Ovarian cancer cell lines including SKOV3, OVCAR3, CaOV3, Kuramochi and OAW28 were purchased from the American Type Culture Collection (ATCC). OVCAR4 and OVCAR8 cell lines were obtained from the National Cancer Institute Division of Cancer Treatment and Diagnosis (NCI/DCTD) repository through Material Transfer Agreement (MTA). Cell lines were grown in recommended media (RPMI supplemented with 10% FBS or DMEM supplemented with 10% FBS) in a 5% CO₂ humidified incubator at 37°C and used within 10 passages. All cell lines used in this study were frequently authenticated by STR analysis. Primary patient tumor samples were obtained from patients after providing informed consent through protocols approved by the UCLA Office of the Human Research Protection Program (IRB# 10-000727). Patients ages ranged from 43 to 75 years. Tumor samples were mechanically and enzymatically dissociated with 1 mg/ml collagenase and dispase solution (Gibco; Thermo Fisher Scientific, Inc.) and cryopreserved in buffer (90% FBS, 10% DMSO) for experimental use.

As previously described (28,29), in this bioassay, 5,000 cells/well were suspended in Matrigel matrix (Corning Inc.) and MammoCult growth medium (STEMCELL Technologies) and plated around the rim of the wells of a 96-well plate. Organoids were allowed to grow for 1-2 days followed by drug treatment in a dose-dependent manner for 3 consecutive days. After drug treatment, cell viability was determined using an ATP luminescence assay (CellTiter-Glo 3D viability Assay kit; Promega Corp.). For testing platinum sensitivity, each cell line was treated with carboplatin at increasing concentrations of 0, 10, 25, 50, 100, 150, 200 and 250 µM. For co-therapy testing, cells were treated with a range of doses of carboplatin (0-50 µM) and birinapant (0-50 nM).

As a measure for quality control, cryopreserved primary tumor samples with cell viability less than 60% after thawing were excluded from the study. Only those primary samples that formed visible organoids in the 3D organoid bioassay were included in this study, confirmed by examination under a microscope. Assay results were rejected if the luminescence value in vehicle-treated wells were found to be similar to the blank control [Matrigel and growth media]. Staurosporine-treated cells were used as a positive control for cell death in each experiment.

Synergy between birinapant and carboplatin was quantified using the Loewe additivity index model available in a web-based package SynergyFinder 2.0 (30). Synergy scores were calculated across all tested drug concentrations and combinations and visualized as a two-dimensional interaction surface over the dose matrix (Loewe synergy plots). Based on the Loewe additivity model, a summary synergy score was generated. Drug interactions were classified as likely synergistic (score >10), antagonistic (score <-10), or additive (score -10 to +10). Synergy scores shown in this study were obtained from the SynergyFinder 2.0 tool as of 10/04/2021.

Information regarding the genetic alterations in cIAP1, cIAP2 and XIAP genes were obtained from the Cancer Genome Atlas (TCGA) database through cBioPortal (http://www.cbioportal.org). Data shown in this study were extracted from PanCancer datasets as of 10/04/2021. We focused on TCGA data for 8 cancer types known to be treated with platinum-based chemotherapy (31-34).

For assessing the endogenous levels of IAP proteins in 7 EOC cell lines, growing cells were harvested, lysed using RIPA lysis buffer supplemented with protease inhibitor cocktail (Thermo Fisher Scientific, Inc.). Cell lysates were centrifuged at 12,000 × g for 10 min at 4°C. The supernatant was collected, and protein concentration was measured using a BCA protein assay kit (Thermo Fisher Scientific, Inc.). Equal amounts (40 µg) of protein were loaded in each lane and resolved on NuPAGE 4-12% Bis-Tris gels (Thermo Fisher Scientific, Inc.). Resolved proteins were transferred to nitrocellulose membranes (Millipore), probed with appropriate antibodies and detected by chemiluminescent reagent (Millipore, WBKLS0500). Protein bands were visualized using a Bio-Rad ChemiDoc Imaging system (Bio-Rad Laboratories).

