Research ArticleCancer

Metabolic reprogramming toward oxidative phosphorylation identifies a therapeutic target for mantle cell lymphoma

See allHide authors and affiliations

Science Translational Medicine  08 May 2019:
Vol. 11, Issue 491, eaau1167
DOI: 10.1126/scitranslmed.aau1167

Dismantling lymphoma metabolism

Mantle cell lymphoma is a B cell malignancy that often responds to initial treatment with ibrutinib, an inhibitor of Bruton’s tyrosine kinase. Unfortunately, the therapeutic response is typically short lived for reasons that are not yet fully understood. Zhang et al. found that resistance to ibrutinib in mantle cell lymphoma can be associated with metabolic reprogramming and a shift toward reliance on glutaminolysis and oxidative phosphorylation by the cancer cells. The authors demonstrated that these drug-resistant cells can be effectively targeted with a small-molecule inhibitor of oxidative phosphorylation, showing promising therapeutic results in patient-derived mouse models.

Abstract

Metabolic reprogramming is linked to cancer cell growth and proliferation, metastasis, and therapeutic resistance in a multitude of cancers. Targeting dysregulated metabolic pathways to overcome resistance, an urgent clinical need in all relapsed/refractory cancers, remains difficult. Through genomic analyses of clinical specimens, we show that metabolic reprogramming toward oxidative phosphorylation (OXPHOS) and glutaminolysis is associated with therapeutic resistance to the Bruton’s tyrosine kinase inhibitor ibrutinib in mantle cell lymphoma (MCL), a B cell lymphoma subtype with poor clinical outcomes. Inhibition of OXPHOS with a clinically applicable small molecule, IACS-010759, which targets complex I of the mitochondrial electron transport chain, results in marked growth inhibition in vitro and in vivo in ibrutinib-resistant patient-derived cancer models. This work suggests that targeting metabolic pathways to subvert therapeutic resistance is a clinically viable approach to treat highly refractory malignancies.

INTRODUCTION

Mantle cell lymphoma (MCL) is a B cell lymphoma subtype that comprises 6 to 8% of non-Hodgkin’s lymphoma (1, 2). B cell receptor (BCR) pathway activation is a hallmark of B cell lymphomas, including MCL (3, 4). Bruton’s tyrosine kinase (BTK), a crucial component of the BCR pathway, is highly expressed in MCL cells, including its active phosphorylated form (5). Ibrutinib, a first-in-class oral covalent inhibitor of BTK, was approved by the U.S. Food and Drug Administration in 2013 to treat relapsed/refractory MCL, and this kinase inhibitor has demonstrated antitumor activity with an overall response rate of 68% in relapsed/refractory MCL and a median duration of response of about 18 months (6). However, the 1-year survival rate is 22% after relapse on ibrutinib (7), prompting an urgent need to identify alternative therapeutic options that will benefit this patient population.

The mechanisms mediating ibrutinib resistance have recently been explored. A relapse-specific C481S mutation in the BTK gene and the PLCG2 mutations associated with acquired ibrutinib resistance have both been identified in multiple chronic lymphocytic leukemia clinical specimens, suggesting that the potential alterations of the BCR-BTK pathway likely play a role in ibrutinib resistance (8, 9). However, BTK and the PLCG2 mutations are infrequent in patients with MCL. On the other hand, the activation of the phosphatidylinositol 3-kinase–AKT–mammalian target of rapamycin (PI3K-AKT-mTOR) pathway and alternative nuclear factor κB (NF-κB) signaling have been implicated as alternative survival mechanisms that override the effects of ibrutinib, initiating resistance (1012). Nevertheless, the mechanisms underlying ibrutinib resistance are most likely complicated, and studies fully elucidating the intricate networks responsible for therapeutic resistance remain lacking.

Although inhibition of the PI3K-AKT-mTOR pathway in preclinical models of MCL has shown strong antitumor activity in vitro (13, 14), this inhibition has not been meaningfully effective in vivo using MCL patient-derived xenograft (PDX) mouse models (12, 15). PI3K inhibitors such as CAL-101 (idelalisib) (16) and several rapamycin analogs (everolimus and temsirolimus) (1721) have resulted in underwhelming clinical outcomes in MCL and many other cancers, suggesting that new inhibitors are needed or that alternative therapeutic targets should be explored and evaluated.

Here, we demonstrate that metabolic reprogramming plays a critical role in ibrutinib-resistant MCL cells, producing a reliance on oxidative phosphorylation (OXPHOS) and glutaminolysis for growth and survival. Furthermore, quenching OXPHOS energy production using a small-molecule inhibitor (IACS-010759) (22) targeting complex I of the mitochondrial electron transport chain (ETC) results in reduced proliferation and increased apoptosis. Overall, our work indicates an opportunity for the clinical treatment of ibrutinib-resistant MCL, and these findings are currently being evaluated in a phase 1 lymphoma clinical trial (NCT03291938).

RESULTS

Whole-exome sequencing reveals the landscape of somatic mutations and DNA copy number alterations in MCL

To decipher the mechanisms underlying ibrutinib resistance, clinical specimens were collected from 37 ibrutinib-treated patients with MCL (fig. S1A). The clinical and pathological information is presented in data file S1. MCL sensitivity to ibrutinib was significantly associated with clinical outcomes, with the ibrutinib-resistant patients with MCL experiencing poorer progression-free survival (PFS) and overall survival (OS) (both PFS and OS, P < 0.0001; fig. S1, B and C).

Of those 37 patients with MCL, whole-exome sequencing (WES) was conducted on the clinical specimens of 14 patients that had sufficient isolated tumor DNA, including 7 ibrutinib-sensitive and 7 ibrutinib-resistant patients (fig. S1A). A corresponding peripheral blood sample was also sequenced and used as the matched germline control for somatic mutation and DNA copy number analyses. For tumors for which a matching germline control was not available from the same patient, pooled sex-matched peripheral blood controls were used for mutation and copy number calling.

Frequent inactivating somatic alterations in ATM, KMT2D, and TP53 were observed in both the ibrutinib-sensitive and ibrutinib-resistant tumors (Fig. 1A). CDKN2A and MTAP (five of seven, 71%) were commonly codeleted in the ibrutinib-resistant tumors (P = 0.010, denoted by an asterisk). Similarly, mutation and amplification of CCND1 were identified only in the ibrutinib-resistant tumors. DNA copy number analysis identified extensive copy number alterations, especially copy number losses in the ibrutinib-resistant tumors, including broad deletions of 6q, 9p, and chromosome 13 (Fig. 1B). The burden of copy number gains and losses was markedly increased in the ibrutinib-resistant tumors (Fig. 1C), and the increased burden of copy number changes was unlikely an artifact caused by differences in the tumor cellularity between samples in the two groups. We observed no difference in the pathological tumor purity (indicated by the percentage of CD19+ cells) between the two groups (fig. S2A). In accordance, no statistical difference was observed in the distribution of tumor variant allelic fractions of all the somatic mutations identified from frequently mutated genes between the two groups (fig. S2B). Furthermore, no mutations in the BCR pathway (BTK and PLCG2) or alternative NF-κB signaling pathway (NF-κB–inducing kinase pathway) were found. Therefore, the increased aneuploidy observed in the resistant group is most likely biologically meaningful and may have played a role in driving tumor progression during the course of treatment with ibrutinib.

Fig. 1 Landscape of somatic mutations and DNA copy number alterations in 14 patients with MCL treated with ibrutinib.

(A) Somatic mutations, where each column represents a patient tumor sample, and the clinical and pathological characteristics are annotated at the top. Genes with nonsynonymous mutations or copy number alterations in two or more patients are listed. The numbers on the left and right sides represent the percentages of MCL tumors carrying a mutation or copy number alteration of each specific gene in the ibrutinib-sensitive and ibrutinib-resistant groups, respectively. Mantle Cell Lymphoma International Prognostic Index (MIPI) score: 0 to 3, low risk; 4 to 5, intermediate risk; 6 to 11, high risk. CR, complete response; PR, partial response; PD, progressive disease; LOH, loss of heterozygosity. The asterisk at CDKN2A and MTAP denotes a statistically significant difference in the deletion of CDKN2A and MTAP in the ibrutinib-resistant cohort compared with the ibrutinib-sensitive cohort (P = 0.010). (B) Somatic copy number alterations in ibrutinib-resistant (top) and ibrutinib-sensitive (bottom) tumors. The chromosome numbers are labeled at the top, and the sample IDs are shown on the left. Blue indicates copy number loss, and red indicates copy number gain, where the intensity corresponds to the log2 ratio of each segment. (C) Copy number gains and losses in ibrutinib-sensitive (blue) and ibrutinib-resistant (pink) groups. The boxes in the box plot represent the interquartile range (IQR), where the centerline depicts the median. The upper whisker indicates the maximum value or 75th percentile +1.5 IQR, whichever is smaller; the lower whisker indicates the minimum value or 25th percentile −1.5 IQR, whichever is greater.

