1. Introduction
Osteosarcoma, the most common primary bone cancer, exemplifies the progress that needs to be made in the approach to discovering new therapies. As the third most common cancer of childhood, osteosarcoma is disproportionately lethal, and patients with advanced or metastatic disease have limited treatment options [ 1 ]. Due to the low incidence of osteosarcoma and the extensive genetic heterogeneity [ 2 ], finding common genetic drivers and common pathways of relevance remains difficult, and the exact etiology remains unknown. Because of these features, progress in identifying new therapies has been slow, and decades of research have brought almost no improvement in patient survival rates [ 3 , 4 ]. Even for those patients who survive, both their life expectancy and quality of life are negatively impacted by the current treatment regimen [ 5 ]. For all of these reasons, there is a persistently unmet need to develop new therapies for this deadly disease.
The story of osteosarcoma is the story of many cancers, especially uncommon ones: Alternative therapeutic approaches are urgently needed, but the path forward is not clear. How can we identify, design, and test new molecular therapies in a disease that is both rare and genetically diverse? Mouse models are a critical tool, but additional translational steps are needed. To complete these additional steps, we are able to look to pet dogs with spontaneous osteosarcoma. While there are approximately 800 cases of human osteosarcoma diagnosed each year, there are at least 30,000 cases of canine osteosarcoma diagnosed each year [ 6 , 7 ]. Treatments in canine and human osteosarcoma patients are identical, and studies have revealed remarkable genomic conservation between canine and human osteosarcoma, with shared molecular alterations in known cancer pathways and shared amplifications of known oncogenes [ 8 , 9 , 10 , 11 ]. Canine patients with spontaneous disease–in contrast to genetically-engineered mouse models–offer a high incidence of tumors that are comparable to humans biologically and genetically [ 7 , 12 ], share environmental factors with humans, have an intact immune system, and possess similar clinical presentation including progression, resistance, recurrence, and metastasis. Most importantly, canine osteosarcoma patients have a shorter course of disease than human osteosarcoma patients, which means therapeutic discoveries could be made more quickly with a platform that integrates canine osteosarcoma into our current disease models (reviewed in [ 13 ]). Naturally-occurring osteosarcoma in dogs offers an unparalleled opportunity to understand the genomics of the disease, to learn about disease progression, and to trial new investigational drugs that would otherwise take too long to accrue in human studies.
While these common features of canine and human osteosarcoma are attractive, it is also important to clearly understand the differences in the biology of canine and human cancers. For example, while osteosarcomas in canines and humans are quite similar in terms of their clinical presentation, progression, host factors (for instance, individuals larger in size are at a higher risk), and molecular profiles (reviewed in [ 13 , 14 ]), there are also some distinct differences, such as the propensity for dogs to be spayed or neutered, which can confound comparisons to human disease (reviewed in [ 14 ]). In addition, while both canines and humans have bimodal age distributions in osteosarcoma incidence, human osteosarcoma is more likely to occur during childhood, while most cases of canine osteosarcoma occur in older dogs (reviewed in [ 14 ]). Indeed, human osteosarcoma is a disease of childhood and adolescence while canine osteosarcoma is a disease of middle to advanced age. Yet despite these differences, studying naturally-occurring osteosarcoma in dogs offers several advantages to studying osteosarcoma in humans: a significantly larger number of patients, treatment features which are extremely similar, and a shorter course of disease. This provides an unparalleled opportunity to understand the genomics of the disease, to learn about disease progression, and to trial new investigational drugs that would otherwise take too long to accrue in human studies.
Patient-derived models of cancer, including low-passage cell lines [ 15 ], patient-derived organoids [ 16 , 17 ], and patient-derived xenografts (PDXs) [ 18 ], are increasingly being used as “standard” preclinical models to identify sensitivities to new candidate therapeutics across cancers. Patient-derived xenografts are also being used to predict drug response [ 19 ] and identify novel drug combinations [ 20 , 21 , 22 , 23 ]. Organoid models are also now being developed to test response to immunotherapies, as the organoids for several cancer types have been shown to contain infiltrating lymphocytes [ 24 , 25 ]. Combinations of these patient-derived models are currently being explored to develop precision medicine strategies for cancer care [ 26 ]. However, translating new discoveries in real time remains a challenge in human patients, in whom disease progression can be slow and whose overall picture can be complicated by a variety of treatments, both for the cancer and for comorbid conditions.
Here we combine the advantages of a comparative oncology approach (e.g., larger numbers of patients, fewer confounding treatment variables, more rapid disease progression) with patient-derived models to develop and refine a cross-species drug discovery pipeline. The pipeline uses patient samples from either pet dogs or humans to generate patient-matched, low-passage cell lines, and PDXs. Cell lines are used to perform high-throughput chemical screens, and top hits are validated in vitro using matched PDX models ( Figure 1 A). Using this approach, we identified the proteasome and CRM1 nuclear export pathways as therapeutic vulnerabilities for osteosarcoma. Using in vitro and in vivo validations, we show that inhibition of both the proteasome and CRM1 pathways acts synergistically to inhibit osteosarcoma growth. Together, these results demonstrate the utility of our cross-species drug discovery pipeline to identify new targets and strategies to treat osteosarcoma and other rare cancers.
