and reactive oxygen species (ROS) (26). In particular, mitochondria undergo changes in shape under oxidative stress (26, 27). To evaluate mitochondrial structural alterations in RT, we performed electron microscopy–based morphometric analysis on murine splenic B cells. A greater percentage of mitochondria in the CLL group had aberrant mitochondria structures (P < 0.001), including broken cristae with reduced width (Fig. 3, A to C, and fig. S7A). RT cell mitochondria also exhibited larger cristae area and wider cristae (P < 0.001) (Fig. 3, A to C, and fig. S7A), indicating altered OXPHOS capability. Consistent with these morphological changes, we also detected increased expression of OPA1 and DRP1, two essential proteins regulating mitochondria fusion and fission (28), in both mouse (CLL and RT) and human (CLL) cells (Fig. 3, D and E, and fig. S7, B to D).

Fig. 3. Mga deletion leads to increased mitochondrial dysregulation and abnormal cristae in murine CLL-RT.

(A) Representative electron micrographs of mouse splenic B cells from control (no disease), CLL, and RT mice. Scale bars, 1000 nm. Black and blue dotted lines indicate normal and abnormally shaped mitochondria. Black and red arrows show normal and abnormal cristae structures. (B and C) Graph showing the quantification of cristae width (B) and cristae area (C) in murine control, CLL, and RT B cells. (D) Immunoblot showing the changes in mitochondrial fusion protein–OPA1 (L-OPA1 long isoform and S-OPA1 short isoforms) and fission protein-DRP1 in murine control, CLL (derived from original CLL1 and CLL2) and RT (derived from original RT1 engrafted into independent NSG mice). (E) Graph showing the quantification of OPA1, DRP1 in murine control, CLL, and RT B cells. (F to H) Mitomass, cellular RO S, and mitochondrial RO S of splenic B cells derived from control (no disease, Cd19-Cre^+/−), CLL (Cd19-Cre^+/− Sf3b1 ^fl/+Mdr^fl/+), and RT mice determined by MitoDeepred-, CellRO X Orange–, and MitoSOX-based flow cytometry, respectively (n = 3 per group). MFI, mean fluorescence intensity. ****P < 0.0001, ***P < 0.001, and *P < 0.05; two-way ANOVA test; Šídák corrected.

The intact mitochondrial structure is essential to mitochondrial function and cellular respiration. Given the distinct mitochondria structural changes, we further evaluated the mitochondrial mass and oxidative stress (cellular and mitochondrial ROS) in splenic B cells derived from control, CLL, and RT mice using flow cytometry–based assays. Murine CLL and RT cells exhibited increased mitochondrial mass compared with normal control B cells, with CLL cells showing greater mitochondrial mass than RT cells (Fig. 3F). Similarly, mitochondrial ROS was higher in murine CLL and RT cells compared with control normal B cells (Fig. 3G). However, we did not see increased mitoROS in RT compared to CLL, which could potentially be due to decreased expression of antioxidant genes such as TNXRD2 in the mitochondria of CLL and increased expression of TFAM (which encodes a mitochondrial transcription factor) in CLL as compared to RT (fig. S7, E and F). Of note, both murine and human RT cells had up-regulated expression of GLUL (glutamine ammonia ligase) (fig. S7G), a glutamine synthetase that is transcriptionally regulated by MYC and supports nucleotide synthesis, amino acid transport, and TCA cycle (29). Cellular ROS was higher in RT cells compared with CLL and normal B cells (Fig. 3H and fig. S8). These data suggest that Mga deletion leads to metabolic alterations. Together, in addition to shared gene and protein expression with human RT cells, Mga-deleted murine RT cells display mitochondria ultrastructure alterations, enhanced OXPHOS, and metabolic reprogramming.