Transcriptomic profiling identifies metabolic reprogramming as a hallmark of ibrutinib resistance

Whole-transcriptome RNA sequencing (RNA-seq) was performed on clinical specimens isolated from 15 ibrutinib-sensitive and 6 ibrutinib-resistant primary MCL samples (fig. S1A). Unsupervised hierarchical clustering of MCL tumors using RNA-seq gene expression data showed a response-specific gene expression signature (Fig. 2A). A total of 63 protein-coding genes were identified as the most differentially expressed genes (DEGs) between the ibrutinib-resistant and ibrutinib-sensitive groups, with a fold change of ≥2 or ≤−2 and a false discovery rate (FDR q value) of ≤0.01. Among the DEGs, 26 genes were up-regulated in ibrutinib-resistant tumors (Fig. 2B). Computational overlapping of those 26 up-regulated genes with the Molecular Signatures Database (MSigDB; Broad Institute) hallmark gene sets suggested a significant enrichment of genes in mTOR signaling (7 of 26, FDR q value = 5.98 × 10−10), cell cycle regulation (5 of 26, FDR q value = 1.84 × 10−6), and MYC targets (3 of 26, FDR q value = 2.9 × 10−3). To further confirm the applicability of these 63 DEGs in delineating ibrutinib sensitivity from resistance, we designed a nanoString nCounter gene expression panel composed of these 63 DEGS and further examined their expression pattern across an additional, independent MCL cohort, including ibrutinib-sensitive (n = 8) and ibrutinib-resistant (n = 7) patients (table S1). Many of these DEGs were differentially expressed between the ibrutinib-sensitive and ibrutinib-resistant groups, and 25 DEGs were statistically significant, with a fold change of ≥2 or ≤−2 and an FDR q value of ≤0.01 (fig. S3), indicating that this expression pattern is consistent across different ibrutinib-sensitive and ibrutinib-resistant MCL samples.

Fig. 2 DEGs between the ibrutinib-sensitive and ibrutinib-resistant tumors.

(A) Unsupervised clustering showing the most DEGs between the ibrutinib-sensitive and ibrutinib-resistant tumors. (B) A cutoff fold change of ≥2 or ≤−2 and an FDR q value of ≤0.01 were applied, and only genes that met these criteria were selected for unsupervised clustering in (A) and labeled in the volcano plot. (C) Box plots showing two representative DEGs of metabolite transporters SLC16A1 and SLC1A5. (D) Immunoblotting showing the differential expression of the metabolite transporters SLC16A1 and SLC1A5 in ibrutinib-resistant MCL cell lines Maver-1, Z-138, Granta-519, and three MCL patient samples (PT4 to PT6) compared with the ibrutinib-sensitive MCL cell lines JeKo-1, Mino, SP-49, and three MCL patient samples (PT1 to PT3). To quantitate the proteins, all bands were first normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) by area using ImageJ software. A ratio was obtained for each band. The band in the first lane was treated as 1, and a fold change for the other bands compared with the first value was calculated to represent the relative protein expression. Ibrutinib-S/R, ibrutinib-sensitive/resistant.

Manual inspection of the remaining up-regulated DEGs against currently available knowledge bases suggested that many of the up-regulated DEGs are metabolism related (labeled with asterisks in Fig. 2B), including SLC16A1, SLC1A5 (Fig. 2C), SLC25A19, and SLC26A. In particular, SLC25A19 is a mitochondrial transporter that mediates the uptake of thiamine pyrophosphate and is also a cofactor for several dehydrogenase enzyme reactions, including pyruvate dehydrogenase [linking glycolysis to the tricarboxylate acid (TCA) cycle], alpha-ketoglutarate (α-KG; linking glutamate to the TCA cycle), transketolase (HMP shunt), and branched-chain ketoacid dehydrogenase. SLC16A1 encodes a proton-coupled monocarboxylate transporter (also known as MCT-1), which is an essential transporter of lactate and pyruvate up-regulated in many solid tumors (23) and maintains the metabolic phenotype of tumor cells (24). SLC1A5, also known as ASCT2, is a heavily studied glutamine-specific transporter regulated by c-MYC (25). In support of these findings, the 63-gene nanoString assay also showed an up-regulation of SLC16A1, SLC1A5, and SLC25A19 in the independent cohort of 15 ibrutinib-sensitive and ibrutinib-resistant MCL samples (fig. S3). Consistent with the RNA-seq and nanoString nCounter analysis, ibrutinib-resistant MCL cells (Maver-1, Z-138, and Granta-519) and primary MCL cells freshly isolated from three patients [patient (PT) 4 to PT6] (11) also showed increased SLC16A1 and SLC1A5 at the protein level compared with the ibrutinib-sensitive cells (SP49, Mino, and JeKo-1) and primary MCL cells (PT1 to PT3) (Fig. 2D).

We then performed gene set enrichment analysis (GSEA) to determine whether an a priori defined set of genes showed statistically significant and concordant differences between the ibrutinib-sensitive and ibrutinib-resistant tumors. As shown in Fig. 3A, oncogenic pathways including c-MYC, mTOR (mTORC1), Wnt, and NF-κB signaling, followed by cell cycle, apoptosis, BCR signaling, and DNA repair, were among the most markedly enriched pathways in the ibrutinib-resistant tumors, some of which were also confirmed by the nanoString nCounter analysis of additional clinical specimens. For example, in the nanoString nCounter analysis, SMC1B, FGFR1, and CDK2 were among the top 20 DEGs and belonged to the cell cycle control, MAPK, and PI3K pathways, respectively (fig. S4A). In addition, ibrutinib-resistant samples had higher expressions of PI3K and NOTCH pathway genes and lower expressions of Janus kinase/signal transducers and activators of transcription (JAK/STAT) pathway genes as compared with the sensitive samples. The DNA damage repair pathway was also significantly up-regulated, as demonstrated by both the RNA-seq and nanoString data (P = 0.031; Fig. 3A and fig. S4B).

Fig. 3 Signaling pathways driving ibrutinib resistance in patients with MCL.

(A) The top significantly enriched signaling pathways in ibrutinib-resistant (versus ibrutinib-sensitive) tumors by GSEA pathway enrichment analysis. The normalized enrichment score (NES) reflects the extent of enrichment and allows comparison across gene sets. Listed pathways are ranked by their NES and colored by their type classification. The FDR q values are labeled on the right. (B) Representative enrichment plots for the hallmark OXPHOS pathway, mTORC1 signaling, and MYC and E2F targets. The top portion of each plot shows the running enrichment score (ES) for the gene set as the analysis walks down the ranked list of genes. ES reflects the degree to which a gene set is overrepresented at the top (positive ES) or bottom (negative ES) of a ranked list of genes. The score at the peak of the plot (the score furthest from 0.0) is the ES for the gene set. The black vertical line at the bottom shows where the members of the gene set appear in the ranked list of genes. The graded red to blue bars on the x axis represent the DESeq2 statistical values (resistant group versus sensitive group), with red and blue denoting up-regulation and down-regulation, respectively.

The expression of genes involved in metabolic pathways such as OXPHOS was significantly enriched in the ibrutinib-resistant tumors (NES > 3 and FDR q value < 1× 10−5) in addition to the noted oncogenic pathways. The representative GSEA enrichment plots of OXPHOS, mTORC1 signaling, MYC targets, and E2F targets are shown in Fig. 3B. The up-regulation of c-MYC targets and mTORC1 signaling components is consistent with the increased activity of OXPHOS because these oncogenic pathways are directly linked to metabolic reprogramming in cancer cells (25). Glutaminolysis fuels OXPHOS in specific tumor cells and different contexts (26), supporting the hypothesis that ibrutinib-resistant MCL cells rely on oxidative glutamine metabolism for energy production and biosynthesis (25).