2. Results
2.1. Development of a Cross-Species Drug Discovery Pipeline
Osteosarcomas and other rare cancers suffer from a lack of access to patient-derived models for study. We reasoned that increased access to a larger patient population with nearly-identical biology could have significant benefits in identifying actionable pathways in osteosarcoma. To this end, we developed a cross-species pipeline that leverages the increased canine patient population and the extensive biological similarities between humans and pet dogs with naturally-occurring sarcomas. The pipeline uses patient tumor tissue from dogs or humans to create patient-derived xenografts (PDXs) that are grown and passaged in immunocompromised mice ( Figure 1 A). These PDXs are used to create matched, low-passage, patient-derived cell lines. The cell lines are applied to high-throughput screens to identify candidate therapies, and top candidates are validated in vivo using patient-matched PDXs ( Figure 1 A) to identify therapeutic vulnerabilities that are shared across species. To date, we have created a total of 9 human ( Figure 1 B) and 20 canine sarcoma PDXs ( Figure 1 C).
2.2. The Cross-Species Pipeline Identified Proteasome and CRM1 Inhibition as Novel Treatments for Osteosarcoma
We applied our pipeline to reveal novel therapeutic vulnerabilities at both the individual patient level and, more importantly, shared across osteosarcomas. To do this, we created matched cell lines from both dog (D418) and human (17-3X) PDXs. Both PDXs were confirmed as osteosarcoma by histopathology ( Figure 2 A,B). Using the PDX tissue, we created clonally-purified cell lines. The 17-3X human osteosarcoma cell line exhibits a more spindle-shaped morphology while the D418 dog osteosarcoma cell line displays a rounded, cobblestone-like morphology ( Figure 2 A,B, right panels). Clonally-derived cell lines were confirmed to be free of detectable mouse fibroblasts using species-specific PCR assays designed to detect human, dog, and mouse DNA ( Figure 2 C,D). The 17-3X cell line has an estimated doubling time of approximately 40 h ( Figure 2 E) while the D418 line grows more rapidly, with an estimated doubling time of approximately 21 h ( Figure 2 F). The 17-3X and D418 low-passage cell lines were combined with a panel of seven additional established osteosarcoma cell lines from both humans and dogs to perform high-throughput screens using 119 FDA-approved small molecule oncology drugs. These screens revealed several trends: (1) cell line-specific variation in responses were observed across the panel, with some cell lines more broadly resistant to drugs in the screen, such as U2OS and D17, and other lines more sensitive, such as 17-3X and Abrams ( Figure 3 A); (2) hierarchical clustering of the nine cell lines using the drug responses for each cell line revealed species-specific clustering of the cell lines, with distinct human and dog clades formed based on response of each cell line to the entire panel of drugs ( Figure 3 A); (3) while cell lines clustered by species, there was overall consistency in the average percent killing across all cell lines, particularly among the top hits ( Figure 3 B). Together, these results suggest that although both individual and species-specific responses exist across osteosarcomas, there are consistent responses to the most effective agents. Importantly, among the top hits were standard-of-care agents such as anthracyclines (e.g., doxorubicin, daunorubicin, idarubicin, epirubicin) and methotrexate, which consistently killed all cell lines, while others showed wider variation (e.g., etoposide) and limited efficacy (e.g., platinum-based chemotherapy) ( Figure 3 C). These analyses indicate that this screening approach is capable of identifying relevant therapies for osteosarcoma.
Unlike standard-of-care therapies, analysis of small molecule sensitivities at the individual cell line level revealed heterogeneous responses across the panel of nine cell lines ( Figure 3 D). Most drugs, such as belinostat (an HDAC inhibitor) or ponatinib (a multi-tyrosine kinase inhibitor), were efficacious across more than one of the nine cell lines ( Figure 3 D). These analyses reveal that substantial heterogeneity in drug response exists across osteosarcomas.
Given the extensive heterogeneity of response, we sought to identify novel compounds with efficacy across the entire panel of cell lines. Interestingly, the two most efficacious inhibitors, with an average of >95% killing for both inhibitors across all nine lines, were compounds that target the proteasome pathway ( Figure 3 E). Together, our results indicate that both human and canine osteosarcomas display heterogeneity in drug response across cell lines, with convergence on the proteasome pathway as a novel target to treat osteosarcoma.
To better understand the landscape of therapeutic vulnerabilities in osteosarcoma, we next performed high-throughput chemical screens in D418 and 17-3X patient-derived lines using 2100 bioactive compounds. This compound library is annotated by both target and pathway, enabling both protein- and pathway-level interrogation of chemical sensitivities. D418 and 17-3X cells displayed similar sensitivity profiles, with just 11.9% and 8.7% of compounds inducing ≥50% killing in D418 and 17-3X, respectively ( Figure 4 A,B; Table S1 ). Consistent with the results from the 119 compound screens, responses of D418 and 17-3X to all 2100 compounds were correlated (R 2 = 0.54; p < 0.0001) ( Figure 4 C). Also consistent with the previous screens, each cell line displayed sensitivity to a subset of agents ( Figure 4 D). For example, D418 displayed sensitivity to multiple MEK and FAK inhibitors while 17-3X was uniquely sensitive to Chk inhibitors ( Figure 4 D). In addition to the unique sensitivities, both cell lines showed common sensitivity to standard-of-care anthracyclines and a number of novel agents ( Figure 4 E). These agents included the zinc pyrithione, the active ingredient in dandruff shampoo, the pan-selective Jumonji histone demethylase inhibitor, JIB-04, an NF-kB inhibitor (WS3), and two CDK inhibitors (alvocidib and SNS-032) ( Figure 4 E). Given that almost all small molecule inhibitors have multiple targets, we focused on targets for which at least three drugs showed >50% killing. We reasoned that filtering by drug targets with multiple hits in the screen would identify the most high-confidence drug targets for downstream validation. From these analyses, we identified the CRM1 nuclear export and proteasome pathways as the top candidate targets ( Figure 4 F). A total of 3 of 4 CRM1 inhibitors and 9 of 11 proteasome inhibitors showed >50% killing in both D418 and 17-3X cell lines ( Figure 4 G,H). Consistent with these analyses, both the CRM1 inhibitor, verdinexor, and the proteasome inhibitor, bortezomib, showed dose-dependent inhibition of 143B and 17-3X human osteosarcomas ( Figure 4 I,J) and D418 and D17 canine osteosarcomas ( Figure 4 K,L), pinpointing the CRM1 and proteasome pathways as lead candidates for in vivo validation.
