Chemoresistance is mediated by ovarian cancer leader cells in vitro

Abstract

Background: Leader cells are a subset of cancer cells that coordinate the complex cell-cell and cell-matrix interactions required for ovarian cancer migration, invasion, tumour deposition and are negatively associated with progression-free survival and response to therapy. Emerging evidence suggests leader cells may be enriched in response to chemotherapy, underlying disease recurrence following treatment.

Methods: CRISPR was used to insert a bicistronic T2A-GFP cassette under the native KRT14 (leader cell) promoter. 2D and 3D drug screens were completed in the presence of chemotherapies used in ovarian cancer management. Leader cell; proliferative (Ki67); and apoptotic status (Cleaved Caspase 3) were defined by live cell imaging and flow cytometry. Quantitative real-time PCR defined "stemness" profiles. Proliferation was assessed on the xCELLigence real time cell analyser. Statistical Analysis was performed using unpaired non-parametric t-tests or one-way ANOVA and Tukey's multiple comparison post hoc.

Results: Leader cells represent a transcriptionally plastic subpopulation of ovarian cancer cells that arise independently of cell division or DNA replication, and exhibit a "stemness" profile that does not correlate with epithelial-to-mesenchymal transition. Chemotherapeutics increased apoptosis-resistant leader cells in vitro, who retained motility and expressed known chemo-resistance markers including ALDH1, Twist and CD44v6. Functional impairment of leader cells restored chemosensitivity, with leader cell-deficient lines failing to recover following chemotherapeutic intervention.

Conclusions: Our data demonstrate that ovarian cancer leader cells are resistant to a diverse array of chemotherapeutic agents, and are likely to play a critical role in the recurrence of chemo-resistant disease as drivers of poor treatment outcomes.

Keywords: Chemo-resistance; Keratin-14; Leader cells; Ovarian Cancer; Recurrence.

Fig. 1

Generation of KRT14-T2A-GFP leader cell lines. A Schematic illustration of the KRT14-T2A-GFP construct. B COV362.4, SKOV3, and OVCAR4-KRT14-T2A-GFP cells, LCs marked by GFP, all cells marked by RFP. C KRT14-T2A-GFP lines were fixed and stained with a polyclonal KRT14 antibody and showed positive co-staining. Scale bar represents 300 μm. D Homozygous KRT14-T2A-GFP cells were transfected with constructs: “KRT14^KO” - targeting exon 1 of the KRT14 gene or the “non-targeting” lentiCRISPR V1 control alone. Cells were imaged at 4X magnification for GFP and RFP 72-h post transfection. E LC^GFP+ counts in KRT14^KO CRISPR targeted lines demonstrated a significant reduction of GFP expression compared to the non-targeting lentiCRISPR V1 control by unpaired t-test and Welch’s correction (p < 0.001) (n = 4 replicates from one representative experiment). Scale represents 100 μm

Fig. 2

Leader cells do not represent a fixed lineage. A LC-T2A-GFP lines were single cell sorted based on LC^GFP+/− status into a single well of a half area 96-well plate and imaged daily at 4X magnification under bright field, for GFP status using the Cytation™ 3 Multimode Imager, scale represents 1000 μm. B LC-T2A-GFP lines were labelled with the pro-dye CellTrace™ Blue (CTB) and routinely imaged for LC^GFP+ and CTB status over a seven-day period using the Cytation™ 3 Multimode Imager. Data are representative of duplicate experiments with 4 imaging areas per triplicate well and are presented as the mean. Representative data from COV362.4-T2A-GFP cells are shown where n = 3 from two experimental replicates. C Flow cytometric analysis of Ki67 surface expression in LC^+/− populations. Cells were seeded at 300,000 cells/well in a 6-well plate, incubated for 18 h, collected and stained for Ki67 using a Ki67-BV786 antibody. GFP (indicative of KRT14 expression) and BV786 fluorescence was acquired using the BD LSRFortessa™ X-20 and data was analysed using FlowJo software (v10.5.0). Analysis was performed using an unpaired non-parametric Mann-Whitney U test to determine statistical significance between groups; n = 2–4/cell line; one representative cell line (SKOV3) is shown (error bars indicate n = 2 replicates); ns = not significant