Ibrutinib-resistant MCL cells rely onglutamine-fueled OXPHOS

In support of the hypothesis that ibrutinib-resistant MCL cells rely on glutamine-fueled OXPHOS, ibrutinib-resistant MCL cell lines (table S2) displayed higher basal, adenosine 5′-triphosphate (ATP)–coupled, and reserve oxygen consumption rates (OCRs) (Fig. 4A), indicators of OXPHOS activity, in comparison to ibrutinib-sensitive MCL cell lines. Furthermore, the ratios of OCR to ECAR (extracellular acidification rate), which suggest preference for OXPHOS versus glycolysis, were examined in both ibrutinib-sensitive and ibrutinib-resistant MCL cell lines and primary MCL cells freshly isolated from patients. Both the ibrutinib-resistant MCL cell lines and primary MCL clinical specimens (n = 3) demonstrated higher OCR:ECAR ratio values, indicating that ibrutinib-resistant MCL primary clinical specimens also rely on glutamine-fueled OXPHOS (Fig. 4B). Targeted metabolomics analysis showed significantly higher cellular abundance of the metabolite α-KG in the resistant cells [fold change, 2.9698 and P = 0.0196 (two-sample t test); Fig. 4C], which was also confirmed in a second targeted metabolomics analysis of additional cell lines (two-sample t test, P = 0.0153; fig. S5), ultimately suggesting enhanced glutamine metabolism. Incorporation of α-KG into the TCA cycle is a major anaplerotic step in proliferating cells and is critical for the maintenance of TCA cycle function (27). In corroboration with our observation, glutaminase (GLS), the enzyme that converts glutamine to glutamate, a precursor of α-KG, was highly expressed in ibrutinib-resistant MCL [fold change, 1.2094 and P < 0.0001 (two-sample t test); Fig. 4D], which was supported by immunoblotting of ibrutinib-resistant and ibrutinib-sensitive MCL cell lines and primary MCL clinical specimens (Fig. 4E). Glutamine uptake was also up-regulated in the ibrutinib-resistant MCL cell lines (fig. S6A), and glutamine deprivation (Fig. 4F) or blockade of glutamine metabolism with aminooxyacetate (fig. S6B), a glutaminolysis inhibitor, resulted in marked induction of reactive oxygen species (ROS) and energy stress compared with the sensitive cell lines. Furthermore, glutamine metabolism inhibition with the allosteric GLS1-selective inhibitor bis-2-(5-phenylacetamido-1,3,4-thiadiazol-2-yl)ethyl sulfide (BPTES) markedly reduced the OCR of the ibrutinib-resistant cells, indicating that glutamine is the primary substrate for OXPHOS (fig. S6C). Last, we conferred ibrutinib resistance in the ibrutinib-sensitive JeKo-1 cell line using CRISPR-Cas9 to alter BTK, the target of ibrutinib (28). In the two genetically manipulated clones, deletions were introduced in one BTK allele, leaving the other allele intact. Because BTK is X linked and the JeKo-1 cells were derived from a female patient, the inactivated X chromosome most likely remained unaffected by the guide RNA and remained a wild type, with very little BTK protein production from this X-inactivated allele, creating a BTK knockdown (KD) model (fig. S7A) that is resistant to ibrutinib. Proteomics analysis of the JeKo-BTK KD clones compared with the JeKo-1 parental line showed increased expression of OXPHOS-associated proteins (fig. S7B), suggesting that ibrutinib resistance is associated with increased OXPHOS. Our results demonstrate that up-regulated mitochondrial OXPHOS energy production may be the primary mechanism that confers ibrutinib resistance in MCL.

Fig. 4 The role of OXPHOS in the energy production of ibrutinib-resistant MCL cells.

(A) The basal, ATP-coupled, and reserve OCR in ibrutinib-sensitive (JeKo-1 and Mino) and ibrutinib-resistant (Z-138 and Maver-1) MCL cell lines (ibrutinib-sensitive comparison versus ibrutinib-resistant comparison: basal OCR, P < 0.0001; ATP-coupled OCR, P = 0.0059; reserve OCR, P = 0.0001, linear regression model; n = 3 biological replicates; means ± SD). (B) OCR:ECAR ratios calculated for MCL cell lines [from left to right: (ibrutinib sensitive) JeKo-1, Mino, and Rec-1 and (ibrutinib resistant) JVM-13, Z-138, and Maver-1; P < 0.0001, linear regression model] and MCL primary clinical specimens (n = 3 sensitive and 3 resistant; P < 0.0001, linear regression model). Each color represents a cell line (JeKo-1, blue; Mino, green; Rec-1, maroon; JVM-13, brown; Z-138, purple; and Maver-1, red) or clinical sample from a distinct patient (each color represents a different patient). (C) Relative abundance of metabolites extracted from ibrutinib-sensitive and ibrutinib-resistant MCL cell lines. Each row represents a single metabolite, and each column represents the indicated MCL cell line. (D) Reverse phase protein array (RPPA) analysis of ibrutinib-resistant MCL cell lines (Z-138 and Maver-1 in triplicate) versus ibrutinib-sensitive MCL cell lines (Rec-1 and Mino in triplicate) was used to generate the depicted volcano plot. A cutoff fold change of ≥2 or ≤−2 and an FDR q value of ≤0.01 were applied, and only proteins that met these criteria were selected for unsupervised clustering. (E) Immunoblotting showing GLS protein amounts in the indicated ibrutinib-sensitive and ibrutinib-resistant MCL cell lines and primary MCL cells freshly isolated from patients. (F) ROS detected by flow cytometry in MCL cell lines as indicated in the absence or presence of 2 mM glutamine (Q). DCF-DA, 2,7-dichlorofluorescein diacetate.

Targeting the OXPHOS pathway overcomes ibrutinib resistance in MCL

Targeting the PI3K/AKT/mTOR pathway in relapsed/refractory lymphoma is being investigated in clinical trials, and it has resulted in only moderate clinical success thus far. Moreover, c-MYC is considered “undruggable” (29, 30). On the basis of our evidence showing that glutamine-fueled OXPHOS appears to be a prominent energy metabolism pathway in ibrutinib-resistant MCL cells, we next assessed the anti-MCL effects of suppressing the OXPHOS pathway in refractory MCL cells. OXPHOS is a critical mitochondrial process that generates ATP to meet the requirements for cell growth. During the process of oxidative phosphorylation, electrons are transferred from electron donors to acceptors through the ETC in redox reactions that release energy to form ATP (31). Therefore, we used an inhibitor of ETC complex I, IACS-010759, developed by the MD Anderson Cancer Center (22) that is currently in phase 1 clinical trials for acute myeloid leukemia (NCT02882321), as well as solid tumors and lymphoma (NCT03291938). IACS-010759 inhibited the proliferation of the ibrutinib-resistant MCL cell lines (Z-138 and Maver-1) in a dose-dependent manner at nanomolar concentrations but had very little effect on the ibrutinib-sensitive MCL cell lines (Fig. 5A), with an about 10-fold difference in median inhibitory concentration (IC50) values between the resistant and sensitive cell lines. This reduction in cell viability of only the ibrutinib-resistant MCL cell lines due to OXPHOS inhibition was verified by treatment with rotenone, another ETC complex I inhibitor (fig. S8). Confirming the specificity and enzyme inhibition of IACS-010759, ETC complex I activity was significantly reduced in an in vitro enzymatic assay in ibrutinib-resistant MCL cell lines after treatment with the inhibitor (P = 0.0041; fig. S9, A and B). In addition, treatment with the OXPHOS inhibitor resulted in a lower OCR in the ibrutinib-resistant MCL cell lines, Z-138 and Maver-1, than in the ibrutinib-sensitive MCL cell lines, Mino and JeKo-1 (Fig. 5B). IACS-010759 reduced not only the OCR of the ibrutinib-resistant MCL cell lines but also the OCR of ibrutinib-resistant primary MCL clinical specimens freshly isolated from patients with MCL, with a much greater reduction compared with the OCR of ibrutinib-sensitive primary MCL cells (Fig. 5C).