2.3. In Vivo Validation of Proteasome and CRM1 Pathway Inhibitors to Treat Osteosarcoma
Our in vitro small molecule screens pinpointed the proteasome and CRM1 nuclear export pathways as two promising therapeutic vulnerabilities for osteosarcoma. Consistent with this, CRM1 protein is highly expressed ( Figure S1A ) and localized within the nucleus of osteosarcoma cells ( Figure S1B ). Moreover, elevated CRM1 expression is prognostic for poorer metastasis-free and overall survival in human osteosarcoma ( Figure S1C,D ). We validated the therapeutic efficacy of proteasome and CRM1 inhibition in D418 PDXs and showed that both CRM1 inhibition and proteasome inhibition significantly reduced tumor growth ( Figure 5 A). CRM1 inhibition also significantly reduced 17-3X PDX tumor growth, while the proteasome inhibitor, bortezomib, had no effect as a single agent ( Figure 5 B). Mouse weights remained unchanged during the course of the treatment ( Figure S1E,F ). Based on the positive results for CRM1 inhibitors in two PDXs, we further verified the efficacy of CRM1 inhibition in two additional PDXs, D071 and D075 ( Figure 5 C,D).
2.4. Combined Proteasome and CRM1 Pathway Inhibition Act Synergistically to Prevent Osteosarcoma Growth
Our cross-species pipeline pinpointed both proteasome and CRM1 pathways as promising single-agent therapies to treat osteosarcoma. However, although CRM1 inhibition was effective across all four of the PDXs tested, the proteasome inhibitor, bortezomib, was only capable of inhibiting growth in one of the two treated PDXs when used as a single agent. Interestingly, these pathways are being targeted with combination therapy for synergistic benefit in several cancer types, including multiple myeloma [ 27 ], colorectal cancer [ 28 ], and fibrosarcoma [ 29 ]. Based on these studies and our results, we hypothesized that combined proteasome and CRM1 blockade would have synergistic benefit for osteosarcoma. Consistent with this hypothesis, the combination of proteasome and CRM1 inhibition induced synergistic cell death for both D418 and 17-3X cells ( Figure 5 E,F), suggesting that this combination may represent a rational strategy to treat osteosarcoma.
3. Discussion
In the present study, we sought to combine the utility of comparative oncology with patient-derived models of cancer and high-throughput small molecule screens to create a cross-species, personalized medicine pipeline. We applied this pipeline to osteosarcoma, a painful bone cancer that occurs predominantly in adolescents and for which almost no treatment progress has been made in nearly four decades. Using our pipeline, we identified therapeutic vulnerabilities for osteosarcoma that are patient-specific, common across patients, and common across species. Among the common therapeutic sensitivities identified were the proteasome and CRM1 nuclear export pathways. Both of these pathways have been explored in depth as potential cancer monotherapies for a range of cancer types [ 30 , 31 ], and both of these therapies are currently approved by the U.S. Food and Drug Administration for the treatment of multiple myeloma [ 32 , 33 ].
The proteasome is a multi-subunit complex that degrades misfolded, damaged, or unused proteins. The proteasome regulates the turnover of thousands of proteins in the cell [ 34 ], and proteasome inhibition creates an imbalance in the levels of misfolded proteins, leading to induction of the unfolded protein response, cellular stress, and apoptosis [ 35 ]. The CRM1 nuclear export pathway transports proteins through the nuclear envelope to the cytoplasm [ 36 ]. Like the proteasome, the nuclear export pathway is responsible for regulating thousands of proteins in the cell [ 37 ]. Despite the fact that CRM1 is a critical mediator of protein localization for normal cells, there is support for targeting CRM1 in osteosarcoma. Our data show that CRM1 is upregulated in osteosarcoma cells as compared to normal osteoblasts ( Figure S1A ). In addition, in vivo treatment of mice bearing D418 and 17-3X PDXs with the CRM1 inhibitor, verdinexor, had no effect on mouse weight ( Figure S1E,F ). Third, CRM1 inhibitors are currently being used to treat relapsed, refractory multiple myeloma [ 38 ]. Together, our results and current, ongoing clinical trials suggest a therapeutic window exists to rationally utilize CRM1 inhibitors to treat osteosarcoma.
Interestingly, the proteasome and CRM1 export pathways are functionally linked. Proteasome inhibition in colorectal cancer cells induces CRM1-dependent nuclear export of ubiquitinated proteins, and inhibition of CRM1 prevents this export, leading to cell cycle arrest and apoptosis [ 28 ]. In addition, CRM1 inhibition re-sensitizes chemo-resistant myeloma cells to proteasome inhibition [ 39 ]. Inhibiting these inter-dependent pathways synergistically inhibits multiple cancers [ 27 , 28 , 29 ]. Consistent with observations in other cancers, our data support the investigation of proteasome and CRM1 pathway inhibitors for osteosarcoma. It is worth noting we observed more consistent evidence for the use of CRM1 inhibitors, as bortezomib was not effective in one of the two matched cell line-PDX models studied. The reasons for the discrepancy between our in vitro and in vivo data on bortezomib are likely due to numerous factors, such as cross-talk with tumor-associated stromal cells [ 40 ]; however, despite this discrepancy we observed that combined proteasome and CRM1 pathway inhibition led to synergistic cytotoxicity in vitro, providing further evidence for the efficacy of dual proteasome and CRM1 pathway inhibition in treating osteosarcoma. Future canine clinical trials will provide important insights into the tolerability and efficacy of this combination therapy.