Fig. 3

Ovarian cancer Leader cells are enriched in response chemotherapeutics commonly used to treat disease. LC-T2A-GFP cells were seeded as a: A-I monolayer; 10,000 cells/single well of a 96-well plate adhered and serum starved overnight or J-S spheroids; 2500 cells/single well in 96-well ultra-low adhesion plates; once formed, spheres were transferred into 96-well flat-bottom imaging plates using wide bore tips. Cells were subsequently treated with olaparib (B, K), rucaparib (C, L), cyclophosphamide (D, M), cisplatin (E, N), paclitaxel (F, O), doxorubicin (G, P), carboplatin (H, Q), and topotecan (I, R) in spheroid and monolayer format for 72 h, with untreated controls (A, J) and a representative spheroid at assay commencement shown (S). 72 h post treatment, duplicate wells were imaged on RFP and GFP channels at 4X magnification to determine LC^GFP+ status and spheroid outgrowth using the Cytation™ 3 Multimode Imager. Spheroids were imaged on 5 focal planes (Z-stacked) and analysis was performed on focus-stacked images. Scale bar 1000 μm

Fig. 4

Ovarian cancer Leader cells represent a population of cells resistant to chemotherapy induced apoptosis. A LC-T2A-GFP cells were treated with the IC⁵⁰ dose of cisplatin and sorted by flow cytometry. Cells were gated based on positive and negative controls where cisplatin treatment increased the percentage of the GFP⁺ LCs significantly. LC and FC percentages are expressed as the percentage of live cells. B LC-T2A-GFP cells were seeded as a monolayer on glass coverslips and treated with IC⁵⁰ concentrations of cisplatin for 48 h. Cells were fixed and stained with the apoptosis maker, Cleaved Caspase-3 and imaged for GFP, Cleaved Caspase-3 and DAPI using the Cytation™ 3 Multimode Imager. C Data was analysed using the Gen5 software (V3.04, Biotek) to calculate the total cell counts, percentage of LCs (GFP⁺), apoptotic cells (cleaved caspase-3⁺) and double positive cell numbers. Statistical analysis was performed using a paired t-test; n = 4 focal areas per well. Scale bar at 200 μm

Fig. 5

Representative immunohistochemical staining of KRT14 in A HGS ovary pre-treatment and B neo adjuvant ovary at 20X magnification, scale 100 μm. C Positive pixel staining count (p = 0.0047, Mann-Whitney U test)

Fig. 6

Leader cells proliferate at the same rate in chemotherapy and do not arise due to cellular division and DNA replication, leader cell deficient lines are more susceptible to cisplatin treatment. A Flow cytometric analysis of Ki67 surface expression in LC^+/− populations in ovarian cancer cell lines in the presence of cisplatin. B Cells were seeded at 300,000 cells/well in a 6-well plate, incubated for 18 h, and treated with 13 μg/ml cisplatin +/− aphidicolin (0.5 μg/ml) or tunicamycin (6 μg/ml) for 24 h. Following treatment, cells were collected and stained for Ki67 using a Ki67-BV786 antibody. GFP (indicative of KRT14 expression) and BV786 fluorescence was acquired using the BD LSRFortessa™ X-20 with data analysed using the FlowJo software (v10.5.0). C Analysis was performed using a non-parametric Mann-Whitney U test to determine statistical significance between groups; n = 2–4/cell line; ns = not significant. D Wild-type, LC-deficient (KRT14^KO) and LC-enriched (KRT14^OE) cells were seeded at 0.2 × 10⁴ cells/0.1 ml/well in 96-well plates. Cells were synchronized in G₀ by overnight incubation in serum-free media prior to treatment and treated with cisplatin concentrations ranging from 1.2–30 μg/ml for 48 h followed by the addition of cell viability dye alamarBlue™ (Invitrogen). Half maximal inhibitory (IC⁵⁰) concentrations of cisplatin were calculated in GraphPad Prism (v9.0) and analysed by non-linear regression fit. E Long-term assessment of cellular proliferation in response to high doses of Cisplatin where 0.2 × 10⁴ cells/0.1 ml/well wild-type, LC deficient (KRT14^KO) and LC enriched (KRT14^OE) cells were seeded onto RTCA xCELLigence E-plates and treated with 30 μg/ml cisplatin, cellular proliferation was monitored over 240-h and impedance readings taken every 15 min

Fig. 7

Leader cells express markers indicative of a “stemness” profile. Cells were sorted by flow cytometry based upon LC status, mRNA was extracted and cDNA synthesized. Gene expression of A KRT14, B CD44v6, C ALDHI, D Twist, E WNT and F Nanog was analyzed by quantitative real-time PCR. Data was normalised to ribosomal 18S and graphed as the mean value ± SEM with a single representative real time experiment shown. Analysis was performed by one-way ANOVA and Tukey’s multiple comparison post hoc test to determine statistically significant differences between groups (* p < 0.05, ** p < 0.01) (n = 3 separate isolations in a single run)