Fig. 5 Targeting the OXPHOS pathway to overcome ibrutinib resistance.

(A) Cell growth inhibition of ibrutinib-resistant (Z-138 and Maver-1) and ibrutinib-sensitive (Mino and JeKo-1) MCL cell lines after a 72-hour incubation with IACS-010759 at the indicated doses (β regression model; P < 0.0001; n = 3 biological replicates, means ± SEM). (B and C) The mitochondria OCR quantitated in ibrutinib-resistant (Z-138 and Maver-1) and ibrutinib-sensitive (Mino and JeKo-1) MCL cell lines (B) and ibrutinib-sensitive and ibrutinib-resistant primary MCL cells (C) treated with 20 nM IACS-010759 for 1 hour (mixed effects regression model; P < 0.0001 between the resistant and the sensitive cell lines and primary MCL cells freshly isolated from patients with MCL; n = 3 biological replicates, means ± SEM). IACS-010759 produced an average reduction of 358.9406 (95% confidence interval, 321.1402-396.741) OCR compared with the control; mixed effects regression model; P < 0.0001. (D) Mitochondrial membrane potential (Δψm) depicted as histograms in the indicated MCL cell lines treated with 20 nM IACS-010759 and/or 5 μM BPTES for 24 hours before staining with tetramethylrhodamine, ethyl ester (TMRE) dye. (E) Quantitative analysis of apoptosis in ibrutinib-resistant Z-138 and Maver-1 compared with ibrutinib-sensitive Mino and JeKo-1 MCL cell lines (n = 3 biological replicates, means ± SD). Single-agent IACS-010759 treatment comparisons between the ibrutinib-sensitive and ibrutinib-resistant MCL cell lines calculated by linear regression P < 0.0001; IACS-010759 + BPTES versus single-agent IACS-010759 treatment in the ibrutinib-resistant MCL cell lines (P = 0.0059).

To probe the mechanisms by which IACS-010759 inhibited proliferation of the ibrutinib-resistant MCL cells, we examined the effects of the complex I inhibitor on mitochondrial potential and ATP production. IACS-010759 treatment decreased both mitochondrial potential (Fig. 5D and fig. S9C) and ATP production (fig. S9D). Moreover, IACS-010759 exposure resulted in greater glutamine uptake inhibition in the ibrutinib-resistant MCL cells compared with the sensitive cells through a currently unknown mechanism (fig. S9E). On the basis of our RPPA profiling, metabolomics, and Western blot analysis of ibrutinib-resistant MCL cell lines versus ibrutinib-sensitive MCL cell lines showing increased amounts of GLS and α-KG (Fig. 4, C to E), we hypothesized that glutamine anaplerosis may provide the major metabolites that fuel OXPHOS for MCL cells. Glutamine-derived α-KG, a TCA intermediate, replenishes the TCA cycle, fueling OXPHOS (25, 32). This metabolite is also an important precursor for other pathways such as nucleotide lipid biosynthesis that promote cell survival and growth (33, 34). As expected, the inhibition of glutamine metabolism with the allosteric GLS1-selective inhibitor BPTES further reduced the mitochondrial potential in combination with IACS-010759 (Fig. 5D and fig. S9C). Inhibition of mitochondrial ETC complex I has been suggested to inhibit growth by positively modulating ROS (35). To determine whether ROS production is altered by IACS-010759, we measured ROS after IACS-010759 treatment (25 nM) in Maver-1 and Z-138 MCL cell lines. As predicted, IACS-010759 induced much higher ROS production in the ibrutinib-resistant cell lines compared with the ibrutinib-sensitive cell lines (fig. S9F), suggesting that IACS-010759 disrupts the critical redox balance that is maintained by the reduced form of nicotinamide adenine dinucleotide (NADH) and the reduced form of nicotinamide adenine dinucleotide phosphate (NADPH) in ibrutinib-resistant MCL. Further suppression of glutamine metabolism with the addition of BPTES resulted in even greater ROS production in the ibrutinib-resistant MCL cell lines Maver-1 and Z-138 (fig. S9F), likely attributable to decreased production of glutathione or depolarization of the mitochondrial membrane potential.

Increases in ROS production are associated with apoptosis induction (35, 36); therefore, the apoptotic effects of IACS-010759 were examined in the ibrutinib-sensitive and ibrutinib-resistant MCL cell lines to determine whether apoptosis underlies the growth inhibition observed with IACS-010759 treatment. As shown in Fig. 5E and fig. S10A, single-agent IACS-010759 treatment for 72 hours resulted in 25 and 43% apoptosis in the ibrutinib-resistant Maver-1 and Z-138 cell lines, respectively, but induced much less apoptosis in the ibrutinib-sensitive Mino and JeKo-1 MCL cell lines (15% each). Single-agent IACS-010759 treatment induced caspase-7 activation and poly(adenosine 5′-diphosphate–ribose) polymerase (PARP) cleavage in ibrutinib-resistant Maver-1 and Z-138 (fig. S10B). However, the observed amounts of apoptosis were not sufficient to explain the drastic growth inhibition observed with IACS-010759 treatment in ibrutinib-resistant MCL cell lines, suggesting that additional mechanisms such as up-regulated glutaminolysis continue to sustain survival. Single-agent treatment with BPTES in ibrutinib-resistant MCL cell lines yielded similar amounts of apoptosis relative to single-agent IACS-010759 (Fig. 5E and fig. S10A) and induced caspase-7 activation and PARP cleavage (fig. S10B). However, coadministration of BPTES to inhibit glutamine metabolism induced caspase-7 activation and PARP cleavage (fig. S10B) and resulted in increased apoptosis in the ibrutinib-resistant MCL cell lines (43% in Maver-1 cells and 53% in Z-138 cells) (P = 0.0059; Fig. 5E and fig. S10A), again suggesting that ibrutinib-resistant MCL cells produce sufficient reducing equivalents to quench ROS; however, combinatory treatment with IACS-010759 and BPTES disrupts this critical balance.

To confirm the observed in vitro anti-MCL effects of IACS-010759, we examined its effects on tumor growth and proliferation in an ibrutinib-resistant MCL PDX mouse model created from a patient with MCL clinically resistant to ibrutinib (Fig. 6A). Treatment with single-agent IACS-010759 (10 mg/kg oral gavage, five consecutive days per week) completely inhibited tumor growth in comparison to the vehicle control as demonstrated by measuring tumor volume (n = 5, P < 0.0001; Fig. 6A) and human β2-microglobulin (β2M) concentration (n = 5, P < 0.0001; Fig. 6B) over the course of treatment. The IACS-010759–treated MCL PDX mice did not display any apparent toxicities. For example, body weight was not significantly different between vehicle control and IACS-010759 treatment group (P = 0.3304; Fig. 6C). Hematoxylin and eosin (H&E) and anti-CD20 staining of dissected tissue after treatment suggested repopulation of the tumor area with fatty deposits; however, a small number of CD20+ cells remained after treatment (Fig. 6D).

Fig. 6 The anti-MCL activity of IACS-010759 in an ibrutinib-resistant MCL PDX model.

Vehicle control, ibrutinib (50 mg/kg oral gavage, daily), or IACS-010759 (10 mg/kg oral gavage, five consecutive days per week) was administered to the mice beginning 5 days after engraftment until the endpoint. (A) Tumor volume calculated to reflect tumor burden [n = 5; P = 0.6397 (ibrutinib versus vehicle) and P < 0.0001 (vehicle versus IACS-010759), mixed effects regression model after logarithmic transformation] as indicated. (B) Human β2M concentrations used to monitor tumor burden [P < 0.0001 (vehicle versus IACS-010759), mixed effects regression model after logarithmic transformation] on days 0, 10, and 20 of treatment. (C) Mouse body weight calculated during drug treatment [P = 0.3304, means ± SEM (vehicle versus IACS-010759), mixed effects regression model after logarithmic transformation]. (D) H&E and CD20 staining of representative mouse tumors dissected at the end of treatment. Scale bars, 100 μM.