Knockout of MGA in human cell lines leads to abnormal cristae structure, hyperactive mitochondrial activity, and hyperproliferation, recapitulating features of murine RT cells

Given that our murine model strongly implicates mitochondrial function as altered in RT cells and this may be involved in the onset of RT, we investigated the molecular circuitry changes underlying MGA deletion–associated mitochondrial dysfunction in human B lymphoid cell lines. We knocked out (KO) MGA in several human cell lines [CLL, MEC1, and HG3; acute lymphoblastic leukemia, Nalm6E; DLBCL and OCI-LY3; mantle cell lymphoma (MCL), JEKO] by CRISPR-Cas9 technology to determine whether the MGA/MYC/NME1 axis can be recapitulated in human cells (fig. S9A). MGA KO Nalm6E and MEC1 cells exhibited a higher oxygen consumption rate (OCR) at baseline and over time in response to modulators of the mitochondrial electron transport chain (ETC) and OXPHOS (Fig. 4, A and B, and fig. S9B). MGA KO Nalm6E cells also showed dependency on all three energy sources measured by the Seahorse substrate oxidation test [glucose, UK5099, glutamine, BPTES, lipids, etomoxir (ETO)] (fig. S9, C and D), indicating the substrates underlying higher OXPHOS. Of note, these cells showed consistently higher steady-state levels of cellular ROS (P < 0.001) coupled with increased mtDNA copy number and amount of adenosine 3′-triphosphate (ATP) (Fig. 4, C to E, and fig. S9E) (P < 0.001). Mitochondrial membrane potential was also elevated upon MGA KO (fig. S9F), consistent with the hyperenergetic state observed in murine RT cells. These observations support that MGA deletion impairs mitochondrial quality control processes. Likewise, we also examined the impact of MGA deletion on mitochondria function in other B cell lines (HG3, OCI-LY3, and JEKO) and found that MGA deletion generated higher cellular ROS, and mitochondrial potential increased mtDNA copy number (P < 0.001) and ATP production (P < 0.01), although mitochondrial membrane potential was not significantly different (fig. S9, E to H). These results indicate that mitochondrial dysregulation is a general feature induced by MGA deletion.

Fig. 4. MGA deletion leads to increased OXPHOS and abnormal cristae structure.

(A and B) OCR was determined by Seahorse Mitostress analysis in Nalm6E-Cas9 (A) and MEC1-Cas9 (B) cells, with and without MGA. (C and D) Cellular RO S (C) and mtDNA (D) in Nalm6E and MEC1-Cas9 cells with or without MGA as determined by H₂D-CFDA flow-based and PCR assays, respectively. (E) Cellular ATP determined by CellTiter-Glo assay in Nalm6E-Cas9 and MEC1-Cas9 cells with or without MGA. Y axis shows ATP concentration (μM) **P < 0.01 and *P < 0.05; Student’s t test. (F) In vitro growth monitoring for 4 days by CC K8 assay in Nalm6E-Cas9 cells with or without MGA. Y axis shows absorbance at 450 nm. ****P < 0.0001, by two-way ANOVA; Šídák corrected. (G) In vivo bioluminescent imaging of NSG mice with engraftment of control (n = 5) and MGA KO (n = 5) Nalm6E cells at days 9, 17, and 34 postengraftments. All of the images were acquired at the same scale. (H) The survival of NSG mice xenografted with control and MGA KO Nalm6E cells was measured and expressed as a percentage of the initial number of mice in each group. **P < 0.01, log-rank test was used. (I to J) Expression of MYC and NME1 in Nalm6E-Cas9 and MEC1-Cas9 cells with different genetic lesions detected by immunoblotting. (K) Expression of MYC, NME1 in Nalm6E-Cas9 cells overexpressing MYC detected by immunoblotting. (L) OCR measured in Nalm6E cells (control and MGA KO) with NME1 KO (K and L) or MYC knockdown (M) by Seahorse Mitostress assay. Y axis represents the OCR, and the x axis represents the time of the experiment. (N) ETC proteins and TFAM in Nalm6E-Cas9 cells with different genetic lesions detected by immunoblotting.

To determine the consequence of the MGA deletion–induced oxidative dysregulation, we measured cell growth in vitro and engrafted cells with stable overexpression of luciferase into immunodeficient NSG mice. MGA KO consistently led to greater cell growth in both Nalm6E and MEC1 cells in vitro (Fig. 4F). Through real-time bioluminescent imaging monitoring over 34 days, we found that MGA KO accelerated cell proliferation in vivo (Fig. 4G). As a result, mice with engraftment of MGA KO cells had poor overall survival compared with mice with engraftment of control cells (Fig. 4, G and H, P = 0.001; log-rank test). Together, these results indicate that MGA deletion led to mitochondria structure change, higher OXPHOS and ROS, higher ATP amount, and cell hyperproliferation, confirming a solid linkage between mitochondrial dysregulation and growth advantage induced by MGA deletion.

