Advanced Xenograft Model with Cotransplantation of Patient-Derived Organoids and Endothelial Colony-Forming Cells for Precision Medicine

Preclinical evaluation models have been developed for precision medicine, with patient-derived xenograft models (PDXs) and patient-derived organoids (PDOs) attracting increasing attention. However, each of these models has application limitations. In this study, an advanced xenograft model was established and used for drug screening. PDO and endothelial colony-forming cells (ECFCs) were cotransplanted in NRGA mice (PDOXwE) to prepare the model, which could also be subcultured in Balb/c nude mice. Our DNA sequencing analysis and immunohistochemistry results indicated that PDOXwE maintained patient genetic information and tumor heterogeneity. Moreover, the model enhanced tumor growth more than the PDO-bearing xenograft model (PDOX). The PDO, PDOXwE, and clinical data were also compared in the liver metastasis of a colorectal cancer patient, demonstrating that the chemosensitivity of PDO and PDOXwE coincided with the clinical data. These results suggest that PDOXwE is an improvement of PDOX and is suitable as an evaluation model for precision medicine.


Introduction
Precision medicine, which encompasses personalized medicine, is important for the implementation of optimized anticancer therapy. Moreover, preclinical evaluation models are indispensable in screening for the sensitivity of anticancer drugs. Patient-derived organoids (PDOs) have been established as an in vitro model for various cancers [1]. is model embodies the function and genetic information of a patient's tissue and could be maintained for an expanded period [2]. PDOs have emerged as a high-throughput screening system in anticancer drug prognosis and development [3]. Although PDOs recapitulate the features of tissues, they cannot emulate the tumor microenvironment. erefore, several anticancer drugs that interrupt the crosstalk between cancer cells and surrounding cells could not be evaluated using this model. Consequently, in vivo animal models are required for an accurate prognosis before clinical application. Recently, patient-derived xenograft models (PDXs) directly transplanted with patient tissue and PDO-bearing xenograft models (PDOXs) have been established [4]. By contrast, PDXs and PDOX can preserve cancer heterogeneity and the genetic information of the patient's tissue, as well as mimic the tumor microenvironment. However, the establishment rate of PDXs remain low [5,6] because patient tissue is composed of cancer cells and stroma and this ratio is not constant. e PDOX could improve on the disadvantages of PDX. Nevertheless, the PDX and PDOX protocols have not yet been optimized because they differ depending on the cancer type. Furthermore, the time required for the establishment of PDX and PDOX remains a limiting factor in their application on anticancer drug screening.
In this study, an advanced xenograft model was developed using colorectal cancer (CRC) patient-derived tissues to overcome the limitations of PDOX. CRC is known to cause liver metastasis, and its progression can lead to death [7]. us, a strategy for the inhibition of CRC progression is important in increasing survival rate. In particular, there is an urgent need for anticancer drugs optimized for accuracy in anticancer therapy applications. Hence, we investigated the sensitivity of anticancer drugs in advanced xenograft models of CRC and liver metastatic CRC.

Human Tissue Acquisition and Patient Treatment.
e protocol for this section of the study was approved by the Ethics Committee of the Korea Cancer Center Hospital (approval no. KIRAMS-2017-07-001 and KIRAMS-2017-09-009) and was performed in accordance with the approved guidelines and regulations of the institution. All samples were obtained from patients who provided written informed consent for the use of their tissues. Surgically resected liver metastatic intestinal cancer tissue (LMT) and endoscopic biopsy intestinal cancer tissues were obtained from patients diagnosed with CRC and treated at the Korea Cancer Center Hospital. e collected samples were also histologically verified as adenocarcinoma by a pathologist using hematoxylin and eosin (H&E) staining. e isolation of the tumor epithelium was performed as previously described with minor modifications [8,9].
For chemotherapy, the LMT patient was treated with an irinotecan-based regimen (FOLFIRI). e FOLFIRI regimen consisted of 180 mg/m 2 irinotecan, 400 mg/m 2 bolus 5fluorouracil (5-FU), and 2400 mg/m 2 infusional 5-FU every two weeks. e patient's response to chemotherapy was evaluated after every three cycles with computed tomography (CT, Ingenuity, Philips Healthcare, Amsterdam, the Netherlands) as scored using Response Evaluation Criteria In Solid Tumors 1.1.