For measuring the IAP protein levels in SKOV3 and OVCAR8 cells after co-therapy treatment, both cell lines were treated in 2D cell culture with carboplatin alone, birinapant alone, or the combination of the two drugs at a concentration corresponding to the 48 h half maximal inhibitory concentration (IC₅₀) of each drug. After 24 h of drug treatment, cells were lysed and processed the same way as mentioned above. For measuring IAP protein levels in PDX subcuticular tumors, an additional step of sonication (Thermo Fisher Scientific sonicator) was performed after tumors were lysed in RIPA lysis buffer.

Antibodies used in this study included: cIAP1 [Cell Signaling Technology, Inc. (CST), cat. no. 7065], cIAP2 (CST, cat. no. 3130), XIAP (CST, cat. no. 14334), GAPDH (CST, cat. no. 5174), Ki67 (Agilent Technologies, cat. no. M724001-2), and Pax8 (Proteintech, cat. no. 10336-1-AP). For western blot experiments, cIAP1, cIAP2, XIAP and GAPDH antibodies were used at a dilution of 1:1,000. For IHC experiments Ki67 and Pax8 antibodies were used at a dilution of 1:100 and 1:500 respectively.

Apoptosis was detected by Annexin V and PI staining using BD Pharmingen Apoptosis Detection Kit (cat. no. 556547). Cells (0.5 million/well) were seeded in 6-well plates. Cells were treated with various concentration of carboplatin, birinapant, or the co-therapy for 72 h. Cells were then harvested and stained with Annexin V/PI following the manufacturer's protocol. Cell death was analyzed using a BD FACS ARIA flow cytometer and FlowJo software (version 10.8.0; BD Biosciences). Cell death was represented as the percentage of Annexin V⁺ cell population after drug treatment. For each cell line, three independent experiments were performed.

Cells were seeded at a density of 0.5 million cells/well in a 6-well plate and allowed to grow until approximately 80% confluency was reached. Cells were then washed with 1X PBS, incubated with fresh media, and treated with carboplatin, birinapant, or co-therapy at a concentration corresponding to the 48 h IC₅₀ of each drug. After 72 h of drug treatment, 100 µl of culture supernatant was collected from each treatment group for TNFα measurement by ELISA kit (Thermo Fisher Scientific, Inc., cat. no. BMS223HS). Recombinant TNFα protein was used as a positive control in these experiments. Two independent experiments were performed with duplicate wells for each condition.

Cells were seeded at a density of 5,000 cells/well in 96-well plate. Cells were then pre-treated with 10 µg/ml of anti-TNFα antibody (R&D Systems, cat. no. MAB610-100) for 2 h followed by addition of carboplatin, birinapant, or combination of two drugs at the 48 h IC₅₀ of each drug. After 72 h of drug treatment, cell viability was measured using an ATP luminescence assay (CellTiter-Glo 3D viability Assay kit, Promega Corp.). IgG antibody was used as a non-specific antibody control. Two independent experiments were performed with triplicate wells for each condition in each cell line tested.

PDX models were established from a primary pleural effusion sample, EOC2, clinically classified as recurrent, platinum-resistant high-grade serous ovarian cancer. For PDX generation, 1 million cryopreserved cells were injected as a cell suspension in the intraperitoneal space of a female NSG mouse. The cells were re-passaged in mice twice. At the end of the second passage, PDX cells were harvested and evaluated by histological and genomic analyses. Short tandem repeat (STR) analyses performed using extracted genomic DNA from PDX cells demonstrated that this PDX retained greater than 90% identity to its parental tumor. Single nucleotide polymorphism analysis was also performed, confirming relatedness of PDX cells and parental tumor cells.

