Published: July 29th, 2016
Human tumor xenografts are vascularized and infiltrated by cells of mouse origin during the growth phase in vivo. To circumvent the bias caused by these contaminating cells during downstream analysis, we developed a method allowing for comprehensive depletion of all mouse cells by magnetic cell sorting.
The use of in vitro cell line models for cancer research has been a useful tool. However, it has been shown that these models fail to reliably mimic patient tumors in different assays1. Human tumor xenografts represent the gold standard with respect to tumor biology, drug discovery, and metastasis research 2-4. Tumor xenografts can be derived from different types of material like tumor cell lines, tumor tissue from primary patient tumors4 or serially transplanted tumors. When propagated in vivo, xenografted tissue is infiltrated and vascularized by cells of mouse origin. Multiple factors such as the tumor entity, the origin of xenografted material, growth rate and region of transplantation influence the composition and the amount of mouse cells present in tumor xenografts. However, even when these factors are kept constant, the degree of mouse cell contamination is highly variable.
Contaminating mouse cells significantly impair downstream analyses of human tumor xenografts. As mouse fibroblasts show high plating efficacies and proliferation rates, they tend to overgrow cultures of human tumor cells, especially slowly proliferating subpopulations. Mouse cell derived DNA, mRNA, and protein components can bias downstream gene expression analysis, next-generation sequencing, as well as proteome analysis 5. To overcome these limitations, we have developed a fast and easy method to isolate untouched human tumor cells from xenografted tumor tissue. This procedure is based on the comprehensive depletion of cells of mouse origin by combining automated tissue dissociation with the benchtop tissue dissociator and magnetic cell sorting. Here, we demonstrate that human target cells can be can be obtained with purities higher than 96% within less than 20 min independent of the tumor type.
Solid human tumors consist of multiple physiological as well as neoplastic cell types. They form heterogeneous tissues with complex biological structures. Biological processes like tumor formation, progression and responses towards therapies are not yet fully understood. In vitro cell culture models represent a useful tool to study and understand tumor biology. However, they can only partly mirror structures and processes found in tumor tissues. Reliable and stable preclinical human tumor models are a prerequisite for the development of anti-tumor drugs and therapies4,6 as well as for understanding tumor biology and the interaction of tumor cells and their environment.
Human tumor xenograft models derived from primary patient tumors show high relations to the tissue of origin regarding histological architecture, interactions with micro-environmental structures, metastatic potential and responses to drugs7. Even when tumor xenografts are derived from cultured cells or cell lines they more closely mimic patient tumors, therefore showing higher validity in most assays compared to in vitro cell culture models. These features make them the gold standard of preclinical models4. Besides its application in cancer research, xenotransplantation of human cells into mice is also frequently used in stem cell research to determine the stemness and differentiation potential of a target population8.
It has been shown that human microvasculature and human immune cells are replaced upon in vivo propagation of human tumors2,9. Factors such as the tumor subtype, growth rate, and region of transplantation have major impacts on the global level of infiltration as well as the composition of the mouse cell types found. However, even when these factors are kept constant, the amount and composition of mouse cells are highly variable.
Downstream analyses of xenografted tissues are often challenged by cells of murine origin. Primary cultures of human tumor cells from xenografts are frequently overgrown by fibroblasts. Besides hampering the generation of tumor cell lines in vitro, downstream assays such as drug cytotoxicity testing or pharmacokinetics are biased since in silico correction for effects originating from contaminating mouse cells is impossible in most cases. Beyond that, the only partly elucidated cross-talk between mouse fibroblasts and human tumor cells has a direct impact on experimental outcomes of studies10. Furthermore, the most widely unpredictable variation of infiltrating mouse cells aggravates accurate molecular downstream analyses. In NGS or proteome analyses each mouse-derived signal measured instead of a human tumor signal directly decreases sequencing sensitivity. Also microarray based expression analyses are challenged by murine nucleic acids possibly cross hybridizing to human probes.
