Researchers find broad-spectrum antibiotic tigecycline selectively kills leukemic stem and progenitor cells.
Scientists have highlighted mitochondrial translation inhibition as a potential therapeutic strategy for treating acute myeloid leukemia (AML). An Ontario Cancer Institute-led research team carried out a screen of FDA-approved drugs to find compounds that are active against AML.
Results showed that the broad-spectrum antibiotic tigecycline selectively killed leukemia stem and progenitor cells without affecting normal hematopoietic cells. Tigecycline in addition showed antileukemic activity in mouse models of human leukemia. The effects of the drug were subsequently found to relate to inhibition of mitochondrial protein translation.
Aaron D. Schimmer, M.D., and colleagues, report their findings in Cancer Cell in a paper titled “Inhibition of Mitochondrial Translation as a Therapeutic Strategy for Human Acute Myeloid Leukemia.”
In series of experiments designed to identify potential AML treatment candidates among existing drugs, the team generated a library of 312 FDA-approved compounds—primarily antimicrobials and metabolic regulators—that are already well-characterized in terms of pharmacokinetics and toxicology. They then screened this library against two AML cell lines, TEX and M9-ENL1, which display features of leukemia stem cells (LSCs).
Tigecycline was found to be one of the most highly effective compounds found to impact on AML cell viability. Further dose-ranging tests showed the drug had antileukemic activity against a panel of human and murine leukemia cells including those that weren’t sensitive to minocycline or tetracycline, which are structural analogs of tigecycline.
Encouragingly, while normal hematopoietic cells were only sensitive to high concentrations of tigecycline, cancer cells taken from 13 of 20 primary AML samples (including those from newly diagnosed and relapsed patients) were highly sensitive to much lower concentrations of the drug. The CD34+38– subset of AML progenitor cells was also similarly sensitive to tigecycline, the team notes.
Further testing showed that at 5 micromolar concentrations, tigecycline reduced the clonogenic growth of primary AML patient samples by 93%, but had only minimal effect on the clonogenic growth of normal hematopoietic cells tested using the same protocol. Tigecycline treatment in addition reduced the ability of human AML cells to repopulate in NOD-SCID mice.
The researchers then moved on to try and identify candidate protein targets of tigecycline. To this end they carried out using haplo-insufficiency profiling (HIP) in yeast, a technique that gives an unbiased in vivo quantitative measure of the relative drug sensitivity of all 6,000 yeast proteins in a single assay. This showed that yeast growth was relatively insensitive to tigecycline under standard fermentation conditions in rich media (YP), where the primary mode of metabolism is glycolysis. In contrast, under respiratory conditions (using YPGE media) that depend on oxidative phosphorylation, tigecycline treatment resulted in increasing, dose-dependent inhibition of yeast growth.
The team ranked the list of drug-sensitive strains generated from the HIP assays and carried out gene set enrichment analysis (GSEA) to identify gene ontology (GO) biological processes that were enriched in the tigecycline screens. This showed that the most significantly enriched GO process was the mitochondrial ribosome.
In support of the role of mitochondrial processes in tigecycline activity, they also found that the mammalian mitochondrial translation inhibitors chloramphenicol and linezolid generated similar GO enrichment results to tigecycline, whereas results for the anthracycline drug doxorubicin were indicative of a different mechanism. Subsequent studies, however, found that while chloramphenicol and linezolid both reduced clonogenic growth of cells from two primary AML samples, higher concentrations were needed than were required to achieve tigecycline-induced killing.
Data thus far suggested that tigecycline inhibits the growth and viability of eukaryotic cells by intereferring with mitochondrial protein translation, which is consistent with the antibiotic’s known ability to block bacterial protein synthesis by reversibly binding bacterial ribosomal RNA, the authors point out.
The next stage was to see whether the drug’s toxicity to leukemic cells specifically was mediated by a similar mechanism. A series of experiments were carried out to test whether exposure of primary AML cells to tigecycline alters expression of proteins known to be dependent on either cytosolic or mitochondrial ribosomes.
They found that tigecycline treatment caused a preferential decrease in the expression of cytochrome c oxidases 1 and 2, which are translated by mitochondrial ribosomes, but had little effect on levels of Cox-4, which is encoded by the nuclear genome and translated by nuclar ribosomes. Tigecycline also had no effect on the expression of other proteins translated by cytosolic ribosomes.