Histologic slides containing PDX subcuticular tumor fragments from each cohort after drug treatment were reviewed by a gynecology pathologist NAM. PAX8 immunostaining was evaluated for nuclear expression to confirm Mullerian origin of the tumors. Ki67 was also evaluated for nuclear staining for proliferation index. Scoring was based on the estimation of the percentage of tumor cells stained by the antibody visualized by light microscopy. The mitotic figures were counted on hematoxylin and eosin stained slides on 10 high power fields (40× objective). Mean number of mitosis between the treatment groups was compared by ANOVA with non-constant variance allowed.

All animal experiments were approved by the UCLA Animal Research Committee (protocol 2008-153) and conducted under the supervision of the UCLA Division of Laboratory Animal Medicine. Six- to eight-week-old female NSG mice (NOD.Cg-Prkd^scid//2rg^tm1Wjl/SzJ) were purchased from Jackson Laboratories for all in vivo experiments. All mice were housed in specific pathogen-free (SPF) facilities. They were kept in autoclaved cages with sterile bedding and food. Mice were anesthetized by isoflurane gas in oxygen. Following the Institutional Animal Care and Use Committee (IACUC) guidelines, 3-5% isoflurane was used for initial induction and 1-3% was used for maintenance. For euthanization, mice were exposed to gradually increasing concentrations of CO₂ at a flow rate ranging between 30-70% chamber volume/min as per the American Veterinary Medical Association (AVMA) guidelines. Cervical dislocation was used as a confirmatory method of euthanasia after CO₂ exposure. At the time of euthanization, tumor volume of each mouse was <1,000 mm³ consistent with (IACUC) guidelines.

Results are reported as means ± standard error. For normally distributed data, the P-values for comparing means were calculated using one-way analysis of variance (ANOVA), two-way ANOVA or two way (mixed) repeated measure ANOVA (tumor mean comparison) when animals were measured repeatedly over time. When data were not normal as indicated by the Shapiro-Wilks test, P-values were computed with non-parametric Kruskal-Wallis test. Calculations were performed using R version 4.0.5 (R Foundation for Statistical Computing, Vienna, Austria).

Tumor volumes were calculated using the modified ellipsoid formula: 1/2 × length × (width)². The baseline volume on day 0 (start of drug treatment) was subtracted from all subsequent tumor volumes of the same animal since day 0 baseline was not the same for all animals. After this standardization, the relation between tumor volume and time (day) was approximately linear in each animal. Thus, the growth rate in mm³ per day for each animal was computed via linear regression on each animal and the mean rates were compared among the four groups using a one-way ANOVA where the variances were allowed to be heterogeneous.

One of the challenges in the treatment of OC is the absence of biomarkers that could prospectively predict patients' response to standard platinum-based chemotherapy. OC tumors are characterized as platinum-sensitive or platinum-resistant only after chemotherapy drugs are administered. Hence, in vitro drug response assays that are designed to test and predict the sensitivity of patient tumor cells to existing chemotherapies may provide a useful tool for treatment of OC patients. Although these assays hold great promise in the field of precision medicine, they have not yet proven efficacious for clinical application and therefore, no such assay has been FDA approved for clinical use at this time (35). In the present study, a 3D organoid-based drug testing platform (28) was utilized to assess the response of OC cell lines to carboplatin. Platinum sensitivity of the OC cell lines measured in this bioassay were correlated with reported response to platinum drugs for each cell line (36,37).

In this bioassay (28,29), OC cells were grown as 3D organoids embedded within an extracellular matrix hydrogel called Matrigel (Fig. 1A). On the first day of this bioassay, a mixture of cells and Matrigel were plated around the rims of wells in a 96-well plate and overlaid with growth media. Cells were allowed to generate organoids for 24 h. Growing organoids were treated with increasing concentrations of carboplatin (0-250 µM) for three consecutive days, with daily drug replenishment. After drug treatment, the viability of the cells was measured using an ATP-based luminescence assay (CellTiter Glo, Promega Corp.). Sensitivity to carboplatin was determined using an IC₅₀ value, defined as the drug concentration causing 50% inhibition. IC₅₀ values for each cell line were obtained from carboplatin dose response curves generated using the ATP-based viability assay.