To further verify the anti-MCL activity of IACS-010759 in vivo, we created a second ibrutinib-resistant MCL PDX mouse model that was treated with the ETC complex I inhibitor. IACS-010759 markedly reduced tumor volume compared with the vehicle control and ibrutinib (n = 5, P < 0.001; fig. S11A). No major decreases in body weight were observed in the single-agent IACS-010759 treatment cohort (P = 0.0025; fig. S11B). Last, IACS-010759 treatment extended the survival of the ibrutinib-resistant MCL PDX mice in comparison to the vehicle- and ibrutinib-treated PDX mice (n = 5, P = 0.0027 for IACS-010759 versus vehicle and P = 0.0019 for IACS-010759 versus ibrutinib; fig. S11C), with IACS-010759 providing a median survival extension of 11 days.

To investigate the therapeutic effects of IACS-010759 in other ibrutinib-resistant B cell lymphomas, we developed an ibrutinib-resistant B cell lymphoma PDX model using tumor cells isolated from the cerebrospinal fluid of a double-hit (defined as the presence of translocations in MYC and BCL-2) patient with B cell lymphoma with CNS involvement who was clinically refractory to multiple therapies such as chemotherapy, venetoclax, and ibrutinib. Single-agent IACS-010759, but not ibrutinib, significantly inhibited tumor growth, as demonstrated by the measurement of tumor volume (P = 0.7227 for ibrutinib versus vehicle and P < 0.0001 for IACS-010759 versus vehicle; Fig. 7A) and human β2M concentration (P = 0.6911 for ibrutinib versus vehicle and P = 0.0003 for IACS-010759 versus vehicle; Fig. 7B). Body weight was not significantly different between the vehicle and IACS-010759 treatment groups (P = 0.1964; Fig. 7C). In addition, IACS-010759 significantly prolonged the survival of the ibrutinib-resistant PDX mice (n = 5, IACS-010759 versus vehicle, P = 0.0035; IACS-010759 versus ibrutinib, P = 0.0035; Fig. 7D) compared with the vehicle- and ibrutinib-treated mice, with IACS-010759 conferring a median survival benefit of more than 20 days.

Fig. 7 The anti-lymphoma effects of IACS-010759 in an ibrutinib-resistant B cell lymphoma PDX model.

Vehicle control, ibrutinib (50 mg/kg oral gavage, daily), or IACS-010759 (10 mg/kg oral gavage, five consecutive days per week) was administered to the mice beginning 5 days after engraftment until the endpoint. (A) Tumor volume calculated to reflect tumor burden [n = 5; P = 0.7227 (ibrutinib versus vehicle) and P < 0.0001 (vehicle versus IACS-010759), mixed effects regression model after logarithmic transformation]. (B) Human β2M concentrations used to monitor tumor burden on days 0, 7, and 15 of treatment [n = 5; P = 0.6911 (ibrutinib versus vehicle) and P = 0.0003 (IACS-010759 versus vehicle), mixed effects regression model after logarithmic transformation]. (C) Mouse body weight calculated during drug treatment [P = 0.1964 (IACS-010759 versus vehicle), mixed effects regression model after logarithmic transformation]. (D) Survival curve of the ibrutinib-resistant PDX model [n = 5; P = 0.0035 (IACS-010759 versus vehicle), log-rank test].

DISCUSSION

The BTK inhibitor ibrutinib is used widely to treat relapsed/refractory MCL and has moved into the frontline setting in multiple clinical trials; however, this agent yields variable therapeutic benefit among patients. The mechanisms underlying intrinsic and acquired resistance to ibrutinib need to be deciphered to further improve the clinical outcomes of MCL. The current preclinical study unravels the essential role of metabolic reprogramming in ibrutinib resistance in which a reliance on the OXPHOS pathway is observed. This finding contrasts with the Warburg effect, asserting that glycolysis drives tumor growth and proliferation (37). OXPHOS pathway activity and glutaminolysis appear to be essential sources of energy to support proliferation in ibrutinib-resistant MCL cells, and the blockade of those pathways results in cell death, suggesting that complex I of the ETC may be a valuable therapeutic target in ibrutinib-resistant MCL cells.

The up-regulation of mTORC1 and c-MYC signaling in ibrutinib-resistant MCL cells revealed by GSEA of the RNA-seq data most likely underlies the observed metabolic reprogramming toward OXPHOS. MYC and mTOR activation reprograms cancer cell metabolism by activating key genes involved in glycolysis, glutaminolysis, and mitochondrial biogenesis (38, 39) to provide an advantage to the resistant cells compared with the sensitive cells in generating energy in the form of ATP for cancer cell growth, survival, and chemoresistance mechanisms. Historically, mTOR inhibitors have resulted in therapeutic responses in lymphomas; however, their benefits are diminished by severe toxicities (4043). IACS-010759 had minimal effect on body weight in the ibrutinib-resistant lymphoma PDX mice, which is encouraging, but more relevant markers of toxicity will be identified in phase 1 studies. Our results suggest that targeting GLS, which is effective in various preclinical cancer models (44, 45), may also be an alternative approach for the treatment of ibrutinib-resistant MCL cells. Another potential therapeutic approach may include combinations with drugs targeting modulators in both the OXPHOS/GLS and mTOR pathways for optimal therapeutic efficacy in ibrutinib-resistant MCL cells.

If the observed reprogramming of the metabolic pathways in ibrutinib-resistant MCL cells is primarily due to the up-regulation of mTOR and MYC signaling pathways (46), then what causes these pathways to be activated? The glutamine transporter SLC1A5, which was up-regulated in the ibrutinib-resistant MCL clinical specimens and cell lines, acts upstream of mTORC1 signaling to activate this pathway (47). mTORC1 signaling also regulates glutamine metabolism via MYC and GLS, increasing glutamine uptake, which suggests that potential feedback mechanisms may be in place (48). SLC1A5 expression correlates with c-MYC expression (49). Whether SLC1A5 and the other identified up-regulated metabolite transporters regulate mTORC1 signaling, and potentially MYC, to produce a reliance on OXPHOS and glutaminolysis requires further investigation. In addition, on the basis of the WES data, CDKN2A deletion and CCND1 amplification (six of seven, 86%) were found only in the ibrutinib-resistant MCL samples. Larger cohorts need to be examined to determine whether these genetic alterations are biomarkers that can predict ibrutinib resistance. The CDKN2A gene codes for two tumor suppressors, p16INK4a and p14ARF, and CCND1 is an oncogene that encodes Cyclin D1. These genes are involved in cell cycle regulation, particularly in cancer cells. However, cell cycle machineries and cancer cell metabolism are interconnected (50). For example, the CCND1-CDK4 complex controls glucose metabolism independently of cell cycle progression (51). Moreover, the deletion of CDKN2A in ibrutinib-resistant MCL tumors may be of interest, primarily due to the role of p14ARF in regulating mitochondria dysfunction (52). Whether CDKN2A and CCND1 genetic alterations are responsible for reprogramming the metabolic pathways in ibrutinib-resistant MCL cells needs to be further examined.

Our study has a number of limitations such as small sample sizes for the WES and RNA-seq cohorts. Several studies validating the activation of OXPHOS in ibrutinib resistance in MCL were conducted in cell lines, which may not accurately represent the complex in vivo changes to the tumor mediating ibrutinib resistance. Although the tumor purity of the samples was quite high for most of the samples used in these studies, a small number of samples had less than 90% tumor purity. Nevertheless, the samples with tumor purity below 60% were excluded from the analysis. Furthermore, paired samples were not used in these studies because of the difficulty in obtaining paired patient samples; however, this will need to be done in future studies. Further mechanistic analyses are needed using primary patient cells, coculture systems, and in vivo models. In addition, metabolic flux experiments that further define the roles of glycolysis and glutaminolysis in MCL and ibrutinib resistance need to be conducted. Although larger cohorts are needed for further investigation, the current study supports the exploitation of active cancer metabolic pathways, especially OXPHOS and glutaminolysis, to improve the clinical outcomes of MCL and additional lymphomas (53). The impact of targeting these pathways is actively being investigated in a phase 1 lymphoma clinical trial (NCT03291938).