MGA deletion modulates OXPHOS via MYC and NME1

To elucidate the molecular regulatory network underlying MGA deletion in driving mitochondrial dysregulation and cell growth, we generated NME1 KO Nalm6E cells with and without MGA. We examined the expression of MYC and NME1 at both RNA and protein levels. Loss of MGA increased MYC and NME1 protein expression but only changed mRNA expression by onefold (Fig. 4I and fig. S10A). We explored the effect of MGA KO in other B cell lines (MEC1 and HG3) and found that both proteins were distinctly up-regulated at the protein level (Fig. 4J), validating MYC and NME1 as targets of MGA. Furthermore, short hairpin RNA (shRNA) knockdown of MGA in different B cell lines (HG3, MEC1, and OCI-LY3) and human embryonic kidney (HEK) 293T cells all confirmed MYC-NME1 to be up-regulated (fig. S10, B to D). Of note, shRNA knockdown of MGA also increased mtDNA copy number (fig. S10E). Moreover, overexpression of MGA in HEK293T cells decreased NME1 expression in a dosage-dependent manner (fig. S10, F and G). We also performed siRNA knockdown to check the quick turnover of MGA on MYC and NME1 in different cell lines (Nalm6E, MEC1, HG3, and 293T). We observed increased MYC and NME1 protein expression after 48 hours of siRNA knockdown of MGA in all cell lines tested (fig. S11). Similarly, the OCR rate was higher in MGA siRNA knockdown cells than in control cells (fig. S11D). Together, these results indicate that the loss of MGA through CRISPR-Cas9 KO, shRNA knockdown, or siRNA knockdown led to the up-regulation of MYC and NME1, supporting NME1 as a target of MGA.

It is worth noting that MYC was up-regulated in Nalm6E cells upon NME1 deletion at both the RNA and protein levels, suggesting that NME1 is a negative regulator of MYC (Fig. 4I and fig. S12A). Similarly, in MEC1 and HG3 cells, NME1 KO increased MYC expression (Fig. 4J and fig. S12B). Although deletion of MGA or NME1 alone can increase MYC expression, double KO of both resulted in less MYC up-regulation across all three cell lines (Fig. 4, I and J, and fig. S12B), suggesting tight control of MYC expression. To further discern the regulation between MYC and NME1, we overexpressed MYC in Nalm6E, MEC1, and OCI-LY3 cells and found that NME1 was up-regulated in these cell lines, with or without MGA (Fig. 4K and fig. S12, C to F), confirming NME1 as a MYC target.

Increased MYC has been associated with proliferation-related pathways such as phosphorylated extracellular signal–regulated kinase (pERK) and mTORC (30, 31), and consistent with this, we detected increased pERK1/2 and mTORC1 signaling (p4E-BP1) in MEC1 and HG3 cells with MGA deletion (fig. S13A). Increased mTORC1 signaling was not accompanied by increased AKT signaling (fig. S13B). Deletion of MYC significantly (P < 0.0001) affected cell growth in all cells (fig. S13C), whereas NME1 KO affected cell growth in Nalm6E-MGA KO cells by CCK8 assay but not by cell counting (fig. S13, D and E). In MEC1 cells, NME1 KO affected cell growth as assessed by cell counting, indicating that mitochondrial respiration and cell growth could not be correlated in some cell lines (fig. S13F). The B cell line data were confirmed in human RT cell line U-RT1 cells, in which KO of MGA increased MYC expression and led to cell growth advantage (P < 0.001) along with increased cellular ROS (P < 0.01) (fig. S13, G to I), highlighting the conserved role of MGA in mitochondrial regulation across different developmental stages of B cells.

