By Richard Simoneaux
Acute myeloid leukemia (AML) is a lethal disease that has a survival rate that leaves only approximately one-fourth of those patients alive 5 years after diagnosis. Standard induction therapy for patients with AML often consists of the '7+3' regimen of standard dose cytarabine for 7 days and 3 days of a cytotoxic anthracycline antibiotic such as daunorubicin, although, doxorubicin or idarubicin or mitoxantrone may be used as well. Then, based upon the characteristics of their disease, they may undergo either consolidation chemotherapy or hematopoietic stem cell transplant.
Despite the 60-70 percent complete response rate noted with induction therapy, a significant portion of these patients will undergo a relapse of their disease within 3 years. These relapses are thought to be due to the evasion of treatment-resistant leukemia stem cells (LSCs). Consequently, therapies which target and effectively eliminate these cells are a large unmet need for this patient population.
In preclinical studies, Irène Baccelli, PhD, from th Institute for Research in Immunology (IRIC) and Cancer, Université de Montréal, and colleagues evaluated mubritinib in a variety of in vitro and mechanistic studies as well as mouse-based patient-derived xenograft models. Results from these studies were published recently (Cancer Cell 2019;36(1):84-99).
"These results demonstrate that the observations we made with our LSC-optimized in vitro screening method are translatable in vivo and may be applied to treat AML (at least in patient-derived xenograft models in mice)," Baccelli observed.
In Vitro Studies
One major hurdle to the development of novel therapies for AML has been the absence of cell culture conditions that maintain LSC activity in vitro. The discovery of culture conditions which prolonged the lifetimes of LSC and prevented their differentiation was made by this same research group (Nat Methods 2014;11:436-442). In those studies, they found that the combination of UM279 with inhibitors of the aryl-hydrocarbon receptor pathway were able to effect ex vivo expansion of these LSCs, thus permitting in vitro assays utilizing these cells.
A panel of several different inhibitors of the RAS and PI3K pathways and tyrosine kinase inhibitors were then screened and GI50 (50% growth inhibition) values were obtained for a small array of primary AMLs using the aforementioned culture conditions for LSCs. These screenings identified mubritinib, a previously described ERBB2 inhibitor (Int J Urol 2006;13:587-592) as the most selective compound against those specimens coming from patients with poor outcomes (p=0.009).
Subsequent validation of these findings was accomplished by testing mubritinib against an array of 200 genetically heterogeneous samples of AMLs. Those cells obtained from patients with poorer outcomes were confirmed as having greater sensitivity to mubritinib than cells derived from patients with better outcomes (p=0.0048). Interestingly, mubritinib showed no proliferative effects on the hematopoietic stem/progenitor CD34+ cord blood cells used as a control, even at concentrations up to 10 mM.
Those patients that provided mubritinib-sensitive samples showed lower overall survival (OS) rates relative to those who had mubritinib-resistant specimens (p=0.0099). It is of interest to note that no difference was observed in complete remission rates for patients with mubritinib-sensitive or resistant AML; however, mubritinib-sensitivity was associated with a significant increase in relapse rates (p=0.0048).
In Vivo Studies
To further assess mubritinib as a potential therapy for poor outcome AMLs, two separate 5-week patient-derived xenograft (PDX) studies were initiated: PDX 1—an adverse cytogenetic risk HOX-high AML, and PDX 2—a poor outcome AML with mutations to NPM1 and DNMT3A, as well as FLT3-ITD.
Analyses performed post-dosing for the PDX 1 mice showed that the levels of the human CD45+ cells found in the bone marrow were decreased by 1.6-fold. For PDX 2 mice, the bone marrow levels of human CD45+ cells were decreased by 6.3-fold. In vitro analyses showed that the cells from PDX 2 were roughly four-fold more sensitive to mubritinib than those from PDX 1. It is of interest to note that this figure was in line with the four-fold difference in efficacy noted in the in vivo PDX studies (bone marrow results), suggesting that the in vitro results may be reliably translated to in vivo efficacy.
One particularly interesting observation was the fact that the signal produced by mubritinib using a molecular probe was not co-localized with the ERBB2 receptor tyrosine kinase in BT474 cells; rather, it was co-localized with a mitochondrial dye, which suggests mitochondrial accumulation of mubritinib.
Upon treatment with mubritinib, the ratios of AMP/ATP and ADP/ATP were both increased in a statistically-significant manner (p<0.0001 each). Oxygen consumption analyses for two different AML cell lines showed that mubritinib inhibited mitochondrial respiration. Concurrently, increased extracellular acidification rates were noted in cells treated with mubritinib; this activity could be rescued by treatment with the glylcolysis inhibitor 2-deoxy-D-glucose, which suggests upregulated glycolytic activity upon treatment. Interestingly, higher intra- and extracellular lactate levels were noted in mubritinib-treated cells. Taken together, these data would seem to suggest that treatment of AML cells with mubritinib causes a switch from oxidative phosphorylation to glycolytic metabolism.