The viability plots for each cell line demonstrated a dose-dependent decrease in cell viability. Differential sensitivity to carboplatin was noted in all cell lines tested (Fig. 1B). Based on the IC₅₀ of carboplatin, OVCAR3 cells were found to be carboplatin sensitive (IC₅₀ <40 µM), Kuramochi and OVCAR8 cells were highly carboplatin resistant (IC₅₀ >85 µM). The remaining cell lines exhibited intermediate carboplatin resistance (Fig. 1C). Platinum sensitivity measured with this bioassay correlated with platinum sensitivity of each cell line reported by others (36,37).

This 3D organoid bioassay was capable of classifying cell lines as carboplatin-sensitive vs. carboplatin-resistant. Since organoids are known to better recapitulate in vivo tumor architecture (38), this bioassay may provide a tool for testing single drug or combination therapies in ovarian cancer. Therefore, next this bioassay was utilized to test the efficacy of a combination therapy in targeting platinum resistant ovarian cancer cells.

Cancer cells are known to overexpress highly conserved anti-apoptotic proteins called inhibitor of apoptosis (IAP) proteins (9). These proteins are predominantly known for inhibiting apoptotic cell death via regulation of caspases and are reported to be involved in tumor cell survival, chemo-resistance, disease progression, and poor prognosis [reviewed in LaCasse et al (39)]. Data from The Cancer Genome Atlas (TCGA) database also demonstrated that gene amplification is a common alteration in key IAPs, namely cIAP1, cIAP2, and XIAP in ovarian cancer and other cancers treated with platinum-based chemotherapy drugs (31-34) (Fig. S1A). Taken together, this evidence provides a rationale to activate apoptosis in platinum-resistant ovarian cancer cells by targeting IAP proteins. We combined birinapant, a small molecule that mimics natural IAP protein antagonist SMAC, with carboplatin to target platinum-resistant OC cell lines in vitro. Due to its pro-apoptotic activity and tolerability, birinapant and several other SMAC mimetic molecules have been evaluated in early clinical trials as a monotherapy or in combination with other chemotherapeutic drugs to target different types of cancers [reviewed in Morrish et al (21)]. The combination of SMAC mimetics, including birinapant and carboplatin, has also been previously explored and demonstrated efficacy in targeting OC cell lines using preclinical models (26,40). In this study, we expanded on these previous investigations by testing the in vitro efficacy of birinapant and carboplatin combination across a panel of 7 EOC cell lines using the 3D organoid bioassay. This 3D organoid bioassay offers several advantages. First, as a preclinical model for drug testing, it retains complex tumor architecture similar to in vivo tumors (28). Second, it is compatible with automation and high throughput drug screening allowing testing of combination therapies in a scalable and reproducible way (28).

Therapeutic target proteins of birinapant (cIAP1/2 and XIAP) were evaluated in these 7 cell lines using western blot analysis (Fig. S1B). For co-therapy testing, cell line-derived organoids were treated with increasing concentrations of carboplatin (0-50 µm) and birinapant (0-50 nM) in the 3D-organoid bioassay (Fig. 2A). After 3 days of drug treatment, cell viability was measured using an ATP-based assay (Fig. S1C). Drug synergy was scored using the Loewe additivity model available in the online package SynergyFinder 2.0 (30). Four out of the 7 cell lines (OVCAR8, SKOV3, Kuramochi, and OAW28) demonstrated a positive synergy score that indicates increased cell death upon co-treatment with birinapant and carboplatin compared to the sum of the single agents (Fig. 2B and C). The combined effects were found to be likely additive in the OVCAR4 cell line while the effects of these two drugs were found to be likely antagonistic in OVCAR3 and CaOV3 cell lines (Fig. 2B and C). Overall, these in vitro results demonstrated that addition of birinapant to carboplatin treatment enhanced cell death in a subset of OC cell lines.