Organoid Viability.
LMT organoids in good condition were harvested, passaged, and seeded in 96-well cell culture plates. e organoid density was adjusted to 50-60/10 μL Matrigel with 200 μL culture medium. For drug testing, the organoid culture medium was removed and replaced with a 200 μL drug-containing culture medium: 2.5 mg/mL cetuximab (Erbitux, Merck), irinotecan (I1406, Merck), or oxaliplatin (O9512, Sigma). Organoids were photographed seven days after drug treatment (EVOS FL Cell Imaging System, ermo Fisher Scientific), and cell viability was also evaluated at seven days by the CellTiter 96 Aqueous One Solution cell assay (Promega, G3580) according to the manufacturer's instructions.

Culture of Human Endothelial Colony-Forming Cells.
Endothelial colony-forming cells (ECFCs) were isolated from the adherent mononuclear cell fraction of human peripheral blood using CD31-coated magnetic beads (Invitrogen, MA, USA) as previously described [11]. Isolated ECFCs were expanded on 1% gelatin-coated plates (BD Biosciences, NJ, USA) using an endothelial cell growth medium MV 2 (EGM-MV 2 without hydrocortisone; Pro-moCell, Heidelberg, Germany) supplemented with 10% fetal bovine serum (Atlas Biologicals, CO, USA) and 1% glutamine-penicillin-streptomycin (Gibco, MA, USA). ECFCs between passages seven and ten were used in all of the experiments. e protocol for this section of the study was approved by the institutional review board of Duksung Women's University (IRB Nos. 2017-002-001 and 2018-007-006). Journal of Oncology to any procedural work.
e room conditions were maintained at 20°C, 50% humidity, and a 12/12 h light/dark cycle.
e diet was provided with drinking water ad libitum.
2.6. Organoid-Derived Xenograft Models. Cultured organoids were collected and implanted into the subcutaneous pockets of NRGA mice. For the coimplantation of organoids and ECFCs, ECFCs were prepared at 1 × 106 cells/100 μL in 10% Matrigel (YoungIn Frontier, Korea) and injected subcutaneously around the implanted organoid. To subculture the organoid-derived xenograft model, organoidderived tumors were isolated and sliced into 1-2 mm 3 sections. One piece of tumor tissue was subcutaneously implanted into the second generation of Balb/c nude mice (G2). Subsequently, the G2 xenograft mouse models were used to investigate the efficacy of anticancer drugs. Tumor size was measured using a caliper (Mitutoyo Corporation, Japan) three times per week. e tumor volume was calculated as follows: (1) When the tumor volume reached approximately 100 mm 3 , the mice were randomly divided into groups (n � 5/group).

Tumor Organoid DNA Sequencing and Analysis.
To analyze the mutational status of patient tissues, organoids, and PDOX tissues, DNA extraction and library construction were performed using the Qiagen Gentra Puregene kit (Valencia, CA, USA) and Agilent SureSelect XT library prep kit (Santa Clara, CA, USA). Deep targeted sequencing using Axen Cancer Panel 2 (170 cancer-related genes, Macrogen) and the NextSeq 500 midoutput system platform (Illumina) was conducted on tumor tissues, organoids, and PDOX samples. Libraries consisting of 150 bp paired-end reads were sequenced by highthroughput sequencing using synthesis technology to a depth coverage of approximately 2000x. An oncoplot was used for the visualization of the mutations of the tissue, organoid, and PDOX.

Drug Treatment.
Intraperitoneal injections of the test drugs were administered following this treatment schedule: oxaliplatin (5 mg/kg, three times/week), irinotecan (20 mg/kg, five times/week), and/or cetuximab (10 mg/kg, twice a week).

Statistics.
Data are presented as the mean ± standard deviation. Statistical significance was set at p < 0.05 and was calculated using Student's t test and one-way ANOVA followed by Tukey's post hoc test.