Both of these targets/pathways are currently being explored as treatments for pediatric cancers, with promising tolerability and efficacy. In a phase I tolerability study, the CRM1 inhibitor selinexor was tolerable at doses as high as 55 mg/m 2 in combination with fludarabine and cytarabine in pediatric patients with relapsed or refractory leukemia [ 41 ]. Likewise, the addition of bortezomib in combination with reinduction therapy led to remissions in 8/10 patients, with grade 3 infections in 4/10 patients [ 42 ]. Unlike selinexor and bortezomib, which appear tolerable when combined with other chemotherapeutic agents, the standard-of-care treatment regimen for osteosarcoma is a combination of cytotoxic methotrexate, adriamycin (doxorubicin), and cisplatin, which is associated with substantial treatment-derived morbidity [ 43 , 44 , 45 ]. Side effects from these agents can delay treatment regimens in pediatric and adolescent patients, and more severe complications, such as anthracycline-induced cardiomyopathy, greatly increase subsequent morbidity [ 46 , 47 ]. Indeed, even osteosarcoma survivors do not have a normal life expectancy due to the morbidity associated with their therapy during childhood.
The establishment of a cross-species drug discovery pipeline provides a robust platform to identify and validate potential new therapies and offers several advantages. First, it capitalizes on the expanded canine patient population with spontaneous disease, thereby substantially improving access to patient samples for the development of patient-derived models of cancer. This is particularly useful for studying rare cancers. Indeed, we established more than double the number of PDXs from canines than from humans. Second, combining low passage and established cell lines with high-throughput chemical screens enables interrogation of hundreds to thousands of compounds simultaneously, pinpointing both patient-specific and population-level therapeutic vulnerabilities. Third, the use of chemical screens with complete target and pathway annotation allows for identification of single agent- and target/pathways-level drug sensitivities. Fourth, the use of matched cell lines and PDXs from the same patient provides a robust system to validate top candidates. While the in vivo studies are a critical validation step in the pipeline, these studies are costly and time consuming. Future iterations of the pipeline that exploit continued improvement in patient-derived models, such as patient-derived organoids [ 17 , 48 , 49 ], are sure to improve the speed and cost-effectiveness of the pipeline and will further enable rapid translation of lead candidates into clinical practice [ 21 ]. Fifth, and perhaps most critically, the ability to test these candidates in veterinary clinical trials “closes the loop” of drug discovery ( Figure 1 A) and enables testing of novel therapeutic strategies at a fraction of the cost and time necessary for human trials (reviewed in [ 13 ]). Using this comparative oncology pipeline we identified two promising inhibitor classes, targeting the proteasome and CRM1 nuclear export pathways, with in vitro efficacy as single agents and in vitro synergy in combination. Together, our data suggest that future studies should be focused on testing the in vivo efficacy of combined targeting of the proteasome and CRM1 pathways in osteosarcoma. Canine clinical trials may pave the way for subsequent human clinical trials aimed at prolonging the lives of osteosarcoma patients.
4. Materials and Methods
4.1. Generation of the Patient-Derived Xenograft Models
Canine sarcoma tumor tissue samples were collected from University of Illinois at Urbana Collage of Veterinary Medicine (Urbana, IL, USA) and Triangle Veterinary Referral Hospital (Durham, NC, USA) under Institutional Animal Care and Use committee (IACUC)-approved protocols. Human sarcoma tumor tissue samples were collected under a Duke IRB approved protocol (Pro00002435). All patients provided written informed consent to participate in the study. PDX models of the human and canine sarcoma were generated as described previously [ 21 ] and all mouse experiments were performed in accordance with the animal guidelines and with Duke University IACUC approval (A063-18-03). To develop PDXs, human and canine tumor tissue samples were washed in phosphate buffered saline (PBS), dissected into small pieces (<2 mm), and injected into the flanks of 8–10-week-old JAX NOD.CB17-PrkdcSCID-J mice obtained from the Duke University Rodent Genetic and Breeding Core.
4.2. Patient-Derived Cell Line Generation, Authentication and Detection of Mouse Cell Contamination
Cell lines were generated from human and canine tumor samples. After washing in phosphate buffered saline (PBS), small pieces (<2 mm) of tumor tissue were mechanically homogenized. The homogenized tissue was then suspended in cell growth media and cultured in 12 well plates with DMEM + 10% FBS + 1% Penicillin/Streptomycin. To isolate tumor cells, growing colonies of cells were isolated by trypsinization using O rings and cultured in new 12 well plates. This process was repeated until a colony of cells that resembled pure tumor cells in morphology was established.
For the cell lines generated from PDX, mouse cell contamination of the PDX cell lines were detected by PCR using human, canine, and mouse specific primers. Because mouse primers easily cross-react with canine and human gDNA, two different mouse primer sets were used. For human cell lines, human reverse (5′ TCC AGG TTT ATG GAG GGT TC), human forward (5′ TAG ACA TCG TAC TAC ACG ACA CG), mouse reverse (5′ CCC AAA GAA TCA GAA CAG ATG C) and mouse forward (5′ ATT ACA GCC GTA CTG CTC CTA T); for canine cell lines, canine reverse (5′ GTA AAG GCT GCC TGA GGA TAA G), canine forward (5′GGT CCA GGG AAG ATC AGA AAT G), mouse reverse (AGG TGT CAC CAG GAC AAA TG) and mouse forward (CTG CTT CGA GCC ATA GAA CTA A) primer sets were used.