Similarly, tigecycline significantly decreased the enzyme activity of respiratory complexes I and IV, both of which contain proteins translated on mitochondrial ribosomes, but had less effect on the enzymatic activity of the respiratory chain complex II.
“These findings were mirrored by treatment with chloramphenicol, a known mitochondrial protein synthesis inhibitor,” the team adds. Moreover, they found that tigecycline treatment led to decreased mitochondrial membrane potential in primary AML cells (preceding the onset of cell death) but not in normal hematopoietic cells.
The notion that tigecycline acts primarily at the level of mitochondrial function was confirmed in two separate experiments. These showed that the drug treatment didn’t have any additional impact either on leukemic cells that were already under hypoxic conditions (which independently reduces mitochondrial membrane potential), or on Burkitt’s lymphoma cells with repressed Myc, which already demonstrate decreased mitochondrial mass, mitochondrial DNA copy number, and oxygen consumption.
The authors wanted to see whether the effects of tigecycline could be mimicked genetically, so they used shRNAs to knock down either mitochondrial elongation factor Tu (EF-Tu) or IF-3 in TEX AML cells. IF-3 plays a role in initiating mitochondrial translation, while EF-Tu is involved the process of brining transfer RNAs to the mitochondrial ribosome for translation. They found that EF-Tu knockdown increased mRNA expression and decreased protein expression of Cox-1 and Cox-2, but not Cox-4. Similarly to the effects of tigecycline treatment, EF-Tu knockdown also reduced the growth and viability of TEX cells and led to decreased mitochondrial membrane potential and oxygen consumption. In contrast, IF-3 knockdown had none of these effects on mRNA expression, protein production or growth/viability and mitochondrial function.
Another question to be addressed, was why did leukemic cells demonstrate hypersensitivity to mitochondrial translation inhibition. Comparing the mitochondrial characteristics of primary normal hematopoietic cells and AML cells showed that patient-derived AML cells including CD34+/CD38+ and CD34+/CD38– subsets had higher mitochondrial DNA copy number compared with normal hematopoietic cells, as well as a larger mitochondrial mass and, as a result, higher rates of oxygen consumption.
Moreover, measurements on leukemic cells from nine different patients indicated that mitochondrial mass was significantly negatively correlated with in vitro sensitivity to tigecycline: i.e. samples with the greatest mitochondrial mass were the most sensitive to tigecycline therapy.
“Taken together, these results suggest that AML progenitors and stem cells are more metabolically active and dependent on mitochondrial function than are normal hematopoietic cells, and provide a mechanism to explain the observed differential activity of mitochondrial translation inhibition in leukemic and normal hematopoietic cells at all levels of differentiation,” the authors state.
When tested in vivo, intraperitoneal tigecycline therapy significantly slowed the growth of OCI-AML2 cells transplanted into SCID mice, and the drug was at least as potent as therapy using daunorubicin or bortezomib at their maximum tolerated doses. The tigecycline treatment has no evident effects on any of the animals organs, and tumors excised from tigecycline-treated mice showed a greater reduction in the expression of Cox-1 and Cox-2, relative to Cox-4, when compared with vehicle-treated controls.
NOD/SCID mice given xenografts of primary AML cells from human patients also responded to tigecyline therapy. A three week course of tigecycline started three 3 weeks post-transplant led to significantly lower levels of leukemic engraftment compared with control-treated mice, and without evidence of toxicity, changes in animal behaviour, or gross changes to organs. “Importantly, leukemic cells harvested from the bone marrow of tigecycline-treated primary mice generated smaller leukemic grafts in untreated secondary mice, compared to cells harvested from control-treated primary mice, indicating that tigecycline was active against AML stem cells,” the authors stress.
Encouragingly, subsequent in vitro and in vivo studies suggested that tigecycline acts synergistically or additively when co-adminstered with existing drugs daunorubicin or cytarabine: in the in the OCI-AML2 xenograft model, animals treated using either tigecyline plus cytarabine or tigecycline plus daunorubicin showed reduced tumor growth incomparison with animals receiving single agent therapy. Again, there were no evident effects on the histology of any organs studies, or alterations in serum levels of liver or muscle/cardiac enzymes.
“Despite the genetic and biological diversity of spontaneously arising human AML clones, some common biochemical pathways accessible to selective targeting appear to still exist and await therapeutic exploitation,” the authors conclude. “we report that the antimicrobial tigecycline has toxicity for human AML cells at all stages of development in both in vitro and in vivo preclinical models, while sparing normal hematopoietic cells.”