MATERIALS AND METHODS

Study design

This study uses genomics and functional genomics approaches to decipher the mechanisms underlying ibrutinib resistance in MCL.To this end, we first performed WES of seven ibrutinib-sensitive and seven ibrutinib-resistant MCL patient samples. As a normal control, we also sequenced a peripheral blood sample absent of any tumor cells collected from a patient with MCL. We also conducted RNA-seq on 15 ibrutinib-sensitive and 6 ibrutinib-resistant MCL patient samples. The data produced from three samples were not analyzed because of poor RNA yield or low percentage of tumor cells in the samples. These samples from patients with relapsed MCL (fresh core biopsy, bone marrow aspirates, and peripheral blood) were collected after the provision of informed consent and approval by the Institutional Review Board at the University of Texas MD Anderson Cancer Center. No statistical methods were used to predetermine the sample size, and the samples were chosen on the basis of availability and known ibrutinib response status. One replicate each for the WES, RNA-seq, and the nanoString experiments was sequenced. Once we observed the up-regulation of OXPHOS signaling based on the RNA-seq and GSEA analysis, we conducted in vitro functional studies to determine whether OXPHOS is indeed activated in MCL ibrutinib resistance. Proteomics, RPPA, and metabolomics experiments were performed in triplicate across multiple different MCL cell lines. For all in vitro experiments shown in the manuscript, at least three replicates were performed. Five mice were included per treatment group for the three demonstrated in vivo experiments, and the statistical findings are provided in the figure legends. The investigators were not blinded during sample collection and preparation, data generation, and data analysis.

Sample collection, processing, and DNA/RNA extraction

Fresh surgical biopsy, bone marrow aspirates, and peripheral blood specimens were collected from patients with relapsed MCL after the provision of informed consent and approval by the Institutional Review Board at the University of Texas MD Anderson Cancer Center. Most of the ibrutinib-resistant MCL samples were collected at relapse, meaning after ibrutinib therapy, and the sensitive samples were collected before ibrutinib therapy. Ficoll-Hypaque density centrifugation and anti-CD19 magnetic microbeads (Miltenyi Biotec) were used to isolate mononuclear cells. For DNA and RNA sequencing, fresh specimens were immediately placed into RNAlater solution after surgical biopsy or Ficoll-Hypaque density centrifugation and selective CD19 magnetic isolation of CD19+ cells from bone marrow aspiration or peripheral blood. All procedures were performed in cold buffer or on ice. DNA and RNA extraction were performed at the MD Anderson Core Facility after standard protocols.

Whole-exome sequencing

Briefly, indexed libraries were generated from 500 ng of sheared, genomic DNA (Bioruptor Ultrasonicator, Diagenode) using the KAPA Hyper Library Preparation Kit (Kapa Biosystems) and were prepared for capture with six cycles of preligation-mediated polymerase chain reaction (PCR) amplification. After amplification and reaction cleanup, the libraries were fluorometrically quantified with the Qubit dsDNA HS (High Sensitivity) Assay (Thermo Fisher Scientific) and analyzed for size distribution (Fragment Analyzer, Advanced Analytical). Normalization was performed, and the libraries were multiplexed at six libraries per pool.

A probe pool was used to hybridize each multiplexed library pool with the SeqCap EZ Human Exome Enrichment Kit v3.0 (Roche NimbleGen). The enriched libraries were amplified with eight cycles of post-capture PCR and then examined for exon target enrichment by quantitative PCR (qPCR). The exon-enriched libraries were evaluated for size distribution (Fragment Analyzer, Advanced Analytical) and quantified by qPCR using the KAPA Library Quantification Kit (Kapa Biosystems). Sequencing was performed on the HiSeq 4000 Sequencer (Illumina), one capture (six samples) per lane using the 76–base pair (bp) paired-end configuration.

WES data processing and genotyping quality check

Using Illumina’s CASAVA (Consensus Assessment of Sequence and Variation) tool (v1.8.2) (http://support.illumina.com/sequencing/sequencing_software/casava.html), raw output of the Illumina exome sequencing data was analyzed for demultiplexing and conversion into FASTQ format. The FASTQ files were aligned to the human reference genome (hg19) using BWA (Burrows-Wheeler Aligner) (v0.7.5) (54) with three mismatches (two mismatches are required in the first 40 seed regions) for a 76-base run. Then, with Picard (v1.112) and GATK (v3.1-1) software tools (55), aligned BAM files underwent base recalibration, mark duplication, and realignment. Downstream analysis was performed on the generated BAM files. To eliminate the possibility of sample swapping or contamination, genotyping quality check was conducted. Briefly, Platypus (v0.8.1) was used to call germline single-nucleotide polymorphisms (SNPs) (56). Samples originating from the same patient were identified by analyzing their overlapping percentage of identity, calculated by the identical germline allele fraction among the SNPs that overlapped between the two samples. All samples passed the quality check, and sample swapping or contamination was not observed.

Somatic mutation calling, filtering, functional annotation, and expression of mutant alleles

MuTect (v1.1.4) (57) was used to identify somatic point mutations, and Pindel (v0.2.4) (58) was used to identify small insertions and deletions (indels). Then, the MuTect and Pindel outputs were analyzed for filtering and annotation through our pipeline. Briefly, only MuTect calls identified as “KEEP” were chosen to proceed to the next step. In the case of both substitutions and indels, mutations with a low variant allelic fraction (<0.02) or with a low total read coverage (<20 reads for tumor samples; <10 reads for germline sample) were discarded. Moreover, indels with an immediate repeat region within 25 bp downstream toward their 3′ region were also discarded. ExAC (the Exome Aggregation Consortium; http://exac.broadinstitute.org), Phase 3 1000 Genome Project (http://phase3browser.1000genomes.org/Homo_sapiens/Info/Index) and the National Heart, Lung, and Blood Institute Gene Ontology Exome Sequencing Project (ESP6500, http://evs.gs.washington.edu/EVS/) databases were used to report common variants, and those with minor allele frequency of >0.5% were discarded. Last, any intronic mutations, 3′ or 5′ untranslated region (UTR) or UTR flanking region mutations, silent mutations, or in-frame small insertions and deletions were discarded.

To evaluate the probability of a missense mutation being functionally deleterious, dbNSFP (v3.0) (59) was applied to add prediction scores for all missense mutations from 12 commonly used functional prediction algorithms: PolyPhen-2 (60), SIFT (61), MutationTaster (62), Mutation Assessor (63), LRT (64), FATHMM-MKL (65) and DANN (66), PROVEAN (67), WEST3 (68), CADD (69), GERP++ (70), MetaSVM, and MetaLR (71). A missense mutation that was called as “deleterious” or “damaging” by five or more algorithms was defined as a “deleterious” mutation.

DNA copy number analysis

DNA copy number analysis was conducted using an in-house application ExomeLyzer (72), followed by Circular Binary Segmentation (73). The segmentation files were loaded to the Integrative Genomics Viewer (74) for visualization. R package was used to identify copy number gains (log2 copy ratio ≥ 0.5) and losses (log2 copy ratio ≤ −0.5).

Whole-transcriptome sequencing

Illumina-compatible, barcoded, and strand-specific total RNA libraries were assembled with the TruSeq Stranded Total RNA Sample Preparation Kit (Illumina). Cytoplasmic and mitochondrial ribosomal RNAs were removed from 250 ng of deoxyribonuclease I–treated total RNA with Ribo-Zero Gold (Illumina). Divalent cations were used to fragment the RNA after purification, and double-stranded complementary DNA (cDNA) was synthesized. After synthesis and repair, Illumina-specific indexed adapters were ligated. Purification and enrichment with 12 cycles of PCR were conducted to prepare the cDNA library.

Each library was then quantified using the Qubit dsDNA HS Assay (Thermo Fisher Scientific) and multiplexed into pools containing 24 libraries. Pooled libraries were quantified using the KAPA Library Quantification Kit (Kapa Biosystems), examined for size distribution using the Fragment Analyzer (Advanced Analytical), and then, using the 76-bp paired-end format, sequenced in four lanes of the Illumina HiSeq 4000 Sequencer.

RNA-seq data processing and quality check

RNA-seq FastQ files were processed through FastQC (v0.11.5) (www.bioinformatics.babraham.ac.uk/projects/fastqc/), a quality control tool to evaluate the quality of sequencing reads at both the base and read levels, and RNA-SeQC (v1.1.8) (75) to create quality control metrics. One RNA sample with a low sequencing yield and poor mapping rate was excluded from this study. STAR 2-pass alignment (v2.5.3) (76) was performed with default parameters to generate RNA-seq BAM files.

