Personalized Medicine-Based Approach to Model Patterns of Chemoresistance and Tumor Recurrence Using Ovarian Cancer Stem Cell Spheroids

Abstract

Purpose: Chemoresistant ovarian cancers grow in suspension within the ascites fluid. To screen the effect of chemotherapeutics and biologics on resistant ovarian cancers with a personalized basis, we developed a 3D hanging drop spheroid platform.Experimental Design: We initiated spheroids with primary aldehyde dehydrogenase-positive (ALDH⁺) CD133⁺ ovarian cancer stem cells (OvCSC) from different patient samples and demonstrated that stem cell progeny from harvested spheroids was similar to the primary tumor. OvCSC spheroids were utilized to initiate tumors in immunodeficient mice. Drug responses to cisplatin and ALDH-targeting compound or JAK2 inhibitor determined whether the OvCSC population within the spheroids could be targeted. Cells that escaped therapy were isolated and used to initiate new spheroids and model tumor reemergence in a personalized manner.Results: OvCSC spheroids from different patients exhibited varying and personalized responses to chemotherapeutics. Xenografts were established from OvCSC spheroids, even with a single spheroid. Distinct responses to therapy were observed in distinct primary tumor xenografts similar to those observed in spheroids. Spheroids resistant to cisplatin/ALDH inhibitor therapy had persistent, albeit lower ALDH expression and complete loss of CD133 expression, whereas those resistant to cisplatin/JAK2 inhibitor therapy were enriched for ALDH⁺ cells.Conclusions: Our 3D hanging drop suspension platform can be used to propagate primary OvCSCs that represent individual patient tumors effectively by differentiating in vitro and initiating tumors in mice. Therefore, our platform can be used to study cancer stem cell biology and model tumor reemergence to identify new targeted therapeutics from an effective personalized medicine standpoint. Clin Cancer Res; 23(22); 6934-45. ©2017 AACR.

Figure 1:. Spheroids generated from malignant ascites-derived OvCSC maintain OvCSC heterogeneity

(A, B, C): OvCSCs were isolated using fluorescence-activated cell sorting from malignant Pt259, Pt224, and Pt152 ascites based on the concurrent elevated activity of aldehyde dehydrogenase (ALDH) and CD133 expression. (D, E, F): Spheroids were initiated in hanging drop arrays using 10 Pt259 OvCSCs, and observed using phase contrast microscopy at Day 7 (scale bar = 100µm). (G, H, I): Following spheroid culture for 7 days, harvested spheroids were processed for flow analysis to identify OvCSC populations in Pt259 OvCSC spheroids – several populations were identified, ALDH⁺, CD133⁺, ALDH^-/CD133^- and ALDH⁺/CD133⁺. OvCSC populations within spheroids mimic the primary patient profile. (J, K, L): Quantification of flow analyses indicated that spheroid culture maintained or enhanced OvCSC populations to similar extents as observed in the primary samples.

Figure 2:. Patient derived OvCSC spheroids exhibit differential response to cisplatin/673A drug treatment

(A): Spheroids were initiated from 10 OvCSCs/drop in hanging drop arrays, and allowed to form over 7 days in serum-free culture. On Day 7, spheroids were treated with a single or combination dose of cisplatin 50µM/compound673A 50µM. Control spheroids were untreated, but maintained for the same period in culture. Phase contrast micrographs at 72 hours following drug treatment demonstrate that control untreated spheroids remain compact, while drug treated spheroids are either progressively smaller and/or appear to have necrotic foci within them (scale bar = 100µm). (B) Alamarblue fluorescence was used to assess viability of drug treated spheroids, indicating that while all OvCSC spheroids were responsive to cisplatin and/or 673A, their chemosensitivity varied from patient to patient. Pt259 OvCSC spheroids were most sensitive to drug treatment, and Pt152 was most resistant to drug treatment; *p<0.05 two-way ANOVA, on all drug treated samples compared to control untreated spheroids. (C): Representative flow cytometry plots on CSC populations within cisplatin/673A treated spheroids from all three patient samples. (D): Summary of flow analysis data indicated that cisplatin/673A treatment targeted ALDH⁺ populations routinely in Pt259 and Pt224, and also additionally targeted CD133⁺ populations in Pt152; ***p<0.001, one-way ANOVA, between control untreated spheroids and cisplatin/673A treated spheroids.

Figure 3:. Patient derived OvCSC spheroids exhibit differential response to cisplatin/ruxolitinib drug treatment

(A) Phase contrast micrographs of control untreated and cisplatin 50µM/ ruxolitinib 500nM treated spheroids from three different patient samples demonstrate that drug treated spheroids have reduced spheroid sizes and cell death can be visually observed (scale bar = 100µm). (B): Viability measurements using alamarblue fluorescence indicated that while all OvCSC spheroids were responsive to cisplatin and/or ruxolitinib, their chemosensitivity varied between patient samples; *p<0.05, two-way ANOVA, on drug treated spheroids compared to control untreated spheroids. (C): Representative flow cytometry plots demonstrate CSC populations remnant in drug treated spheroids. (D): Graphical summary of flow analysis of OvCSC populations in cisplatin/ruxolitinib treated versus control untreated spheroids, indicating that cisplatin/ruxolitinib treatment effectively targeted CD133⁺ populations routinely in all patient samples; *p<0.05, ***p<0.001, one-way ANOVA, between control untreated spheroids and cisplatin/ruxolitinib treated spheroids. (E): Immunoblot analysis of phosphorylated STAT3 (pSTAT3), compared to total STAT3 (tSTAT3) in control untreated (No Tx) or ruxolitinib 500nM treated (Rux) OvCSC spheroids. Against a GAPDH loading control, STAT3 phosphorylation is inhibited with ruxolitinib treatment in Pt259 and Pt152 OvCSC spheroids.

