Bioinformatics, Computational and Systems Biology
Tyler Johnston (he/him/his)
Undergraduate Researcher
University of Pittsburgh Medical Center
Canonsburg, Pennsylvania, United States
Recurrence in cancer is a complication all too familiar to many oncologic patients. Of the 80% of ovarian cancer patients who achieve remission through diamminedichloroplatinum(II) (cisplatin) treatment, 70% see recurrent tumors. Though effective in the first round of treatment, chemotherapies like cisplatin often lose their potency in subsequent phases of treatment due to resistance developed in recurrent tumors. This often leads to terminal outcomes in recurrent patients, as seen by the 50% decrease in survival rate from 1 to 5 years after diagnosis with ovarian cancer.
Currently, the vehicle in which cell fate decisions are made after interaction with chemotherapy is poorly understood. It is unknown how a cell chooses between cell-mediated death versus a continued state of proliferation. Due to proliferation’s dependence on the cell cycle, studying the cell cycle of cancer could lead to further understanding of this decision—and thus, a way to modulate this decision, leading to more effective treatments.
The aim of this experiment is to produce an intuitive, interactive map of the ovarian cancer cell cycle using highly multiplexed immunofluorescence to obtain a detailed molecular signature for each cell following drug treatment followed by a non-linear dimensionality-reduction approach to project these data onto a 2-d map. Combining datasets of populations from various treatment conditions will directly visualize the effect of cisplatin on the cell cycle and can explain the timing and mechanism of how cancer can survive the results of cisplatin treatment.
The OVCAR-8 cell line used throughout this experiment was provided in partnership with the Aird Laboratory of UPMC Hillman Cancer Center. 8-well chamber slides were prepared with cells being treated with either 0.2 or 1 μM of cisplatin over a 3- or 5-day time span. This was followed by chemical fixation with 4% paraformaldehyde and permeabilization with 0.1% Triton X-100. The slides were mounted in mounting media after Hoechst staining for multiplex immunofluorescence imaging-- the iterative staining of fluorescently-tagged antibodies to directly measure the level and localization of over 25 proteins within the populations. The CellDIVE Multiplexed Imager was used to produce this raw data.
Data from raw images was processed through a custom Cellpose segmentation method, optimized for the OVCAR-8 cell line treatment with chemotherapy. Analysis of derived data was performed using Potential of Heat-diffusion for Affinity-based Trajectory Embedding (PHATE), providing an intuitive, low-dimensional trajectory for the cell populations.
A cell cycle map was produced from combining three separate treatment regimens: one with no treatment, and two with a high dose of treatment over a 3- and 5-day period.
Progression through the canonical cell cycle phases was inferred by the characteristic dynamics of cyclin proteins. While G1 was singular and characterized by a clear high Cyclin D1 state, other phases identified through cyclin dynamics split into two states populated completely by either treated or untreated cells. Within the treated trajectory, cell size and DNA content was notably elevated compared to the non-treated group. A ridge near the outset of G1-phase and beginning of S-phase was observed to have low cyclin content, thus a lack of propulsion through the cell cycle, and has considerable levels of cell cycle arrest markers.
Along both M-phase trajectories, a population of cells was identified as apoptotic through visual confirmation when viewing image data. This region also contained considerably higher levels of most cyclins, uncharacteristic of the Cyclin B-centered nature of M-phase. A higher level of treated cells resides in this apoptotic area than non-treated, which is consistent with the apoptosis-inducing properties of cisplatin in high doses.
This initial production of a cell cycle map, consistent with the already modeled cell cycle, provides preliminary insight to the response of cancer to chemotherapies. After interaction with cisplatin, the behavior of treated populations to reside mainly in the later phases of the division process could provide more insight to the mechanisms governing oncologic response to chemotherapy.
Further investigations into alternative oncologic treatments, such as PARP inhibitors, could prove useful in exploring alternative mechanisms regarding treatment response. Other factors such as cell density, drug concentration, and treatment duration will also be explored in the next steps of this experiment—hopefully to provide a variety of trajectories visualizing response to clinical treatment.