To the best of our knowledge, this study is the first reported investigation comparing bioluminescence imaging using a firefly luciferase system in monoclonal and polyclonal human leukemia cell populations in vitro and in a xenograft mouse model. We have shown that the bioluminescence signal intensity and the dynamics of luciferase activity in vitro were cell line dependent (Figure ). Moreover, bioluminescence signal intensity was unstable in polyclonal populations and decreased significantly with repeated passage in culture. The decreasing signal observed may be due to survival and growth advantages for the non-transduced cells within the polyclonal cell population. We also observed dramatic variability in signal intensity in sub-lethally irradiated NSG mice transplanted with a polyclonal luciferase-transduced cell population, suggesting that in a polyclonal population, different clones contribute to establishment of disease in different mice (Figure ). This source of heterogeneity significantly decreases the power of this model system to determine differences in disease progression related to experimental treatments. In addition, the variability in bioluminescence signal we observed in polyclonal luciferase-transduced populations limits side-by-side comparison of cell line derivatives with targeted genetic manipulations, where differences in signal intensity between cell lines will optimally be due solely to differences in disease progression, rather than cell line dependent differences in transduction efficiency and/or luciferase activity.
With the generation of monoclonal luciferase-expressing leukemia cell lines, stability of luciferase activity over a long period of time (> 4 months) was obtained. Comparable bioluminescence intensity in monoclonal luciferase-tagged cell lines with targeted genetic modifications was also observed. Moreover, monoclonal cell populations showed the same dynamic of bioluminescence intensity after administration of D-luciferin in vitro, revealing a significant advantage of this method (Figure ). Most importantly, transplantation of monoclonal luciferase-transduced cell lines in a xenograft mouse model resulted in genetically and phenotypically identical disease with comparable disease kinetics, thereby significantly improving the utility and sensitivity of this model (Figure ).
In the system used for our investigations, we demonstrated that the selection of monoclonal luciferase-expressing populations based on equal luciferase activity resulted in isolation of cell lines that were directly comparable, both in vitro and in vivo. Commonly the use of an antibiotic resistance marker would likely be sufficient for in vitro models. However, we avoided the treatment of leukemia cells with additional antibiotic agents to minimize putative in vivo drug interactions in future translational drug studies. In addition, antibiotic selected cell populations maintain a heterogeneous expression of luciferase, which may introduce variability to the transplanted cell population and may affect the consistency during the experiment.
The method to transplant monoclonal luciferase-transduced cell populations in murine xenograft models presented here has several advantages, but there are also some limitations which should be mentioned. First, we have seen that survival in xenograft models can be impacted by lentivirus transduction and/or luciferase expression in leukaemia cells and thus, these models may less accurately represent the normal biology of leukemogenesis (data not shown). Second, our data indicate that only monoclonal populations with the same growth characteristics, luciferase activity and phenotype can be directly compared in luciferase-based murine xenograft models, limiting the ease of their utility for comparison of genetically modified cell lines and their parental counterparts.