To dissect the relationship between MYC and NME in MGA deletion–mediated mitochondria dysregulation, we performed a Seahorse Mito stress assay on Nalm6E cells with or without NME1 or MYC. NME1 KO cells appeared to have an increased basal and maximal OCR compared with control cells, possibly owing to increased MYC in these cells (Fig. 4L). MGA KO cells showed a dependency on NME1 for OXPHOS (Fig. 4L). Concurrent KO of MGA and NME1 resulted in partial normalization of metabolic activity, consistent with our notion that NME1 is required for cell proliferation in MGA KO cells (Fig. 4L). In line with the role of MYC in promoting high OXPHOS in NME1 KO cells, control cells demonstrated reductions in basal and maximal OCRs upon MYC deletion, whereas MGA KO cells showed dependency on MYC for OXPHOS (Fig. 4M). These results reveal that MGA deletion shifts OXPHOS from a MYC-dependent mode to a MYC and NME1 codependent mode.

To determine the molecular basis for MYC/NME1 OXPHOS dependency in MGA KO cells, we examined the OXPHOS mitochondrial ETC complex and TFAM (transcription factor A, mitochondrial), a key regulator for mitochondrial DNA replication (32). Mitochondrial ETC complex is the machine used for cells to generate ATP from the oxidation of carbohydrates, fats, and proteins and is composed of five separate protein complexes [I (NDUFB8), II (SDHB), III (UQCRC2), IV (MTCOX1), and V (ATP5α)] (27). NME1 loss resulted in a substantial decrease in subunit I, whereas MGA KO cells increased subunit II (Fig. 4N). An increase in complex II in MGA KO Nalm6E and a decrease in complex I in NME1 KO cells were further confirmed using immunoblotting on fractionated mitochondrial protein isolates (fig. S13J). Concurrent deletions of MGA and NME1 led to a subunit II down-regulation and a subunit I reduction compared with control cells, which could form the basis of NME1-dependent OXPHOS in MGA KO cells (Fig. 4N). Moreover, we observed an NME1- and MYC-dependent TFAM expression pattern in MGA KO cells (Fig. 4N and fig. S13K). Of note, the increase of complex II was also seen in MEC1 cells with MGA KO (fig. S13L) and primary RT samples (fig. S14, A and B). This distinct mitochondrial ETC complex and TFAM expression pattern upon MYC or NME1 deletion corroborated our observation that both NME1 and MYC have roles in regulating mitochondrial OXPHOS in MGA KO cells.

AZ5576 and TTFA treatment provide therapeutic benefits for RT in vivo

The MYC and NME1 coregulation mode of OXPHOS upon MGA deletion provides RT cells a maximum potential for mitochondrial alterations; however, it also posits a challenge for treating RT disease. Thus, better treatment of RT is likely to be achieved through targeting mitochondrial dysregulation in MGA KO–induced disease. We tested inhibitors of ETC complex II [thenoyltrifluoroacetone (TTFA)] and CDK9 (AZ5576, targeting MYC) in the Nalm6E cell line and MEC1 (Fig. 5, A and B). TTFA and CDK9 inhibited the expression of complex II/succinate dehydrogenase, MYC, NME1, pERK1/2, and mTOR (both phospho and total 4E-BP1) pathways in Nalm6E and MEC1 cells (Fig. 5, A and B, and fig. S15, A to D), with decreased OCR in Nalm6E cell lines (fig. S15, E to H), suggesting that perturbation of mitochondrial dysregulation could affect cell proliferation.

Fig. 5. AZ5576 and TTFA prolong the survival of RT mice by inhibiting MYC and NME1 expression.

(A and B) MYC, p4E-BP1, and total 4E-BP1 expression detected by immunoblotting in Nalm6E-Cas9 cells treated with CDK9i AZ5576 (1 μM) (A) or complex II TTFA (100 μM) (B) overnight. (C) Schema for in vivo drug treatment experiments. (D) Kaplan-Meier survival curves of RT mice treated with AZ5576 or TTFA alone or in combination, log-rank test Bonferroni corrected. (E) Images of spleen and spleen weight (mg) from RT mice with different treatments collected at the end point. (F) Immunoblotting of MYC, NME1, p4E-BP1, and 4E-BP1 protein in splenic B cells derived from RT mice at the end point. (G) Protein quantification of MYC, NME1, and p4EBP1 in different groups with respect to loading control from (F), two-way ANOVA; Šídák corrected. ****P < 0.0001, ***P < 0.001, **P < 0.01, and *P < 0.05; ns, not significant.