PDOX Maintains Patient-Derived Properties.
e sensitivity of anticancer drugs was predicted by screening using the PDO and PDOX models (Figure 1(a)). In our study, we cotransplanted PDO with ECFCs in NRGA mice (G1) and subcultured PDOX (G1) with ECFCs in Balb/c nude mice (G2). First, we investigated whether PDOX maintained the characteristics of PDO. As shown in Figure 1(b), the gene expression of PDO, PDOX (G1), and PDOX (G2) coincided with each other. Moreover, the establishment period of PDO correlated with that of PDOX (R � 0.6007) (Figure 1(c)). e establishment period of an in vivo model is their limitation in precision medicine applications. Hence, we investigated whether advanced xenograft models can improve the original PDOX.

PDOX with ECFCs Overcomes the Obstacles of PDOX.
e tumor growth of PDOX with ECFCs (PDOXwE) was compared with that of PDOX, because the establishment period of PDOX is an obstacle for its utilization. In 19T-PDO, the establishment of PDOX (G1) failed, but cotransplantation POD with ECFCs showed tumorigenicity (Figure 2(a), left). Furthermore, PDOXwE stimulated tumor growth more than PDOXs in the case of 5T-PDO and 8T-PDO (Figure 2(a), middle and right, respectively). Among them, 5T-PDO was also confirmed to maintain the 5T patient's properties (Figure 2(b)). Additionally, gene expression in PDOXwE coincided with that in PDO (Figure 2(c)).
ese results indicate that PDOXwE overcomes the obstacle of PDOX by enhancing tumorigenicity and tumor growth while maintaining the advantages of PDOX.

Drug Sensitivity Is Consistent in PDO and PDOXwE.
Our results indicate that PDOX drug sensitivity was consistent with that of the patient. e chemotherapeutic efficacy of anticancer drugs was evaluated in PDOXwE and PDO, and the application validity of PDOXwE as an advanced xenograft model is shown in Figure 2. To compare preclinical data with clinical data, we used liver metastatic CRC patient-derived organoids.
As shown in Figures 3(a) and 3(b), the histopathology and DNA sequence analyses demonstrate that PDO and PDOXwE also coincided with the LMT patient's tissue. e expression of several genes was different among the tissue, PDO, and PDOXwE; nevertheless, the gene profile of PDOXwE (G2) for preclinical evaluation was almost similar to that of the tissue. After seven days of observation of the PDO model, the cytotoxicity of cetuximab was not significantly enhanced; by contrast, the combination of cetuximab and irinotecan significantly enhanced cytotoxicity compared to cetuximab alone (Figure 3(c)). On the other hand, the combination of cetuximab and oxaliplatin showed no difference with the use of cetuximab alone. e tumor growth of PDOXwE was significantly suppressed only when the combination of cetuximab and irinotecan was used (Figure 3(d)). Moreover, the chemotherapeutic efficacy of PDOXwE was the same as that in the PDO model. However, tumor size and weight significantly decreased in all drugtreated groups on the final day after the 3-week treatment period (Figure 3(e)). e combination of cetuximab and irinotecan inhibited the suppression of tumor growth, tumor size, and tumor weight. Figure 3, our results suggest that irinotecan is more effective than oxaliplatin in the LMT organoid and the LMT organoid-bearing xenograft models. In the LMT organoid-supplied patient, a liver metastasis of approximately 2 cm was detected at the edge of liver segment IIb.

Monitoring of the LMT Patient Receiving the Irinotecan-Based Regimen. As shown in
us, we decided to use the FOLFIRI regimen for palliative chemotherapy based on the results of preclinical tests. We monitored the chemotherapeutic efficacy every 3 cycles using CT (Figure 4(a)). Four lesions were analyzed in every detection, and the total lesion size was calculated (Figure 4(b)). e best response to chemotherapy was achieved after the 6th cycle, and the patient remained at the stable disease status until the 9th cycle. After the 12th cycle, the size of the target lesions increased by more than 20% of the size of the best response, and we determined that the disease has progressed.