In order to further confirm the mechanism by which mubritinib acts on AML cells, whole genome CRISPR/Cas9 screens were performed in the presence and absence of mubritinib. These studies were performed with the mubritinib-sensitive NALM-6 B-cell precursor leukemia cell line. Importantly, this cell line was resistant to lapatinib, a known ERBB2 inhibitor.
These studies identified a synthetic lethal interaction between mubritinib treatment and loss of glutamic-oxaloacetic transaminase 1 (GOT1) expression. This relationship is particularly interesting, as a similar synthetic interaction had been noted in earlier studies for GOT1 knockout and electron transport chain (ETC) inhibitors (Cell 2015;162:540-551, 552-563).
In order to verify that mubritinib did inhibit the ETC, enzymatic assays were performed for each complex involved. These analyses showed that mubritinib did not diminish the activities of ETC complexes II-V; however, it was an efficient inhibitor of ETC complex I (NADH dehydrogenase), with an IC50 (in vitro half-maximal inhibitory concentration) value of 51 nM. This level of activity was somewhat similar to that of the standard ETC complex I inhibitor rotenone, which had an IC50 value of 17 nM. Mubritinib treatment did not appear to affect the ubiquinone-independent diaphorase activity of ETC complex I, which suggests that perhaps, similar to rotenone, mubritinib may bind at or in the vicinity of the ubiquinone binding site.
"Although we have not yet identified which subunit of ETC complex I mubritinib binds to (this is quite difficult as the complex is composed of 45 different units)," Baccelli explained, "we have noticed that its mechanism of action is dependent on ubiquinone, quite similar to the reference ETC complex I inhibitor, rotenone. We therefore hypothesize that mubritinib may bind near or at the ubiquinone binding site of the complex."
To further verify if the anti-AML activity of mubritinib is a direct result of its ETC complex I inhibition, the activities of 15 structurally unique mubritinib analogs were evaluated in cellular assays to obtain GI50 values; in addition, IC50 values were obtained for these derivatives in an enzymatic assay for ETC complex I inhibition. Strong correlations were observed for both GI50 and IC50 values in human and murine AML cells (r=0.9; p<0.0001).
In this study, it was shown that chemotherapy-sensitive AMLs with favorable cytogenetic risk did not require ETC complex I activity for energy production, which was consistent prior observations made in AML xenograft studies (Cancer Discov 2017;7;716-735).
Rather interestingly, a strong correlation was noted between sensitivity to ETC complex I targeting in primary AML specimens and increased expression of mitochondrial activity-associated gene modules. In addition, direct metabolic profiling of patient-derived AML samples showed an association for mubritinib sensitivity with oxidative phosphorylation hyperactivity; consequently this may therefore serve as a biomarker for ETC complex I addiction in AML.
"Several recent studies would seem to suggest that LSCs are metabolically remarkable," Baccelli noted, "in the sense that, contrary to normal hematopoietic stem cells (HSCs), which use mainly glycolysis as a source of energy, LSCs rely heavily on oxidative phosphorylation for their survival. As mubritinib impairs oxidative phosphorylation through ETC complex I targeting, we believe that this is the way LSCs are efficiently killed when exposed to mubritinib."
"We do find that mubritinib has an effect on those patient specimens, which have an abundance of the difficult-to-treat leukemic stem cells, particularly within the relatively homogeneous genetic subgroup of normal karyotype AMLs. LSC frequencies are higher within sensitive patient specimens compared to resistant samples, as assessed by limiting dilution assays in mice," Baccelli explained. "Interestingly, we find that, in the context of the MLL-AF9 murine AML in vivo model, cells expressing markers that characterize LSCs are efficiently targeted by mubritinib treatment in mice."
When asked to summarize the main findings of their study, she stated, "First of all, we identified an extreme dependency towards oxidative phosphorylation in the leukemic cells of a subgroup of poor outcome AML patients.
"Second, we demonstrate that mubritinib, a compound previously known as an ERBB2 inhibitor, which completed a phase I trial for the treatment of ERBB2 overexpressing solid tumors, is a direct inhibitor of the mitochondrial ETC complex I.
"Third, we find that mubritinib exerts a strong in vitro and in vivo anti-leukemic effect in oxidative phosphorylation-dependent AMLs through ETC complex I inhibition (and not through ERBB2 blockade)."
Regarding the AML subtypes identified in this study, Baccelli stated, "We find that AMLs most vulnerable to oxidative phosphorylation targeting (using the ETC complex I inhibitor mubritinib as a sensor) are those belonging to the traditional intermediate cytogenetic risk group that are doing poorly in the clinic: specimens with mutations affecting NPM1, DNMT3A or FLT3-ITD, including dismal survivor AMLs that carry simultaneously these three mutations are particularly sensitive. High expression of oxidative phosphorylation/mitochondrial metabolism related genes, as well as high oxidative phosphorylation activity are hallmarks of such leukemic cells."
When asked how the development of mubritinib was proceeding, Baccelli noted, "The small molecule mubritinib has been modified and improved, with the IRIC medicinal chemistry platform before being patented. A large pharmaceutical company has just signed a licensing agreement with us to mobilize researchers with the aim of accelerating clinical trials in AML."
Richard Simoneaux is a contributing writer.