We then explored whether these inhibitors provide therapeutic benefits for RT mice in vivo. We first engrafted RT cells into NSG mice, established stable RT mice in 2 weeks (fig. S16A), and then treated these mice with oral administration of vehicle (n = 8), AZ5576 (60 mg/kg) (n = 5), TTFA (25 mg/kg) (n = 8), or AZ + TTFA (60 and 25 mg/kg) (n = 8). AZ5576 was administered once a week, whereas TTFA was given 5 days a week (Fig. 5C). AZ5576 and TTFA treatment significantly prolonged the survival of RT mice (P = 0.0007) with a significant (P < 0.0001) reduction in total spleen weight. The AZ + TTFA combination was further effective in increasing the survival of RT mice by 10 days (P = 0.0003) and reducing spleen weight (P < 0.0001) compared with single treatments (Fig. 5, D and E). AZ5576- and TTFA-treated RT mice B cells had reduced MYC, NME1, SDHB, SDHA (complex II), NDUFB8 (complex I), and AT-P5A (complex V) at the RNA level (fig. S16B). RT cells derived from mice with AZ5576 and TTFA treatment had reduced MYC, NME1, and phospho-4E-BP1 T37/46 (mTOR pathway) protein expression (Fig. 5, F and G), confirming the effective targeting of the MGA/MYC/NME1 axis in vivo. Together, concurrent targeting of MYC and OXPHOS provides enhances treatment efficacy for RT in vivo, highlighting the MGA/MYC/NME1 regulatory axis as a target for RT treatment (fig. S16C).

DISCUSSION

Here, we established an RT murine model faithful to human genetics by silencing Mga in an existing CLL model. Through extensive characterization of this murine RT model and functional experiments of human cell lines with MGA deletion, we found that the MGA-NME1 axis drives RT through mitochondrial OXPHOS up-regulation and that concurrent targeting of MYC and OXPHOS pathways provides therapeutic benefits for RT mice in vivo, suggesting a therapeutic avenue for patients with RT.

Genetic manipulation recently established several RT models based on an aggressive Eμ-TCL1–transgenic CLL mouse model. These models include B cell–specific deletion of TP53/CDKN2A/CDKN2B; deletion of NFAT2 (characterized by down-regulation of CDKN2A or TP53); and overexpression of activated AKT, activated NOTCH1, or c-MYC to develop aggressive lymphomas with DLBCL morphology (9–11). During our manuscript revision, RT models were also established using a similar method as ours, in which they simultaneously silenced Tp53/Mga/Chd2, and these combinations resulted in the onset of RT with high penetrance (9). Of note, RT mice based on either Eu-TCL1 transgene or codeletion of Tp53/Mga/Chd2 all had activated phosphatidylinositol 3-kinase (PI3K) pathway; however, our RT mice showed no PI3K pathway activation but mTOR activation, suggesting that our model may represent another subtype of RT. A recent study clustered patients with RT into five subtypes on the basis of recurrent genetic events, in which del(13q) and del(15q) (MGA gene resides) co-occur in the RS3 subtype, and del(13q), mutations in SF3B1, and MGA also co-occur in patients with subtype RS4. Our model thus may genetically recapitulate these subtypes of human RS, which could have different cellular circuitry changes from other subtypes. Nonetheless, our model provides a tool to study disease biology and serves as a pre-clinical model for RT.

Mga deletion–induced mitochondrial ROS accumulation and mitochondrial alterations underlie the CLL-to-RT transition. In CLL, mitochondria are dysregulated, with increased OXPHOS as the vital source of ROS (33). However, until now, the underlying mechanisms of oxidative stress in CLL and their contributions to RT onset were elusive. Our results support that the MGA/MYC/NME1 axis induces extensive mitochondria structure changes, increased OXPHOS, and ROS accumulation through up-regulation of ETC complex II and TFAM that eventually leads to activation of mTOR pathway and ERK pathway to contribute to cell proliferation. Consistently, a recent longitudinal study of CLL to RT samples (n = 54) identified a (OXPHOS)^high B cell receptor (BCR)^low transcriptional signature and showed OXPHOS as a potential therapeutic target in RT (15). Also, a recent multi-omics analysis of primary CLL samples suggests that the mTOR-MYC-OXPHOS axis is significantly associated with an aggressive type of CLL (34). Our murine model and cell line characterization recapitulates the observation of MYC-OXPHOS-mTOR activation in human hyperproliferative CLL and further suggests that this axis underlies CLL-to-RT transformation.