Discussion
In this study, chemotherapeutic efficacy was evaluated in an in vitro and an in vivo model. Drug sensitivity of the LMT patient was extrapolated based on these results and monitored using CT.
In the preclinical test, these models were expected to predict chemosensitivity in cancer patients. PDO and PDOX models must represent some of the cancer patients' attributes (growth and gene expression); thus, PDOX exhibits different sensitivities to anticancer drugs depending on a patient's organoid (Supplementary Figure 1). Practically, time constraints are addressed to applicate the results of these preclinical assessments for cancer patients.
us, advanced xenograft models were developed through the cotransplantation of PDO and ECFCs (Figure 1(a)). e PDOXwE model improved the period of establishment, which is a limitation in the utilization of such models for preclinical evaluation (Figure 2). Moreover, our results suggest that PDOXwE could have an edge as an in vivo model and, particularly, as an anticancer drug screening system for precision medicine.
PDO is emerging as a model of pathophysiology because it exhibits intratumor heterogeneity [12]. Furthermore, PDO has maintainability with long-term expansion culture [2].
us, PDO could be used for high-throughput screening in an in vitro model. PDO must be an attractive in vitro model for development of anticancer drugs. Nevertheless, PDO could not show tumor-stroma interaction and the integratable immune system [13]. erefore, indirect targeted anticancer agents, such as antiangiogenic agents and inhibitors of crosstalk between cancer cells and surrounding cells, are not suitable for evaluation in PDO.
To remedy PDO's shortcomings, the evaluation of anticancer agents in an in vivo model was required for development of chemotherapeutic agents. Transplanted materials of xenograft models for anticancer drug screening have been developed from human cancer cell lines to PDOs [4,14]. Xenograft models could effectively evaluate the chemotherapeutic efficacy. In general, standard protocols have been established for human-cancer-cell-derived xenograft models. us, this model has been used easily for a long time in the field of anticancer drug development. However, this model could not show the diverse characteristics of cancer patients [15]. e PDX model improves the obstacles of the human-cancer-cell-derived xenograft model [14]. e PDX model as an avatar model represents genetic alterations and pathohistological characteristics of cancer patients [16]. Unfortunately, this model has several limitations including long establishment period and low engraftment rate [16], which may be one of the major hurdles to apply PDX models to the effective anticancer drug screening system. As an improving model, transplantation of PDO into immunodeficient mouse has been tried. e PDOX model retains the advantages of PDX. us, it could predict anticancer drug susceptibility just like in patients. However, these models could only be used with some organoids. Furthermore, an optimized protocol of PDOX for stable engraftment rate and rapid establishment period is not yet found. Unsolved limitations may be due to the insufficient blood supply to the cells within the PDO after implantation. Current methodologies have been improving on    ECFCs are circulating endothelial progenitor cells and contribute to neovascularization in many postnatal pathophysiological conditions. For example, circulating ECFCs are recruited into the ischemic tissues, where they are incorporated into the vascular endothelial lining and differentiate into endothelial cells to form new blood vessels [17]. Furthermore, it has been reported that about 40% of vascular endothelial cells within the tumor region are derived from ECFCs originated from the bone marrow [18]. Moreover, ECFCs have adhesiveness and migratory activities toward tumor [19]. Human-originated blood vessels are made to the vasculature of xenograft models by ECFCs around PDO. Similar to our results, human-derived blood vessels could be observed in tumor tissues of a breast cancer xenograft model by coinjection of MDA-MB-231 cells and ECFCs [20]. In our study, the transplanted PDO exhibited faster tumorigenicity and tumor growth through the blood vessels newly formed by ECFCs ( Figure 2). us, by application of ECFCs to the PDOX models, PDOXwE can be a novel strategy to establish an effective and practicable   screening system for the personalized cancer medicine. Furthermore, our results showed that PDOXwE preserved patient genetic information, and some of the variations in gene expression were negligible (Figure 2). During anticancer drug screening, the drug sensitivity was observed to be coincident between PDO and PDOXwE (Figures 3(c) and  3(d)).
When the result of preclinical assessment was applied in chemotherapy, irinotecan was effective in the chemotherapy of the patient (Figure 4). ese results indicated that PDO and PDOXwE models could predict chemotherapeutic efficacy in a patient.

Conclusions
In this study, PDOXwE as an advanced xenograft model was established by cotransplantation of organoids and ECFCs. e advanced xenograft model has a short establishment period and high success rate. e advanced xenograft model is an edge in preclinical modeling for precision medicine.
us, the PDOXwE model is anticipated to be applied in precision medicine in the field of chemotherapy.

Data Availability
e data used to support the findings of this study are included within the article and the supplementary information file.

Conflicts of Interest
e authors declare that there are no conflicts of interest regarding the publication of this paper.