All PDX and patient-derived cell lines were authenticated using the Duke University DNA Analysis Facility cell line authentication service by analyzing DNA samples from each individual cell line for polymorphic short tandem repeat markers using the GenePrint 10 kit from Promega (Madison, WI, USA).
4.3. In Vitro Studies and High-Throughput Drug Screen
All human (143B, MG63, SAOS, U2OS, 17-3X) and canine (Abrams, Moresco, D17, D418) osteosarcoma cell lines were cultured in DMEM + 10% FBS + 1% Penicillin/Streptomycin. The NIH Approved Oncology set (119 compounds) and Selleck Bioactives collection (2100 compounds) were screened in the Duke Functional Genomics Shared Resource as described previously [ 21 , 50 ]. Briefly, compounds were first stamped in triplicate into 384 well plates at a final concentration of 1 μM using an Echo Acoustic Dispenser (Labcyte, Indianapolis, IN, USA). Cells and media were then dispensed into plates using a WellMate (Thermo Fisher, Waltham, MA, USA) at a density of 2000 cells/well for each cell line. CellTiter-Glo (Promega, Madison, WI, USA) viability assays were performed after incubation of cells with compounds for 72 h and luminescence was read using a Clariostar plate reader (BMG, Berlin, Germany). Top drug targets as identified by the high-throughput drug screens, Bortezomib (PS-341), Verdinexor (KPT-335) and 17-DMAG (Alvespimycin) HCl were purchased from Selleck Chemicals (Houston, TX, USA) and were solubilized in DMSO at 10 mM concentration to use for in vitro IC50 studies. Dose-response curves were generated with four-fold serial dilutions from a starting concentration of 20 μM.
4.4. In Vivo Drug Sensitivity Validation
To validate in vitro drug screen results in vivo, 150 μL homogenized PDX tissue-PBS suspensions at 150 mg/mL concentration were injected subcutaneously into the right flanks of the 8 weeks old JAX NOD.CB17-PrkdcSCID-J mice. When the tumor volumes reached 100 mm 3 , mice were randomized ( n = 5 mice for each treatment group and n = 5 for the control group) and 1 mg/kg bortezomib, 25 mg/kg alvespimycin and 5 mg/kg verdinexor intraperitoneal injections were initiated twice a week. Tumor volumes were measured three times a week using calipers, and ( L e n g t h × W i d t h 2 ) 2 was used to calculate the tumor size. Mice were sacrificed on day 18 or if the tumor volume reached 1500 mm 3 .
4.5. Data Analysis and Statistics
JMP from SAS software (Cary, NC, USA) was used for the high-throughput drug screen data analysis. Analysis of Means was used to identify the top drug candidates from the 119-compound drug screen and the 2100-compound screen. Tumor volumes were recorded in GraphPad Prism 6 software (La Jolla, CA, USA). Two-way ANOVA analysis was used to compare the tumor volumes among the control and treatment groups. Kaplan Meier curves for CRM1 expression were generated using the R2 genomics software ( https://hgserver1.amc.nl/cgi-bin/r2/main.cgi ) with the “Kaplan Meier by gene expression” tool and the cutoff/modus set to “scan”.
5. Conclusions
We demonstrate here the utility of combining patient-derived models of cancer with comparative oncology approaches to discover potential new therapies for osteosarcoma. The increased access to patient samples with biologically-similar disease not only means we can make discoveries in the lab faster, but we can also validate lead candidates in veterinary clinical trials at a fraction of the time and cost. Increasing collaboration between veterinarians, cancer researchers, and human physicians is needed to apply this pipeline and identify new therapies to treat osteosarcoma and other cancers across species.
Acknowledgments
The authors acknowledge and thank the BioRepository & Precision Pathology Center (BRPC), a shared resource of the Duke University School of Medicine and Duke Cancer Institute, for providing access to human biospecimens used under institutional review board oversight in this work. The BRPC receives support from the P30 Cancer Center Support Education Foundation Resident Research Grant (P30 CA014236ALL). The authors acknowledge the Duke Functional Genomics Shared Resource for assistance with compound screens.
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Materials
The following are available online at https://www.mdpi.com/2072-6694/12/11/3335/s1 , Figure S1: CRM1 upregulation is prognostic for poorer clinical outcomes in osteosarcomas, Table S1: Drug screen data for 17-3X and D418.
Click here for additional data file.
Author Contributions
J.A.S., W.C.E., and D.S.H. conceived of the project and planned experiments. G.R., E.A., E.M.F., S.R., D.S., A.L.L., S.M.H., M.U.S., S.C., S.Y.K., K.E.W., A.A., and M.M.C. carried out the experiments, collected, and analyzed the data. L.E.S., C.E., J.I.E., and S.J.M. provided samples and helped interpret results. J.A.S. and G.R. wrote the first draft of the manuscript, and all authors provided critical feedback and helped shape the research, analysis and manuscript. All authors have read and agreed to the published version of the manuscript.
Funding
This work was supported by the Triangle Center for Evolutionary Medicine (J.A.S.), Hyundai Hope on Wheels (W.C.E.), the Ryan Foundation (W.C.E.), the Consortium for Canine Comparative Oncology (W.C.E./J.A.S.), and an Orthopedic Research and Education Foundation Resident Research Grant (A.L.L.).
Conflicts of Interest
The authors declare no conflict of interest.