Both birinapant and carboplatin are known to induce cell death. Birinapant-mediated cell death occurs by inhibition of IAP proteins (21) while carboplatin forms DNA lesions that block DNA synthesis and results in cell death (41). Therefore, the combined effect of carboplatin and birinapant on cellular apoptosis was investigated by flow cytometry. We selected a highly platinum-resistant cell line, OVCAR8, and a moderately platinum resistant cell line, SKOV3, as these two cell lines demonstrated Loewe synergy scores greater than 10 when treated with carboplatin and birinapant combination in vitro (Fig. 2C).

Quantification of drug-induced cell death was facilitated by Annexin V/PI staining using flow cytometry. Both SKOV3 and OVCAR8 cells were treated with carboplatin, birinapant, or the combination of both drugs in 2D cell culture at concentrations corresponding to IC₅₀ values of each drug. After 72 h of drug treatment, cells were harvested, stained with Annexin V/PI and analyzed by flow cytometry (Fig. 3A). The percentage of dead cells (Annexin V⁺) was significantly increased when either cell line was treated with the co-therapy (CP+BP) compared to single agents (CP or BP) (Fig. 3B and C). Degradation of cIAP1 was detected in both cell lines by birinapant alone, or in combination with carboplatin (Fig. S2). However, birinapant had no effect on XIAP expression levels in these cell lines.

Several studies have reported that SMAC mimetics including birinapant stimulate cells to release cytokine TNFα which binds to membrane-bound TNF receptors and triggers an apoptosis-signaling pathway (42,43). The role of TNFα in co-therapy targeted SKOV3 and OVCAR8 cells was examined by measuring the release of TNFα molecules using ELISA. In this experiment, both cell lines were treated with carboplatin (CP), birinapant (BP), or the combination of the two agents (CP+BP) in 2D cell culture at concentrations corresponding to the IC₅₀ of each drug (Fig. 3D). After 72 h of drug treatment, 100 µl of culture supernatant was collected from each treatment group for TNFα measurement by ELISA. Results demonstrated that co-therapy treated OVCAR8 and SKOV3 cells secreted significantly more TNFα compared to carboplatin alone (Fig. 3D, right panel). This suggests that TNFα signaling may also contribute to overall drug synergy observed in these cell lines. To further confirm the role of TNFα in mediating pro-apoptotic signaling, anti-TNFα neutralizing antibody was used to block the interaction between TNFα and its membrane bound TNF receptor. In this experiment, both cell lines were pre-treated with anti-TNFα antibody for 2 h followed by addition of carboplatin (CP), birinapant (BP), or the combination of the two agents (CP+BP) for 72 h (Fig. 3E). After drug treatment, cell viability was measured using an ATP-based luminescence assay. Neutralization of TNFα molecules with anti-TNFα antibody was found to partially reverse co-therapy-induced cytotoxicity. An increase in cell viability was observed in SKOV3 and OVCAR8 cells treated with the co-therapy following pre-treatment with the anti-TNFα antibody (Fig. 3E, right panel).

Overall, the findings demonstrated that apoptosis was induced in the co-therapy-treated platinum-resistant OVCAR8 and SKOV3 cells. It also suggests that mechanisms of drug synergy may be TNFα-dependent in these cell lines.

Molecular alterations that drive OC progression and chemotherapy resistance are highly variable between individual patients. The heterogenous nature of OC tumors poses a challenge in selecting the most effective treatment for individual patients based on their unique tumor traits. This challenge has steered the field of cancer therapeutics towards precision medicine approaches whereby the selection of therapy is tailored to each individual patient's tumor. Exploring the potential of this approach, we utilized the 3D organoid bioassay as a precision medicine tool to test the therapeutic efficacy of carboplatin and birinapant combination in a panel of platinum-resistant primary OC tumor samples (Fig. 4A). A total of 10 platinum-resistant EOC specimens were tested in this study (Fig. 4B).