Identification of DEGs and enriched signaling pathways

HTSeq-count (v0.9.1) tool was applied to the RNA-seq BAM files to count how many aligned reads overlap with the exons of each gene. The HTSeq raw count data were processed by DESeq2 (v3.6) software to identify DEGs between the ibrutinib-sensitive and ibrutinib-resistant phenotypes. A cutoff fold change of ≥2 or ≤−2 and an FDR q value of ≤0.01 were applied to select the most DEGs. A ranked list of genes was generated on the basis of DESeq2 FDR q values for all coding genes and processed by GSEA (77) against the curated gene sets from MSigDB (78) to identify significantly enriched signaling pathways. A cutoff FDR q value of ≤0.01 was applied to select the most significantly enriched signaling pathways.

Cell lines

The MCL cell lines Rec-1, JeKo-1, Z-138, Maver-1, JVM-2, and JVM-13 were obtained from the American Type Culture Collection. The Granta-519 cell line was originally established by the Leibniz-Institut DSMZ and was a gift from F. Samaniego at the MD Anderson Cancer Center. The Mino cell line was originally established and provided by R. Ford at the MD Anderson Cancer Center. The SP49 cell line was a gift from J. Tao at the Moffitt Cancer Center. The JeKo-BTK KD cell line was generated by the MD Anderson Core Facility and previously verified and published (28).

Cell growth assay

Seeded cells (triplicate, 2 × 105 cells per well) were treated for 72 hours with the indicated doses of single agent. For the last 30 min of the reaction, the cells were incubated with 50 μl of CellTiter 96 AQueous One Solution Reagent in 5% CO2 at 37°C. Formazan light absorbance was recorded at 495 nm (BioTek Instruments).

Apoptosis assays

An annexin V/propidium iodide–binding assay was performed to detect apoptosis induction. Seeded cells were incubated with single-agent or combination therapy in different combinations for 48 hours. Annexin V–fluorescein isothiocyanate (FITC) was used to measure the percentages of cells undergoing apoptosis. Flow cytometric analysis was conducted with the NovoCyte Flow Cytometer (ACEA Biosciences), and the data were examined with NovoExpress (ACEA Biosciences) or FlowJo 10 (Tree Star).

RPPA assay

The RPPA assay was performed by the MD Anderson Core Facility as previously described (15). Briefly, slides with 5808 array spots were probed with validated primary and biotin-conjugated secondary antibodies (www.mdanderson.org/research/research-resources/core-facilities/functional-proteomics-rppa-core/antibody-information-and-protocols.html). The signals were amplified with a DakoCytomation-catalyzed system (Dako) and visualized with a 3,3′-diaminobenzidine (DAB) colorimetric reaction. To generate spot intensity, the slides were analyzed using Microvigene software (VigeneTech Inc.) and “supercurve fitting” R package developed by the MD Anderson Cancer Center (15, 79). The fitted curves were plotted as signal intensities (y axis) versus log2 protein concentration (x axis). Protein concentration normalization was performed by median polish and was corrected with linear expression values.

Western blot analysis

Cells were treated with single agents or two agents, harvested, washed, and lysed in radioimmunoprecipitation assay (RIPA) buffer (Thermo Fisher Scientific) with protease inhibitors. Cell lysates were maintained for 30 min on ice and centrifuged at 13,000g for 10 min at 4°C. A Bio-Rad Bradford assay (Bio-Rad) was used to determine the protein content. Sample preparation electrophoresis on 10% SDS–polyacrylamide gel electrophoresis was conducted before transfer onto a nitrocellulose membrane (Bio-Rad). The membranes were blocked [5% nonfat dry milk in phosphate-buffered saline (PBS) containing 0.05% Tween 20] for 2 hours and immunoblotted with SLC16A1 (Novus), SLC1A5, BTK, GLS, Caspase-7, cCaspase-7, PARP, cPARP, and GAPDH (Cell Signaling Technology) antibodies before visualization with chemiluminescence (Pierce Biotechnology).

Histopathological analysis

Tissues were excised from the PDX mice, fixed in 10% formalin, processed, paraffin-embedded, sectioned (5 μm), and stained with H&E as previously described (15). CD20 was detected with human anti-CD20 (Dako) and visualized with an Olympus BX51TF microscope (Olympus).

PDX models

The Institutional Animal Care and Use Committee of the University of Texas MD Anderson Cancer Center approved the experimental protocols as previously described (15, 80). To create the PDX models, 6- to 8-week-old NSG mice (the Jackson Laboratory) were anesthetized with 5% isoflurane vaporizer and then injected with 5 × 106 freshly isolated lymphoma cells, either into human fetal bone implants within NSG-hu hosts or into the subrenal capsule as previously described (81). Concentrations of circulating human β2M in the collected mouse serum, as analyzed with a human β2M enzyme-linked immunosorbent assay (ELISA) kit (Abnova Corporation), were used to monitor engraftment and tumor burden. After detection of engraftment [β2M concentration (30 ng/ml) detected in collected mouse serum], human anti-CD20 expression was detected in the tumor by flow cytometry, and then, the tumors were transferred to additional NSG mice to create the second generation. After tumor passage and subsequent tumor growth, the mice were randomly divided into five mice per group for in vivo treatment. Five days after tumor implantation, mice were treated with vehicle control, ibrutinib (50 mg/kg, oral gavage daily), or IACS-010759 (10 mg/kg, oral gavage for five consecutive days per week) until the endpoint, which was identified as the point at which one diameter of tumor mass was measured at 15 mm or greater. Mouse serum was collected throughout and after treatment with either the vehicle control or the indicated drug, and by measuring either tumor volume (using a caliper) or human β2M concentration, tumor burden was assessed and used to determine the therapies’ efficacy in the PDX mouse models. Tumor cells collected from each generation and treatment groups were labeled with FITC/phycoerythrin (PE)-conjugated anti-human CD5, CD19, and CD20 monoclonal antibodies (BD Biosciences) to determine the human lymphoma cell population using flow cytometry.

Mitochondrial membrane potential, mitochondria mass, ROS, and apoptosis analysis

MCL cells were seeded and cultured overnight before treatment to reach log phase. The cell lines were plated at 2 × 106 cells per well in six-well plates in phenol red and pyruvate-free RPMI 1640 assay medium containing 5 mM glucose (Sigma-Aldrich), 2 mM GlutaMAX (Gibco, Thermo Fisher Scientific), and 5% dialyzed fetal calf serum (Gibco). Vehicle, IACS-0107059 (20 nM), or BPTES (1 to 5 μM; Sigma-Aldrich) was added and replenished every day until the indicated time points. ROS and mitochondrial membrane potential (Δψm) were evaluated 24 hours after treatment. To determine mitochondrial membrane potential (Δψm), treated cells were loaded with MitoStatus TMRE (BD Biosciences) at 37°C for 15 min and washed with PBS twice to remove residual TMRE. TMRE dye fluorescence retained in the cells was detected by flow cytometry with excitation and emission at 548/574 nm. Data were analyzed with NovoExpress (ACEA Biosciences) or FlowJo 10 (Tree Star).

To measure ROS, treated cells were harvested, stained with 0.5 mM CM-H2DCF-DA (C6827, Molecular Probes, Invitrogen) in PBS for 30 min at 37°C, washed twice with PBS, and analyzed with flow cytometry (NovoExpress, ACEA Biosciences) by measuring the green fluorescent output of oxidized DCF. The mean fluorescence of 10,000 cells was used as the measurement of intracellular ROS.