Our results further suggest that NME1 has a critical role in mediating mitochondrial alterations upon MGA deletion. We found that NME1 is a negative regulator of MYC and a direct target of MYC in cells without MGA deletion. The codependent mode on MYC and NME1 in MGA KO cells maximizes the cell’s potential to tolerate ROS accumulation and activate downstream oncogenic signaling, promoting cell growth via modulating mitochondria function. How NME1 interacts with mitochondria is not well explored. NMEs directly interact with and channel guanosine 5′-triphosphate (GTP) to specific dynamin-related GTPases to drive membrane remodeling and mitochondrial fusion (35, 36). It has been reported that long-chain fatty acyl coenzyme A inhibits NME1/2 (37). Our results also indicated that NME1 regulates a few mitochondrial genes, suggesting that NME1 may not directly interact with mitochondria. Future molecular studies and biochemistry assays are anticipated to determine the mechanism of how NME1 exerts its impact on mitochondria.

Our study has several limitations. First, the RT murine model we developed was only genetically faithful to a subset of patients with RT carrying MGA mutations, making the findings potentially nonge-neralizable to RTs with different genetic makeup. Further, the link between mitochondrial OXPHOS and cell proliferation might not directly correlate across all cell lines, requiring additional research. In addition, although the combination of CDK9i and TTFA improved the overall survival of RT mice, mice ultimately succumbed to the disease, indicating the need to identify more effective targets. Given that NME1 lacks commercial inhibitors, discovering direct inhibitors of NME1 could be more beneficial, in particular when combined with CDK9 or complex II inhibitors.

Targeting the MGA/MYC/NME1 axis provides therapeutic benefits for RT. Our murine RT model allowed us to explore new treatment options for this aggressive disease. Although inhibitors for either AZ5576 or TTFA can improve the survival of RT mice in vivo, combination treatment showed enhanced efficacy, suggesting that cellular metabolic function is a tractable target in RT.

MATERIALS AND METHODS

Study design

The goal of this study was to evaluate the role of Mga in CLL and RT. To achieve this, we silenced Mga using CRISPR-Cas9 technology in murine B cells with 13q (Mdr) and Sf3b1 K700E, two of the most recurrently genetic lesions in human CLL. We transduced progenitor cells (LSK) derived from Cd19-Cre^+/−Sf3b1 ^fl/+Mdr ^fl/+ Cas9 ^fl/+ (CD45.2) mice with sgRNAs against Mga or control, engrafted them in sublethally radiated recipient mice (CD45.1), and followed the mice for the development of the disease by flow cytometry for 22 to 23 months. The splenic B cells from the CLL mice were serially transplanted in either CD45.1 C57BL/6 or NSG mice, and the development of CLL/RT was followed by flow cytometry analysis. We performed morphological and immunohistochemical staining on spleen, bone marrow, and liver sections to verify murine CLL and RT, in addition to splenic cell size analysis by flow cytometry. To understand the molecular mechanism underlying Mga-driven CLL-to-RT, we performed transcriptome analysis using RNA derived from murine splenic CLL and RT samples (n = 2) and normal B cells from wild-type mice (n = 4). Cellular ROS and mitochondrial structure, ROS, volume, and DNA copy number were analyzed in murine CLL and RT splenic B cells by electron microscopy, flow cytometry, and PCR-based methods. To evaluate the function of MGA in human B cell lines [NALM6 (pre-B ALL), MEC1, HG3 (CLL), OCI-LY3 (DLBCL), Jeko1 (MCL), and U-RT1 (RT)], we performed MGA deletion via CRISPR-Cas9, knockdown using shMGA controls/siRNA controls in different cell lines. We validated downstream MGA, MYC, and NME1 targets in human CLL-RT cell lines and mouse CLL-RT using immunoblot and quantitative PCR–based assays. We conducted phenotypic experiments such as growth in vitro and in vivo and mitochondrial functional assays. For in vivo treatment experiments, we injected RT cells into NSG mice via tail vein injection and randomized the mice for CDK9i (AZ5576), complex IIi (TTFA), and both single and combination treatments. For in vivo experiments, investigators were blinded to treatment allocations.