Figure 4:. Remnant viable cells cause re-emergence of ovarian tumor and CSC populations in cisplatin/673A treated OvCSC spheroids

(A) Phase contrast micrographs of spheroids generated from viable cells surviving cisplatin/673A treatment. Insets show spheroids at Day 1 following cell seeding (scale bar = 100µm). (B): Representative flow cytometry plots of OvCSC populations in the re-emerging spheroids at Day 7, following cisplatin/673A treatment. (C): Quantification of proliferation within spheroids over 7 days, formed from viable cells after cisplatin/673 treatment, indicating robust proliferation and recovery following cisplatin/673A therapy. No statistical significance was observed in the proliferation rates between patient samples. (D): Graphical quantification of OvCSC populations derived from flow analysis, indicating the presence of ALDH⁺ populations, minimal to no recovery of ALDH⁺/CD133⁺ or CD133⁺ populations. (E): Flow analysis of serially passaged recovered Pt259 OvCSC spheroids indicates an increasing expression of ALDH⁺, CD133⁺ and CD133⁺/ALDH⁺ populations, over seven generations. OvCSC populations approach and even seem to surpass original populations. (F): Flow analysis of serially passaged recovered Pt152 OvCSC spheroids demonstrates an increase in ALDH⁺, CD133⁺ and CD133⁺/ALDH⁺ OvCSC progeny. CD133⁺ progeny do not approach original levels in Pt152 spheroids even after seven generations.

Figure 5:. Patient malignant ascites-derived OvCSC spheroids initiate tumors in xenografts

(A): Macroscopic pictures of tumors observed in NSG mice with subcutaneous injections of 5000 OvCSCs or 10 injected spheroids. (B): Photomicrographs of hematoxylin and eosin stained 6 µM sections, obtained from primary patient debulking specimens and OvCSC derived tumor xenograft from the sample patient (Primary patient scale bar = 200µm; OvCSC derived tumor xenograft scale bar = 100µm). (C): Tumor initiation for the varying experimental conditions, indicating that 500 injected spheroids initiated tumors in less than a week of injection, while 50 injected spheroids, and 5000 OvCSCs initiated tumors within 7 days. (D): Comparison of the growth rate and tumor volumes of tumors initiated using 1–500 spheroids or 5000 freshly isolated OvCSCs. All groups contained 4–8 injected tumors. (E): Phase contrast micrographs of spheroids initiated from tumor cells isolated from xenografted tumors, using 50 cells/drop (scale bar = 100µm). (F): Follow up phase contrast micrographs of spheroids at Day 7, indicating that spheroid forming abilities were retained even upon xenograft (scale bar = 100µm).

Figure 6:. Drug response of OvCSC xenografts match drug response of patient malignant ascites-derived OvCSC spheroids treated with cisplatin/673A

(A): Tumors were initiated from Pt259 and Pt152 spheroids using 10 injected spheroids. Once palpable tumors were obtained, control groups received a sham saline injection, while the experimental drug group received cisplatin/673A. Tumor volumes of control (black trace) and drug treated (red trace) Pt259 xenografts are plotted, indicating a decrease in volume of drug treated Pt259 xenografts compared to control. Shaded gray area indicates the window for tumor burden (1500–2000mm³) upon which mice were sacrificed. (B): Hematoxylin and eosin and proliferative index marker Ki67 staining of control Pt259 xenograft tumor cross-sections (scale bar = 100µm). (C): H&E staining and cleaved caspase 3 staining of drug treated Pt259 xenografts indicating cell death in the cisplatin/673A treated tumors (scale bar = 100µm). (D): Representative flow analysis plots for control untreated and drug treated tumors, where drug treated tumors demonstrate reduced OvCSC populations. (E) Tumor volumes of control (black trace) and drug treated (red trace) Pt152 tumors, where there is minimal difference in tumor volumes between the two experimental conditions. Shaded gray area indicates the window for tumor burden (1500–2000mm³) upon which mice were sacrificed. (F) H&E and Ki67 staining of Pt152 xenografts visually demonstrate no difference in proliferative index when compared to Pt259 xenografts (scale bar = 100µm). (G): Hematoxylin and eosin and cleaved caspase-3 staining of Pt152 xenograft tumor cross sections indicate minimal cell death in cisplatin/673A treated tumors compared to control untreated tumors (scale bar = 100µm). (H): Representative flow analysis plots for drug treated and control untreated tumors, where marked losses of CD133⁺ and ALDH⁺ CD133⁺ populations are observed.