In order to circumvent the obstacles of contaminating mouse cells in downstream analyses of human tumor cells from xenograft models different approaches have been proposed. In many studies desired target cells for downstream analysis are isolated by utilizing markers or combinations of markers exclusively expressed on human tumor cells. However, the lack of reliable markers for positive identification of human tumor cells frequently represents a big hurdle. Even broadly expressed markers, such as EpCAM on human carcinomas, frequently show tumor intrinsic expression differences11. This increases the risk that low expressing subpopulations, e.g., tumor cells undergone epithelial-to-mesenchymal transition, are lost during isolation. In addition, the direct binding of a selection agent to the target cell might influence the subsequent analysis results. Attempts of depleting mouse cells by using a combination of antibodies specific for murine CD45 and MHC class I epitopes have also been made12. However, this marker combination only detects a subset of mouse cells. Another approach is to enhance analyzing processes by software. Nevertheless, all these approaches are either not suitable for any kind of experimental setup or for any tumor entity and transplantation method.
In this study we present a novel and fast method for comprehensively depleting mouse cells from xenografted tissues independent of the mouse strain and tissue of origin. In screenings on multiple target organs and tissues for transplantation of xenografts (including skin, lung, brain, kidney and skeletal muscle) we were able to identify a combination of antibodies allowing for comprehensively detecting mouse cells. These antibodies were coupled to superparamagnetic particles, titrated and optimized for depletion by using magnetic cell sorting. As only mouse specific antibodies are used, human target cells stay "untouched" and the method is not limited to xenotransplanted tumor tissue. We demonstrate that comprehensively removing mouse cells from xenografted tumor samples standardizes and facilitates cultures of human tumor cells. Furthermore, we show that the analysis of human tumor xenografts by next generation sequencing is significantly improved. As this effect was observed although a human sequence specific selection has been carried out during exome enrichment, the influence on whole genome and whole transcriptome sequencing are expected to be even more prominent. Taken together, removal of mouse cells using this novel method facilitates cultivation of human tumor cells and significantly improves the downstream analyses of human tumor xenografts.
All animal experiments were performed in accordance with local ethical and legal guidelines of Regierungspräsidium Freiburg Referat 35.
1. Reagent Preparation
NOTE: Before starting the experiment, prepare the following reagents from the dissociation kit:
2. Tumor Dissociation Protocol
NOTE: In this study patient-derived xenografts of non-small cell lung cancer (NSCLC), renal cancer and bladder were used. The tumors were generated by implanting xenografts of non-small cell lung cancer (NSCLC) renal cancer and bladder in 4 - 6 weeks old female NMRI nu/nu mice as described previously13.
NOTE: As this protocol is completely independent of the tumor type, it can be used for any kind of patient or cell line-derived tumor as well as other types of human xenotransplanted tissue without modification of the protocol.
3. Mouse Cell Depletion by Magnetic Cell Separation Technology
NOTE: Volumes for magnetic labeling given below are for up to 2 x 106 tumor cells or 107 total cells including red blood cells. Depending on the tumor size lower or higher cell numbers will be obtained. In case of fewer cells, use the same volumes as indicated. When using higher cell numbers, scale up all reagent volumes and total volumes accordingly (e.g., for 4 x 106 tumor cells or 2 x 107 total cells, use twice the volume of all indicated reagent volumes and total volumes).
4. Downstream Analyses
Different approaches have been proposed to identify or deplete mouse cells from xenografted human tumors, for example a combination of mouse-specific antibodies against CD45 and MHC class I. However, only subpopulations of mouse cells were detected using these combinations whereas the proposed combination detected CD45 and MHC class I expressing mouse cells as well as the rest of the mouse cells found in target tissues for transplantation (Figure 1). Utilizing this novel combination of antibodies for magnetic cell sorting, human tumor cells could be isolated independent of tumor type (Figure 2).
Cells obtained from magnetic cell separation procedure-based mouse cell removal could be cultured, leading to pure cultures of human tumor cells (Figure 4). In addition, viable mouse cells could be obtained and cultured from the same sample. Besides removal of mouse cells, also debris was removed upon mouse cell depletion (MCD) (Figure 3).