References
1. Mirabello L. Troisi R.J. Savage S.A. International osteosarcoma incidence patterns in children and adolescents, middle ages and elderly persons Int. J. Cancer 2009 125 229 234 10.1002/ijc.24320 19330840 PMC3048853
2. Egas-Bejar D. Anderson P.M. Agarwal R. Corrales-Medina F. Devarajan E. Huh W.W. Brown R.E. Subbiah V. Theranostic profiling for actionable aberrations in advanced high risk osteosarcoma with aggressive biology reveals high molecular diversity: The human fingerprint hypothesis Oncoscience 2014 1 167 179 10.18632/oncoscience.21 25126591 PMC4128257
3. Friebele J.C. Peck J. Pan X. Abdel-Rasoul M. Mayerson J.L. Osteosarcoma: A Meta-Analysis and Review of the Literature Am. J. Orthop. 2015 44 547 553 26665241
4. He H. Ni J. Huang J. Molecular mechanisms of chemoresistance in osteosarcoma (Review) Oncol. Lett. 2014 7 1352 1362 10.3892/ol.2014.1935 24765137 PMC3997672
5. Armenian S. Bhatia S. Predicting and Preventing Anthracycline-Related Cardiotoxicity Am. Soc. Clin. Oncol. Educ. Book 2018 38 3 12 10.1200/EDBK_100015 30231396
6. Rodriguez C.O. Jr. Using Canine Osteosarcoma as a Model to Assess Efficacy of Novel Therapies: Can Old Dogs Teach Us New Tricks? Current Advances in Osteosarcoma Advances in Experimental Medicine and Biology Springer Cham, Switzerland 2014 Volume 804 237 256 10.1007/978-3-319-04843-7_13 24924178
7. Schiffman J.D. Breen M. Comparative oncology: What dogs and other species can teach us about humans with cancer Philos. Trans. R. Soc. B Biol. Sci. 2015 370 20140231 10.1098/rstb.2014.0231 PMC4581033 26056372
8. Paoloni M. Davis S. Lana S. Withrow S. Sangiorgi L. Picci P. Hewitt S.M. Triche T. Meltzer P. Khanna C. Canine tumor cross-species genomics uncovers targets linked to osteosarcoma progression BMC Genomics 2009 10 625 10.1186/1471-2164-10-625 20028558 PMC2803201
9. Karlsson E.K. Sigurdsson S. Ivansson E. Thomas R. Elvers I. Wright J. Howald C. Tonomura N. Perloski M. Swofford R. Genome-wide analyses implicate 33 loci in heritable dog osteosarcoma, including regulatory variants near CDKN2A/B Genome Biol. 2013 14 R132 10.1186/gb-2013-14-12-r132 24330828 PMC4053774
10. Angstadt A.Y. Motsinger-Reif A. Thomas R. Kisseberth W.C. Couto C.G. Duval D.L. Nielsen D.M. Modiano J.F. Breen M. Characterization of canine osteosarcoma by array comparative genomic hybridization and RT-qPCR: Signatures of genomic imbalance in canine osteosarcoma parallel the human counterpart Genes Chromosomes Cancer 2011 50 859 874 10.1002/gcc.20908 21837709
11. Angstadt A.Y. Thayanithy V. Subramanian S. Modiano J.F. Breen M. A genome-wide approach to comparative oncology: High-resolution oligonucleotide aCGH of canine and human osteosarcoma pinpoints shared microaberrations Cancer Genet. 2012 205 572 587 10.1016/j.cancergen.2012.09.005 23137772
12. Lindblad-Toh K. Wade C.M. Mikkelsen T.S. Karlsson E.K. Jaffe D.B. Kamal M. Clamp M. Chang J.L. Kulbokas E.J. 3rd Zody M.C. Genome sequence, comparative analysis and haplotype structure of the domestic dog Nature 2005 438 803 819 10.1038/nature04338 16341006
13. Somarelli J.A. Boddy A.M. Gardner H.L. DeWitt S.B. Tuohy J. Megquier K. Sheth M.U. Hsu S.D. Thorne J.L. London C.A. Improving Cancer Drug Discovery by Studying Cancer across the Tree of Life Mol. Biol. Evol. 2020 37 11 17 10.1093/molbev/msz254 31688937 PMC8204703
14. Makielski K.M. Mills L.J. Sarver A.L. Henson M.S. Spector L.G. Naik S. Modiano J.F. Risk Factors for Development of Canine and Human Osteosarcoma: A Comparative Review Vet. Sci. 2019 6 48 10.3390/vetsci6020048 31130627 PMC6631450
15. Van De Wetering M. Francies H.E. Francis J.M. Bounova G. Iorio F. Pronk A. Van Houdt W. Van Gorp J. Taylor-Weiner A. Kester L. Prospective derivation of a living organoid biobank of colorectal cancer patients Cell 2015 161 933 945 10.1016/j.cell.2015.03.053 25957691 PMC6428276
16. Barretina J. Caponigro G. Stransky N. Venkatesan K. Margolin A.A. Kim S. Wilson C.J. Lehar J. Kryukov G.V. Sonkin D. The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity Nature 2012 483 603 607 10.1038/nature11003 22460905 PMC3320027
17. Vlachogiannis G. Hedayat S. Vatsiou A. Jamin Y. Fernández-Mateos J. Khan K. Lampis A. Eason K. Huntingford I. Burke R. Patient-derived organoids model treatment response of metastatic gastrointestinal cancers Science 2018 359 920 926 10.1126/science.aao2774 29472484 PMC6112415