In this bioassay, dissociated tumor cells either freshly processed or cryopreserved were mixed with Matrigel and plated in tissue culture plates similar to experiments performed using the OC cell lines. Patient-derived organoids were then treated with carboplatin (0-50 µM), birinapant (0-50 nM), or co-therapy for 3 consecutive days followed by assessment of cell viability and synergy score calculation (Fig. S3A). Loewe synergy scores calculated by the SynergyFinder tool demonstrated that co-therapy treatment was likely synergistic in targeting 1/10, additive in 7/10, and antagonistic in 2/10 platinum resistant primary tumor samples (Fig. 4B and C). The IC₅₀ of carboplatin and birinapant was also measured for each sample (Fig. S3B). Clinically classified platinum-resistant tumor samples tested in this bioassay demonstrated high IC₅₀ values of carboplatin (above 40 µM).

To examine the therapeutic efficacy of the co-therapy in vivo, one primary tumor sample that demonstrated maximum Loewe synergy score in vitro (EOC2) was used to generate a platinum-resistant patient-derived xenograft (PDX) model in immunocompromised mice (Figs. 4D and S4A). The efficacy of the co-therapy was confirmed ex vivo in the 3D organoid bioassay. Similar to its parental patient tumor sample (EOC2), PDX cells were found sensitive to co-therapy (Fig. S4B). Levels of IAP proteins in PDX cells were also measured using western blot analysis (Fig. S4C).

To assess the efficacy of birinapant, carboplatin, and the co-therapy in vivo, PDXs were generated in n=17 female NSG mice by subcuticular injection of one million PDX cells/mouse (Fig. S4D). One mouse was randomly selected and euthanized before treatment to confirm establishment of subcuticular tumors (Fig. S4E). The remaining mice were then randomized into treatment groups (n=4/group). Treatment was initiated when the average tumor volume of all PDX-bearing mice reached approximately 125 mm³. Mice were treated with either vehicle, carboplatin (50 mg/kg i.p. 1×/week), birinapant (30 mg/kg i.p. 2×/week), or the combination of both drugs for 4 weeks. In the co-therapy-treated mice, birinapant was administered 4 h before carboplatin. During the treatment period, the mouse body weight and size of tumors were recorded twice a week. After 4 weeks of drug treatment, mice were euthanized, and tumors were harvested.

Tumor volume of each mouse was adjusted by subtracting the baseline tumor volume as measured at the start of treatment (day 0). At the end of treatment (day 25), the smallest mean tumor volume (adjusted to baseline) was observed in the co-therapy-treated mice compared to vehicle, carboplatin-treated or birinapant-treated group, with mean difference statistically significant with the vehicle-treated mice only (P=0.0390) (Fig. 4E).

Comparison of the mean tumor growth rates between treatment groups demonstrated that the mean tumor volume in the vehicle and carboplatin-treated groups increased with time at the rate of (2.03±1.58 mm³/day) and (2.22±0.30 mm³/day) respectively whereas it decreased in the birinapant-treated and co-therapy-treated mice at the rate of (-1.42±0.23 mm³/day) and (-1.94±0.59 mm³/day) respectively (Fig. 4F). Tumors harvested after treatment were also weighed and histologically analyzed (Fig. S4F and S4G). IHC staining for the cell proliferation biomarker Ki-67 in these tumor fragments demonstrated no significant differences in the percentage of Ki67-positive cells across the 4 treatment groups (Fig. S4H). The mean number of mitosis (per high power field) was found lower in the birinapant (0.6377) and co-therapy (0.720)-treated groups compared to the vehicle (1.533) and carboplatin (1.388) groups, although it did not reach statistical significance (Fig. S4I). No signs of drug toxicity as measured by mouse body weight were observed (Fig. S4J).

Overall, the results highlight consistency between in vitro and in vivo therapeutic responses of cancer cells to combination therapy as measured by the 3D organoid bioassay. This suggests that the 3D organoid bioassay may be used as valuable preclinical research tool in the field of cancer therapeutics to evaluate the efficacy of targeted therapies. Additionally, the results indicate the efficacy of carboplatin and birinapant combination in targeting a subset of platinum-resistant primary human ovarian cancers.