The comprehensive depletion of mouse cells as well as debris removal lead to improved molecular downstream analyses. This was demonstrated by comparing results obtained from whole exome sequencing (WES) carried out on bulk tumor pieces and mouse cell depleted samples from three different patient-derived xenograft tumor models (Figures 5 - 9). Our data indicate that the removal of mouse cells before performing WES on xenograft tumor samples not only significantly increases the total amount of reads (Figure 5) but also considerably reduces the number of host derived reads mapped to the human reference genome (Figure 6B). As these erroneously mapped mouse reads frequently interfere with SNP calling, we observed a strong reduction of falsely predicted SNPs after MCD (Figures 7 - 8). Finally, we showed that the in silico removal of mouse-derived reads by bioinformatics methods could not fully replace the experimental procedure, as an unambiguous sequence-based assignment to the species of origin was not possible for all reads. Furthermore, the in silico procedure could not correct for the enhanced quality and concomitant higher read coverage of the in vitro depleted samples (Figure 9).
Figure 1. Establishing an Antibody Cocktail Recognizing All Mouse Cells across Multiple Organs14. Screenings in multiple organs, including skin, lung, brain, kidney and skeletal muscle, have been performed. These organs represent major target tissues for xenotransplantation. Combinations such as those recognizing mouse CD45 and MHC class I epitopes have already been used in order to deplete mouse cells after xenotransplantation. However, these marker combinations recognized only a subset of mouse cells in all analyzed tissues. In previous screenings a combination of five antibodies has been identified allowing for the comprehensive detection of all cells from mouse origin. This panel includes red blood cells and is independent of the tissue of origin (A, B, and data not shown). Modified from14. Please click here to view a larger version of this figure.
Figure 2. Reliable and Fast Depletion of Mouse Cells15. Conjugates of the antibody combination with superparamagnetic nanoparticles were used to develop an optimized protocol for depletion of mouse cells from human tumor xenografts by magnetic cell sorting (magnetic cell separation) (A). This novel protocol allowed for the elimination of > 99% of contaminating mouse cells in less than 20 min, irrespective of the tumor type. Cell fractions obtained from magnetic cell separation procedure were labeled with the pan-mouse antibody cocktail and a human specific antibody against CD326 (EpCAM) (B). Modified from15. Please click here to view a larger version of this figure.
Figure 3. Isolation of Human Glioblastoma Cells. Mouse Cell Depletion Kit was used to isolate human glioblastoma cells from adult mouse brain. During dissociation of neural tissue high amounts of debris is commonly generated. MCDK efficiently depletes dead cells and debris as seen by the plot showing cell size versus cell granularity. With this conjugate cocktail, it was possible to eliminate > 99% of the contaminating mouse cells and > 60% of the debris. Please click here to view a larger version of this figure.
Figure 4. Depletion of Mouse Cells Facilitates Downstream Cultures of Human Tumor Cells14. Mouse fibroblasts frequently hamper the cultivation of human tumor cells after dissociation of transplanted tissues or lead to heterogeneous cultures. Fibroblasts attach and expand more efficiently, thereby overgrowing target cells. This perturbs in vitro cell culture assays (e.g., drug cytotoxicity testing), since mathematical correction for effects arising from contaminating mouse cells is impossible in most cases. The negative (B) and positive fractions (C) after mouse cell depletion were cultured for three days, fixed and stained for a mouse-specific fibroblast marker (Vimentin) and the human-specific tumor marker CD326 (EpCAM). As a control, the original fraction containing unseparated cells (A) was cultured and stained as well. Even after three days of culture, a nearly pure population of human tumor cells was observed in the negative fraction (B), whereas the unsorted fraction (A) was almost overgrown by fibroblasts. Only a minor portion of target cells was lost to the positive fraction (C). Modified from14. Please click here to view a larger version of this figure.