Mitochondria respiration

The OCR and ECAR were measured using the Seahorse XF 96 Metabolic Flux Analyzer (Seahorse Bioscience). Briefly, MCL cells were treated with or without IACS-010759 (20 nM) for 1 hour. On the day of the analysis, 2 to 4 × 105 MCL cells per well were adhered to XF 96-well cell culture microplates (Seahorse Bioscience) precoated with Cell-Tak (BD Biosciences) by centrifugation at 1000 g without brakes for 5 min and then changed to mitochondria stress medium supplemented with 25 mM glucose, 2 mM sodium pyruvate, and 2 mM glutamine (pH 7.4), followed by incubation at 37°C in a non-CO2 incubator for 1 hour. Oligomycin (1 μM), carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP) (2 μM), and rotenone/antimycin A (0.5 μM) were loaded into the drug delivery ports and sequentially injected. After each injection, four time points were recorded at 35-min intervals. For the glycolysis stress test, glucose, oligomycin, and 2-deoxyglucose were injected to final concentrations of 10 mM, 1 μM, and 50 mM, respectively. OCR and ECAR were normalized to protein content measured by sulforhodamine B staining and plotted using GraphPad Prism. Basal respiration, spare capacity, ATP production, glycolysis, and glycolytic capacity were calculated on the basis of the OCR and ECAR readings. The ratio of OCR to ECAR at basal respiration was calculated and used to assess cellular preference for OXPHOS versus glycolysis.

Targeted metabolomics analysis

MCL cell lines were grown in complete RPMI 1640 medium for 2 days. Next, 1 × 107 cells were seeded in fresh medium in T75 flasks and then harvested either immediately (0 time point) or 24 hours later after cell counting and washed with PBS. Metabolites were immediately extracted and subjected to targeted metabolomics analysis for amino acids and TCA metabolites. Electrospray ionization mass spectrometry (cation) or Agilent 6460 Triple Quad LC/MS (anion) was conducted. Using the peak information including mass/charge ratio, migration time, and peak area, each metabolite was identified and quantified. The peak area was then converted to relative peak area by normalization with internal standard peak area and total cell number. Relative concentrations of metabolites were plotted as a heat map using GraphPad Prism.

Glutamine consumption and ATP analysis

MCL cell lines were seeded at 1 to 2 × 106 cells per well in triplicate in six-well plates and incubated overnight. Next, the media were replaced with 3 ml of complete RPMI 1640 medium with or without inhibitors. Cell culture supernatants were collected 2 days after treatment with IACS-010759 or BPTES and analyzed for their glutamine content using the Glutamine Colorimetric Assay Kit (BioVision). Glucose or glutamine consumption was determined by subtracting the measured glucose concentrations in the media from the initial concentrations (11.1 mM glucose and 2 mM glutamine). Intracellular ATP contents from the above treated cells were determined using the ATP Colorimetric Assay Kit (BioVision) as outlined in the manufacturer’s protocol. Relative concentrations of all metabolites were then normalized to the relative sum of total cell counts over a 48-hour treatment period.

Statistical analysis

All assays were performed in triplicate and expressed as mean ± SEM or SD. Statistical significance of differences was determined by Student’s t test, linear regression models, mixed-effects regression models after appropriate transformation, and b regression models. OS and PFS were calculated with the Kaplan-Meier method and compared with the log-rank test. All analyses were performed using statistical software R v3.4.3 with packages betareg v3.1-0, nlme v3.1-131, and survival v2.41-3 and were plotted using package survminer v0.4.2 and software GraphPad Prism v7.03. P values less than 0.05 were considered statistically significant.

SUPPLEMENTARY MATERIALS

stm.sciencemag.org/cgi/content/full/11/491/eaau1167/DC1

Materials and Methods

Fig. S1. Survival outcomes of patients with MCL relative to ibrutinib response.

Fig. S2. Tumor cellularity comparison between the ibrutinib-sensitive and ibrutinib-resistant clinical specimens.

Fig. S3. DEGs between ibrutinib-sensitive and ibrutinib-resistant tumors using the nanoString nCounter system 63-gene ibrutinib resistance panel.

Fig. S4. DEGs between the ibrutinib-sensitive and ibrutinib-resistant tumors using the nanoString nCounter system panCancer panel.

Fig. S5. Relative abundance of metabolites in ibrutinib-sensitive and ibrutinib-resistant MCL cell lines.

Fig. S6. OXPHOS pathway activation in ibrutinib-resistant MCL cells.

Fig. S7. OXPHOS pathway member up-regulation in BTK KD ibrutinib-resistant MCL cell lines.

Fig. S8. The growth inhibition effects of rotenone treatment on ibrutinib-resistant MCL cells.

Fig. S9. Reductions in mitochondrial activity and ATP production and induction of ROS production in ibrutinib-resistant MCL cells.

Fig. S10. Induction of apoptosis in ibrutinib-resistant MCL cells.

Fig. S11. The anti-cancer effects of IACS-010759 in an ibrutinib-resistant MCL PDX model.

Table S1. Information on the samples used for the nanoString nCounter system assay using the 63-gene ibrutinib resistance panel.

Table S2. The IC50 values of ibrutinib across MCL cell lines.

Data file S1. The clinical and pathological data of patients with MCL.

References (82, 83)

REFERENCES AND NOTES

Acknowledgments: We thank the patients and their families who contributed to this research study. We also thank F. Samaniego at the MD Anderson Cancer Center, R. Ford at the MD Anderson Cancer Center, and J. Tao at the Moffitt Cancer Center for providing us with the cell lines Granta-519, Mino, and SP49, respectively. Funding: This study was supported by the philanthropic support to the MD Anderson B Cell Lymphoma Moon Shot Project; philanthropy funds from the Gary Rogers Foundation, Kinder Foundation, and the Cullen Foundation; and the start-up research funds provided to L.W. by the Department of Genomic Medicine, Division of Cancer Medicine, MD Anderson Cancer Center. This study was also supported by the Cancer Prevention Research Institute of Texas (CPRIT) Grant RP130397, the NIH-funded Cancer Center Support Grant (CCSG) P30 CA016672 [P. Pisters, principal investigator (PI)], the NIH Core Grant for the Sequencing and Microarray Facility (CA016672), the NIH High-End Instrument Grant 1S10OD012304-01, and the Grant R21 CA202104 (to M.W., PI). The Institute for Applied Cancer Science at the MD Anderson Cancer Center provided IACS-010759 for these studies. The Cancer Genetics Laboratory at the MD Anderson Cancer Center, a Moon Shots–supported platform, conducted WES of the patient samples. The Leukemia & Lymphoma Society supports the IACS-010759 solid tumor and lymphoma clinical trial (NCT03291938). The development of IACS-010759 was supported, in part, by the Leukemia & Lymphoma Society through its therapy acceleration program (TAP) and by the MD Anderson Moon Shots Program. Author contributions: M.W. and L.W. conceived and jointly supervised the study. L.Z., K.N., L.W., Y.Y., and M.W. conceived the experiments. Y.Y., Y.L., T.B., M.B., S. Zhang., H.G., H.Z., E.L., G.H., L.W., M.B., L.Z., K.N., M.A., L.V.P., and M.W. carried out the experiments and data analysis. J.Z., X.S., and X.M. contributed to the raw sequencing data processing. S. Zhang. contributed to the DNA and RNA-seq data processing, quality check, mutation and expression analysis, and generation of figures and tables for the manuscript. G.H. assisted with the RNA-seq analysis. S. Zhou. contributed to the biostatistical analyses. Y.S., M.E.D.F., N.F., X.M., X.S., J.Z., J.M., T.H., G.D., P.J., and A.F., contributed to the creation of IACS-010759 and bioinformatics analysis. P.L.L. supervised the targeted metabolomics and contributed to the manuscript preparation. R.H. supervised and R.L. and S.G.M. conducted the proteomic study and data analysis. L.Z., K.N., L.W., Y.Y., Y.L., L.V.P., and M.W. wrote and revised the manuscript and prepared the figures. Competing interests: P.J. and M.E.D.F. are inventors of the following patent: P. Jones, M. E. Di Francesco, T. McAfoos; Salts of the Heterocyclic Modulators of HIF Activity for the Treatment of Diseases (publication no. WO 2015/130790; applicants: Board of Regents, University of Texas System). All other authors declare that they have no competing interests. Data and materials availability: All data associated with this study are present in the paper and/or the Supplementary Materials. All WES and RNA-seq data have been deposited at the European Genome-phenome archive (EGA). The datasets can be fully accessed under the accession number EGAS00001003418. Further information about the EGA can be found on https://ega-archive.org: “The European Genome-Phenome Archive of Human Data Consented for Biomedical Research” (www.nature.com/ng/journal/v47/n7/full/ng.3312.html).
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article