18. Koga Y. Ochiai A. Systematic Review of Patient-Derived Xenograft Models for Preclinical Studies of Anti-Cancer Drugs in Solid Tumors Cells 2019 8 418 10.3390/cells8050418 PMC6562882 31064068
19. Gao H. Korn J.M. Ferretti S. Monahan J.E. Wang Y. Singh M. Zhang C. Schnell C. Yang G. Zhang Y. High-throughput screening using patient-derived tumor xenografts to predict clinical trial drug response Nat. Med. 2015 21 1318 1325 10.1038/nm.3954 26479923
20. Lu M. Zessin A.S. Glover W. Hsu D.S. Activation of the mTOR Pathway by Oxaliplatin in the Treatment of Colorectal Cancer Liver Metastasis PLoS ONE 2017 12 e0169439 10.1371/journal.pone.0169439 28060954 PMC5218497
21. Rao S.R. Somarelli J.A. Altunel E. Selmic L.E. Byrum M. Sheth M.U. Cheng S. Ware K.E. Kim S.Y. Prinz J.A. From the Clinic to the Bench and Back Again in One Dog Year: How a Cross-Species Pipeline to Identify New Treatments for Sarcoma Illuminates the Path Forward in Precision Medicine Front. Oncol. 2020 10 117 10.3389/fonc.2020.00117 32117764 PMC7026496
22. Somarelli J.A. Roghani R.S. Moghaddam A.S. Thomas B.C. Rupprecht G. Ware K.E. Altunel E. Mantyh J.B. Kim S.Y. McCall S.J. A Precision Medicine Drug Discovery Pipeline Identifies Combined CDK2 and 9 Inhibition as a Novel Therapeutic Strategy in Colorectal Cancer Mol. Cancer Ther. 2020 10.1158/1535-7163.MCT-20-0454 33158998 PMC7718319
23. Altunel E. Roghani R.S. Chen K.Y. Kim S.Y. McCall S. Ware K.E. Shen X. Somarelli J.A. Hsu D.S. Development of a precision medicine pipeline to identify personalized treatments for colorectal cancer BMC Cancer 2020 20 592 10.1186/s12885-020-07090-y 32580713 PMC7313200
24. Scognamiglio G. De Chiara A. Parafioriti A. Armiraglio E. Fazioli F. Gallo M. Aversa L. Camerlingo R. Cacciatore F. Colella G. Patient-derived organoids as a potential model to predict response to PD-1/PD-L1 checkpoint inhibitors Br. J. Cancer 2019 121 979 982 10.1038/s41416-019-0616-1 31666667 PMC6889147
25. Neal J.T. Li X. Zhu J. Giangarra V. Grzeskowiak C.L. Ju J. Liu I.H. Chiou S.-H. Salahudeen A.A. Smith A.R. Organoid Modeling of the Tumor Immune Microenvironment Cell 2018 175 1972 1988.e16 10.1016/j.cell.2018.11.021 30550791 PMC6656687
26. Pauli C. Hopkins B.D. Prandi D. Shaw R. Fedrizzi T. Sboner A. Sailer V. Augello M. Puca L. Rosati R. Personalized In Vitro and In Vivo Cancer Models to Guide Precision Medicine Cancer Discov. 2017 7 462 477 10.1158/2159-8290.CD-16-1154 28331002 PMC5413423
27. Turner J.G. Kashyap T. Dawson J.L. Gomez J. Bauer A.A. Grant S. Dai Y. Shain K.H. Meads M. Landesman Y. XPO1 inhibitor combination therapy with bortezomib or carfilzomib induces nuclear localization of IkappaBalpha and overcomes acquired proteasome inhibitor resistance in human multiple myeloma Oncotarget 2016 7 78896 78909 10.18632/oncotarget.12969 27806331 PMC5340237
28. Wu T. Chen W. Zhong Y. Hou X. Fang S. Liu C.-Y. Wang G. Yu T. Huang Y.-Y. Ouyang X. Nuclear Export of Ubiquitinated Proteins Determines the Sensitivity of Colorectal Cancer to Proteasome Inhibitor Mol. Cancer Ther. 2017 16 717 728 10.1158/1535-7163.MCT-16-0553 27903750
29. Kashyap T. Argueta C. Aboukameel A. Unger T.J. Klebanov B. Mohammad R.M. Muqbil I. Azmi A.S. Drolen C. Senapedis W. Selinexor, a Selective Inhibitor of Nuclear Export (SINE) compound, acts through NF-kappaB deactivation and combines with proteasome inhibitors to synergistically induce tumor cell death Oncotarget 2016 7 78883 78895 10.18632/oncotarget.12428 27713151 PMC5346685
30. Mofers A. Pellegrini P. Linder S. D’Arcy P. Proteasome-associated deubiquitinases and cancer Cancer Metastasis Rev. 2017 36 635 653 10.1007/s10555-017-9697-6 29134486 PMC5721125
31. Sun Q. Chen X. Zhou Q. Burstein E. Yang S. Jia D. Inhibiting cancer cell hallmark features through nuclear export inhibition Signal Transduct. Target. Ther. 2016 1 16010 10.1038/sigtrans.2016.10 29263896 PMC5661660
32. XPO1 Inhibitor Approved for Multiple Myeloma Cancer Discov. 2019 9 1150 1151 10.1158/2159-8290.CD-NB2019-085 31332020
33. Kane R.C. Farrell A.T. Sridhara R. Pazdur R. United States Food and Drug Administration Approval Summary: Bortezomib for the Treatment of Progressive Multiple Myeloma after One Prior Therapy Clin. Cancer Res. 2006 12 2955 2960 10.1158/1078-0432.CCR-06-0170 16707588
34. Kim W. Bennett E.J. Huttlin E.L. Guo A. Li J. Possemato A. Sowa M.E. Rad R. Rush J. Comb M.J. Systematic and Quantitative Assessment of the Ubiquitin-Modified Proteome Mol. Cell 2011 44 325 340 10.1016/j.molcel.2011.08.025 21906983 PMC3200427