Figure 5. Whole Exome Sequencing (WES) of Tumor Samples Prior to and after Mouse Cell Depletion (MCD)15. We conducted WES on three different xenograft models derived from human kidney, lung and bladder cancer in order to assess the impact of MCD on the quality of next-generation sequencing data. DNA from bulk tumor and from isolated tumor cells after mouse cell depletion was used to generate exome-captured sequencing libraries applying an exome capture kit. For sequencing on a desktop sequencer instrument, a desktop sequencer reagent kit 150 cycles, was utilized to generate 75-bp paired-end reads. A significant increase (p < 0.05) in cluster density (A) as well as an average increase in read counts of 33% (B) was observed for the samples depleted of mouse cells, indicating improved sample quality. Correspondingly, a strong reduction of debris and dead cells upon MCD could also be demonstrated by flow cytometry analysis (see Figure 3). Modified from15. Please click here to view a larger version of this figure.
Figure 6. Mouse Cell Depletion (MCD) Strongly Reduces Number of Erroneously Mapped Mouse Reads after Whole Exome Sequencing (WES)15. As capture oligonucleotides used for targeted enrichment of protein-coding sequences were designed based on the human genome, an initial pre-enrichment of DNA fragments of human origin from the mixture of mouse and human cells was expected. In order to assess the number of capture oligonucleotides that might cross-hybridize with mouse genomic DNA, we conducted BLAST searches of each single rapid capture exome probe against mouse genome. The resulting alignment parameters were used to determine possible cross-hybridization. Depending on the selection thresholds (alignment length, no. of mismatches, no. of gaps), we predicted a cross-reactivity of 5 - 10% of capture probes with mouse transcripts (data not shown). After adapter clipping (trimmomatic v0.3216), we mapped the reads of all samples against human and mouse genomes (bwa v0.7.1217) and determined their putative origin based on the respective alignment parameters (LINUX shell, command-line Perl) (A). An average of 12% of reads derived from bulk tumor samples was attributed to mouse cells. This amount could be reduced to 0.3% by prior depletion of mouse cells (B). As on average 15% of the mouse-derived reads mapped erroneously to the human genome, corresponding to 1.9% of total reads in the bulk tumor samples and 0.04% in the isolated tumor cells, a strong positive influence of mouse cell depletion on downstream analyses can be expected. Figure 6B exemplifies the detailed read assignment for bulk tumor and isolated human tumor cells derived from the bladder cancer xenograft. Modified from15. Please click here to view a larger version of this figure.
Figure 7. MCD Strongly Reduces the Number of Falsely Predicted SNPs15. In order to determine the impact of mouse reads mapped to the human genome, we determined the number of predicted SNPs for the xenograft samples prior to and after MCD. As no healthy tissue was available for comparison, a SNP was defined as a difference between the sequenced sample and the reference genome (hg19). After removal of duplicate reads, SNP and INDEL calling was conducted18 and was restricted to the regions targeted by an exome capture kit. 63 ± 10% of all SNPs predicted for the bulk tumor samples were no longer detected after mouse cell depletion, 18 ± 1% were specific for the isolated human tumor cells (A). While the former were mainly caused by erroneously mapped mouse reads the latter seemed to be detected due to higher read counts and accordingly higher coverage within the isolated human tumor cell samples. This effect was also visible for predicted INDELs (B). Modified from15. Please click here to view a larger version of this figure.
Figure 8. MCD Improves the Prediction of High-impact SNPs15. (A) Impact of MCD on SNP prediction within a protein-coding exon of the POLA1 gene19 is exemplified. While erroneously mapped mouse reads caused a number of falsely predicted SNPs in the bulk kidney cancer xenograft, these SNPs were missing after MCD. In addition, MCD also improved the prediction of high-impact SNPs. For example, mouse reads mapped to the human reference genome in the bulk tumor sample resulted in the wrongly predicted destruction of the start codon of the GRIA3 gene (B). Modified from15. Please click here to view a larger version of this figure.
Figure 9. In Silico Depletion of Mouse-derived Reads Cannot Fully Replace In Vitro MCD15. The reads predicted to be of mouse origin were computationally removed from the sequencing data, and SNP calling was repeated. By this approach, the number of SNPs predicted for the bulk tumor samples was considerably reduced from 63% to 8.5% of the total number of SNPs (compare with Figure 8). However, this in silico approach could not completely replace in vitro MCD, especially if the improved sample quality and the concomitant increase in read counts and coverage are considered. Modified from15. Please click here to view a larger version of this figure.
