Mechanical phenotyping of acute promyelocytic leukemia reveals unique biomechanical responses in retinoic acid-resistant populations

Abstract

While all-trans retinoic acid (ATRA) is an essential therapy in the treatment of acute promyelocytic leukemia (APL), an aggressive subtype of acute myeloid leukemia, nearly 20% of APL patients are resistant to ATRA. As no biomarkers for ATRA resistance yet exist, we investigated whether cell mechanics could be associated with this pathological phenotype. Using mechano-node-pore sensing, a single-cell mechanical phenotyping platform, and patient-derived APL cell lines, NB4 (ATRA-sensitive) and AP-1060 (ATRA-resistant), we discovered that ATRA-resistant APL cells are less mechanically pliable. By investigating how different subcellular components of APL cells contribute to whole-cell mechanical phenotype, we determined that nuclear mechanics strongly influence APL cell mechanical responses. By arresting APL cells in S-phase or M-phase in the cell cycle, we found cell pliability to be inversely related to DNA content. In addition to DNA content affecting cell pliability, we observed that chromatin condensation also affects nuclear mechanics: decondensing chromatin with trichostatin A is especially effective in softening ATRA-resistant APL cells. RNA-Seq allowed us to compare the transcriptomic differences between ATRA-resistant and ATRA-responsive APL cells and highlighted gene expression changes that could be associated with mechanical changes. Overall, we demonstrate the potential of physical biomarkers in identifying APL resistance.

Evaluating sources of technical variability in the mechano-node-pore sensing pipeline and their effect on the reproducibility of single-cell mechanical phenotyping

“…The geometry of each mechano-NPS device is optimized to measure a particular cell size. To assess the reproducibility of previous studies using mechano-NPS, we employed the same device designs from Kim et al and Li et al to measure MCF-10A and AP-1060 cells, respectively [ 5 , 6 ] ( Fig 1B ). Both devices assess a cell’s elastic deformability according to its transit time through the contraction segment using a unitless number referred to as the whole-cell deformability index ( wCDI ): where v c is the cell’s velocity in the contraction segment, v 0 is the cell’s velocity in the reference pore, d 0 is the cell’s initial diameter, and h is the height of the channel.…”

“…Device-to-device variability can arise from manufacturing, resulting in differences in channel geometry (e.g., channel height and width) that in turn may affect flow rates through a given device. We mechanically phenotyped small samples of AP-1060 cells (127 ≤ n ≤ 184), determining the wCDI and recovery time constant using seven different mechano-NPS devices as described by Li et al [ 6 ] ( Fig 1B ).…”

“…For typical mechano-NPS experiments, the data from replicate devices are pooled before comparisons are made among experimental conditions [ 5 , 6 ]. The discrepancy among wCDI data from the different devices highlights the importance of this data pooling, which reduces the influence of device-to-device variability.…”

“…More recently, there has been increasing research on the use of “mechanical phenotyping” to identify and screen single cells for mechanical properties associated with malignancies [ 3 – 5 ]. In this work, we focus specifically on the use of mechano-node-pore sensing (mechano-NPS) to analyze cells from breast epithelial tissue and from a drug-resistant leukemia cell line [ 5 , 6 ]. To extract mechanical information from cells, mechano-NPS uses a four-terminal measurement across a microchannel to track cell transit times as they are deformed by a narrow constriction [ 7 , 8 ] ( Fig 1A ).…”

“…First, to characterize device-to-device variability, we quantified the differences in mechano-NPS results when a single biological sample is tested across multiple replicate devices with relatively low sample sizes. Next, we examined the intra- and inter-user reliability of the semi-manual mechano-NPS data processing pipeline by analyzing results obtained by several researchers using our MATLAB command-line interface (CLI) [ 6 ] to process identical mechano-NPS raw data sets. This CLI replaces a time-consuming, fully manual data processing pipeline that took, at minimum, one minute to find and analyze a cell.…”