Animals

All animals were housed at the City of Hope National Medical Center (COH). The animals were housed in groups (n = 5), following a standard 12-hour light/dark cycle. They were provided with auto-claved tap water and sterile pellets available ad libitum. All animal procedures were completed in accordance with the guidelines for the Care and Use of Laboratory Animals. All protocols were approved by the Institutional Animal Care and Use Committees at COH. To obtain heterozygous expression of Sf3b1 mutations and Mdr deletion in B cells, we crossed Sf3b1-K700E floxed mice (38) with Mdr floxed mice (39) to generate Sf3b1^fl/+Mdr^fl/fl mice, which were then crossed with CD19Cre (Cd19-Cre^+/+) to obtain double-mutant mice (Cd19-Cre^+/−Sf3b1 ^fl/+Mdr^fl/+). The double-mutant mice were further crossed with CRISPR-Cas9 green fluorescent protein mice (19) to obtain (Cd19-Cre^+/−Sf3b1 ^fl/+Mdr^fl/+ Cas9 ^fl/+).

Treatment in vivo

In murine RT cell drug experiments, 1 million cells in 100 to 150 μl of phosphate-buffered saline (PBS) were injected into NSG mice across four groups, vehicle (n = 8), AZ5576 (n = 8), TTFA (n = 8), and combined AZ5576 + TTFA (n = 8). AZ5576, in B-cyclodextrin, was given orally at 60 mg/kg biweekly. TTFA, in 2% dimethyl sulfoxide/30% polyethylene glycol 300, was given at 25 mg/kg, 5 days weekly. Treatments began 7 days after RT cell introduction. Euthanasia was based on criteria such as hunched back, lethargic movement, breathing difficulties, and leg paresis.

Mouse serial transplantation

Mouse transplantation studies were conducted on 6- to 10-week-old immunocompetent CD45.1 C57BL/6 (sublethally irradiated with 400 cGy) or immunodeficient NSG mice using viably cryopre-served splenic cells from CLL animals. One million CLL cells were resuspended in 100 to 150 μl of PBS and injected intravenously into CD45.1 C57BL/6 or NSG mice. We named three original CLLs: CLL1, CLL2, and CLL3. The subsequent RTs derived from these CLLs are RT1, RT2, and RT3. We usually engrafted each RT mouse into up to five NSG mice to serve as technical replicates. Because RT cells derived from RT1, RT2, and RT3 have similar growth patterns and molecular changes, unless mentioned in the legend, we used RT cells derived from RT1 for experiments. For RT mice, 1 million cells were resuspended in 100 to 150 μl of PBS and injected intravenously into NSG mice. CLL/RT burden in the peripheral blood was monitored by flow cytometry. The criteria for euthanasia were hunched back, lethargic movement, breathing difficulties, and leg paresis.

Statistical analysis

Statistical analysis was performed using GraphPad Prism 9.3.1. For two sample groups, means were compared by unpaired two-tailed Student’s t test for normally distributed data. For multiple group comparisons, P values were calculated using a one-way or two-way analysis of variance (ANOVA) test followed by a Dunnett, Tukey, and Šidák post hoc multiple comparison test. A P value <0.05 was considered statistically significant. Statistical significance for differential gene expression was adjusted with Benjamin Hochberg correction at a 5% level. The type of statistical test used and the results, including P values, means, medians, and SEs, are shown in the figures and figure legends.

Supplementary Material

Supplementary Materials

This PDF file includes:

Materials and Methods

Figs. S1 to S16

Table S1

References (40–46)

Other Supplementary Material for this manuscript includes the following:

Data file S1

MDAR Reproducibility Checklist

Data and materials availability:

All data associated with this study are available in the paper or the Supplementary Materials. RNA-seq data related to murine CLL and RT samples are available in NCBI GEO under accession GSE263238. Further information and requests for resources and reagents should be directed to the corresponding author. Raw data from figures are available in data file S1.