We have developed a fast and easy method to isolate untouched human tumor cells from xenografted tumor tissue. This procedure is based on the comprehensive depletion of cells of mouse origin by combining automated tissue dissociation and magnetic cell sorting. Besides depletion of all mouse cells also the amount of dead cells and debris is significantly reduced in the target cell fraction. Taken together, this method significantly improves molecular downstream analyses and cultivation of human tumor cells.
In contrast to alternative methods, there is no need for knowledge of tumor cell specific markers, adjustment to varying compositions of mouse cells or establishment of algorithms. The presented method is independent of the mouse strain and tumor type. Therefore, a bias through isolation of human tumor subpopulations on the basis of single human specific markers which may vary in expression, as shown for EpCAM11, is avoided. Furthermore, it is not limited to tumor material but can be used for any kind of xenotransplanted tissue as mouse cells are targeted instead of specific human tumor subpopulations (data not shown).
Comprehensive removal of mouse cells is facilitated by targeting surface epitopes exclusively expressed on cells of mouse origin. Apart from the well-established method for tumor dissociation as presented in this study, there are many procedures using different kinds of enzymes. Preservation of cell surface epitopes is a prerequisite for this novel method, but also for accurate analyses of human tumor cell populations. Therefore, it is crucial that gentle enzymes are used for digestion of tumor tissue. Thus, mouse cell depletion can only be used in combination with enzymatic digestion procedures that evidently do not affect epitopes targeted during mouse cell depletion procedure. In case of doubt this can only be verified by the respective manufacturer.
This method also works for human circulating tumor cells (CTCs). However, as frequencies of target cells are usually < 0.1% removal of mouse cells additionally requires red blood cell lysis and purities of more than 50% cannot be expected. In this case, mouse cell depletion can be used as pre-enrichment of CTCs followed by an enrichment using positive markers (data not shown).
Removal of mouse cells significantly improves the culture of human tumor cells by avoiding culture overgrow of mouse fibroblasts. In addition, the analysis of human tumor xenografts by next-generation sequencing is significantly improved and standardized. As this effect was observed although a targeted human sequence-specific selection has been carried out during exome enrichment, the influence on whole genome and whole transcriptome sequencing are expected to be even more prominent.
Moreover, the presented method facilitates accurate molecular studies of tumor subpopulations within xenotransplanted human tumors. Removal of mouse cells from a respective sample in the first step and subsequently sorting human tumor cells in two different subpopulations enables direct molecular comparison of the tumor subpopulation of interest to human bulk tumor cells20. Differential gene expression between tumor cell subtypes can be more reliably assessed as it is not affected by cross-hybridization of mouse-derived molecules.
David Agorku, Stefan Tomiuk, Stefan Wild, Silvia Rüberg, Lisa Zatrieb, Andreas Bosio and Olaf Hardt are or were employees of Miltenyi Biotec GmbH, Germany. Kerstin Klingner and Julia Schüler are employees of Oncotest GmbH, Germany.
We are grateful to Janina Kuhl, Nadine Chelius, Petra Kussmann, Lena Willnow and Dorothee Lenhard for excellent technical assistance.
|Tumor Dissociatio Kit, human
|contains enzymes H, R and A; see data sheet for reconstitution instructions
|gentleMACS Octo Dissociator with Heaters
|MACS SmartStrainer (70 µm)
|Petri dish 100 mm
|Mouse Cell Depletion Kit
|use 1x PBS for PEB buffer
|use final concentration of 2 mM in PEB buffer
|or use 5 mg/L BSA in PEB buffer
|MACS LS columns
|PreSep filters (70 µm)
|MACSmix Tube Rotator
|use 1:40 for immunofluorescence staining
|dilute to a final concentration of 0.1 µg/ml in 0.1 M sodium bicarbonate
|Inside Stain Kit
|InsideFix from this kit can be used for fixation step during immunofluorescence staining
|FcR Blocking Reagent, mouse
|Nextera Rapid Capture Enrichment Kit
|MiSeq Reagent Kit v3
Copyright © 2024 MyJoVE Corporation. All rights reserved