35. Obeng E.A. Carlson L.M. Gutman D.M. Harrington W.J. Jr. Lee K.P. Boise L.H. Proteasome inhibitors induce a terminal unfolded protein response in multiple myeloma cells Blood 2006 107 4907 4916 10.1182/blood-2005-08-3531 16507771 PMC1895817
36. Stade K. Ford C.S. Guthrie C. Weis K. Exportin 1 (Crm1p) Is an Essential Nuclear Export Factor Cell 1997 90 1041 1050 10.1016/S0092-8674(00)80370-0 9323132
37. Taylor J. Sendino M. Gorelick A.N. Pastore A. Chang M.T. Penson A.V. Gavrila E.I. Stewart C. Melnik E.M. Chavez F.H. Altered Nuclear Export Signal Recognition as a Driver of Oncogenesis Cancer Discov. 2019 9 1452 1467 10.1158/2159-8290.CD-19-0298 31285298 PMC6774834
38. Theodoropoulos N. Lancman G. Chari A. Targeting Nuclear Export Proteins in Multiple Myeloma Therapy Target. Oncol. 2020 1 12 10.1007/s11523-020-00758-2 33074469 PMC7570401
39. Chanukuppa V. Paul D. Taunk K. Chatterjee T. Sharma S. Kumar S. Santra M.K. Rapole S. XPO1 is a critical player for bortezomib resistance in multiple myeloma: A quantitative proteomic approach J. Proteomics 2019 209 103504 10.1016/j.jprot.2019.103504 31465861
40. Markovina S. Callander N.S. O’Connor S.L. Xu G. Shi Y. Leith C.P. Kim K. Trivedi P. Kim J. Hematti P. Bone marrow stromal cells from multiple myeloma patients uniquely induce bortezomib resistant NF-kappaB activity in myeloma cells Mol. Cancer 2010 9 176 10.1186/1476-4598-9-176 20604947 PMC3095250
41. Alexander T.B. Lacayo N.J. Choi J.K. Ribeiro R.C. Pui C.-H. Rubnitz J.E. Phase I Study of Selinexor, a Selective Inhibitor of Nuclear Export, in Combination With Fludarabine and Cytarabine, in Pediatric Relapsed or Refractory Acute Leukemia J. Clin. Oncol. 2016 34 4094 4101 10.1200/JCO.2016.67.5066 27507877 PMC5477824
42. August K.J. Guest E.M. Lewing K. Hays J.A. Gamis A.S. Treatment of children with relapsed and refractory acute lymphoblastic leukemia with mitoxantrone, vincristine, pegaspargase, dexamethasone, and bortezomib Pediatr. Blood Cancer 2020 67 e28062 10.1002/pbc.28062 31724803
43. Whelan J.S. Bielack S.S. Marina N. Smeland S. Jovic G. Hook J.M. Krailo M. Anninga J.K. Butterfass-Bahloul T. Böhling T. EURAMOS-1, an international randomised study for osteosarcoma: Results from pre-randomisation treatment Ann. Oncol. 2015 26 407 414 10.1093/annonc/mdu526 25421877 PMC4304379
44. Marina N.M. Smeland S. Bielack S.S. Bernstein M. Jovic G. Krailo M.D. Hook J.M. Arndt C. Berg H.V.D. Brennan B. Comparison of MAPIE versus MAP in patients with a poor response to preoperative chemotherapy for newly diagnosed high-grade osteosarcoma (EURAMOS-1): An open-label, international, randomised controlled trial Lancet Oncol. 2016 17 1396 1408 10.1016/S1470-2045(16)30214-5 27569442 PMC5052459
45. Ferrari S. Ruggieri P. Cefalo G. Tamburini A. Capanna R. Fagioli F. Comandone A. Bertulli R. Bisogno G. Palmerini E. Neoadjuvant Chemotherapy With Methotrexate, Cisplatin, and Doxorubicin With or Without Ifosfamide in Nonmetastatic Osteosarcoma of the Extremity: An Italian Sarcoma Group Trial ISG/OS-1 J. Clin. Oncol. 2012 30 2112 2118 10.1200/JCO.2011.38.4420 22564997
46. Taran S.J. Taran R. Malipatil N.B. Pediatric Osteosarcoma: An Updated Review Indian J. Med. Paediatr. Oncol. 2017 38 33 43 10.4103/0971-5851.203513 28469335 PMC5398104
47. Nagarajan R. Kamruzzaman A. Ness K.K. Marchese V.G. Sklar C.A. Mertens A.C. Yasui Y. Robison L.L. Marina N. Twenty years of follow-up of survivors of childhood osteosarcoma Cancer 2011 117 625 634 10.1002/cncr.25446 20922787 PMC3025070
48. Bruun J. Kryeziu K. Eide P.W. Moosavi S.H. Eilertsen I.A. Langerud J. Røsok B.I. Totland M.Z. Brunsell T.H. Pellinen T. Patient-Derived Organoids from Multiple Colorectal Cancer Liver Metastases Reveal Moderate Intra-patient Pharmacotranscriptomic Heterogeneity Clin. Cancer Res. 2020 26 4107 4119 10.1158/1078-0432.CCR-19-3637 32299813
49. Ooft S.N. Weeber F. Dijkstra K.K. McLean C.M. Kaing S. Van Werkhoven E. Schipper L. Hoes L. Vis D.J. Van De Haar J. Patient-derived organoids can predict response to chemotherapy in metastatic colorectal cancer patients Sci. Transl. Med. 2019 11 eaay2574 10.1126/scitranslmed.aay2574 31597751
50. Xu S. Ware K.E. Ding Y. Kim S.Y. Sheth M.U. Rao S.R. Chan W. Armstrong A.J. Eward W.C. Jolly M.K. An Integrative Systems Biology and Experimental Approach Identifies Convergence of Epithelial Plasticity, Metabolism, and Autophagy to Promote Chemoresistance J. Clin. Med. 2019 8 205 10.3390/jcm8020205 PMC6406733 30736412