Testing Targeted Therapies in Cancer using Structural DNA Alteration Analysis and Patient-Derived Xenografts

Summary

Here we present a protocol to test the efficacy of targeted therapies selected based on the genomic makeup of a tumor. The protocol describes identification and validation of structural DNA rearrangements, engraftment of patients’ tumors into mice and testing responses to corresponding drugs.

Abstract

We present here an integrative approach for testing efficacy of targeted therapies that combines the next generation sequencing technolo-gies, therapeutic target analyses and drug response monitoring using patient derived xenografts (PDX). This strategy was validated using ovarian tumors as an example. The mate-pair next generation sequencing (MPseq) protocol was used to identify structural alterations and followed by analysis of potentially targetable alterations. Human tumors grown in immunocompromised mice were treated with drugs selected based on the genomic analyses. Results demonstrated a good correlation between the predicted and the observed responses in the PDX model. The presented approach can be used to test the efficacy of combination treatments and aid personalized treatment for patients with recurrent cancer, specifically in cases when standard therapy fails and there is a need to use drugs off label.

Introduction

Patient-derived xenografts (PDXs), which are generated from the implantation of patient tumor pieces into immunodeficient mice, have emerged as a powerful preclinical model to aid personalized anti-cancer care. PDX models have been successfully developed for a variety of human malignancies. These include breast and ovarian cancers, malignant melanoma, colorectal cancer, pancreatic adenocarcinoma, and non-small cell lung cancer^1,2,3,4,5. Tumor tissue can be implanted orthotopically or heterotopically. The former, considered more accurate but technically difficult, involves transplantation directly into the organ of tumor origin. These types of models are believed to precisely mimic histology of the original tumor due to the “natural’ microenvironment for the tumor^6,7. For example, orthotopic transplantation into the bursa of the mouse ovary resulted in tumor dissemination into the peritoneal cavity and the production of ascites, typical of ovarian cancer⁸. Similarly, injection of breast tumors into the thoracic instead of the abdominal mammary gland affected the PDX success rate and behavior⁹. However, orthotopic models require sophisticated imaging systems to monitor tumor growth. Heterotopic implantation of solid tumor is typically performed by implanting tissue into the subcutaneous flank of a mouse which allows for easier monitoring of tumor growth and is less expensive and time consuming⁷. However, tumors grown subcutaneously rarely metastasize unlike as observed in the case of orthotopic implantation¹⁰.

The success rate of engraftment has been shown to vary and greatly depend on the tumor type. More aggressive tumors and tissue specimens containing a higher percent of tumor cells were reported to have better success rates^12,13. Consistent with this, tumors derived from metastatic sites were shown to engraft at frequencies of 50–80%, while those from primary sites engraft at frequencies as low as 14%¹². In contrast, tissue containing necrotic cells and fewer viable tumor cells engraft poorly. Tumor growth can also be promoted by the addition of basement membrane matrix proteins into the tissue mix at the time of the injection into mice¹⁴ without compromising properties of the original tumor. The size and number of tissue pieces intended for implantation were also found to affect the success rate of engraftment. Greater tumor take-rates were reported for implantation in the sub-renal capsule compared to subcutaneous implantation due to the ability of the sub-renal capsule to maintain the original tumor stroma and provide the host stromal cells as well¹⁵.

Most studies use NOD/SCID immunodeficient mice, which lack natural killer cells¹⁶ and have been shown to increase the tumor engraftment, growth and metastasis compared to other strains¹⁴. However, additional monitoring is required as they may develop thymic lymphomas as early as 3-4 month of age¹³. In ovarian tumor transplants grown in SCID mice, the outgrowth of B cells was successfully inhibited by rituximab, preventing the development of lymphomas but without impacting the engraftment of ovarian tumors¹⁷.

More recently, NSG (NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ) mice, carrying a null mutation in the gene encoding the interleukin 2 receptor gamma chain¹⁸, became a frequently used strain for the generation of PDX models. Tumors from established PDX models passaged to future generations of mice are reported to retain histological and molecular properties for 3 to 6 generations^19,20. Numerous studies have shown that the treatment outcomes in PDX models mimic those of their corresponding patients^{2,3,4,21,22,23}. The response rate to chemotherapy in PDX models for non-small lung cancer and colorectal carcinomas was similar to that in clinical trials for the same drugs^24,25. Studies conducted in PDX models, developed for patients enrolled in clinical trials, demonstrated responses to tested drugs similar to those observed clinically in corresponding patients^2,3,4.

High–throughput genomic analyses of a patient tumor in conjunction with PDX models provide a powerful tool to study correlations between specific genomic alterations and a therapeutic response. These have been described in a few publications^26,27. For example, therapeutic responses to the EGFR inhibitor cetuximab in a set of colorectal PDX models carrying EGFR amplification, paralleled clinical responses to cetuximab in patients²⁸.

There are a few challenges associated with the development and application of PDX models. Among those is tumor heterogeneity^29,30 that may compromise the accuracy of treatment response interpretation as a single cell clone with higher proliferative capacity within a PDX can outgrow the other ones³¹, thus resulting in a loss of heterogeneity. Additionally, when single tumor biopsies are used to develop PDX, some of the cell populations may be missed and will not be represented in the final graft. Multiple samples from the same tumor are recommended for implantation to resolve this issue. Although PDX tumors tend to contain all the cell types of the original donor tumor, these cells are gradually substituted by those of murine origin³. The interplay between murine stroma and human tumor cells in PDX models is not well understood. Nevertheless, stromal cells were shown to recapitulate tumor microenvironment³³.

Despite these limitations, PDX models remain among the most valuable tools for translational research as well as personalized medicine for selecting patient therapies. Major applications of PDXs include biomarker discovery and drug testing. PDX models are also successfully used to study drug resistance mechanisms and identify strategies to overcome drug resistance^34,35. The approach described in the present manuscript allows the researcher to identify potential therapeutic targets in human tumors and to assess the efficacy of corresponding drugs in vivo, in mice harboring engrafted tumors which were initially genomically characterized. The protocol uses ovarian tumors engrafted intraperitoneally but is applicable to any type of tumor sufficiently aggressive to grow in mice^2,3,12.

Protocol

Fresh tissues from consenting patients with ovarian cancer were collected at the time of debulking surgery according to a protocol approved by Mayo Clinic Institutional Review Board (IRB). All animal procedures and treatments used in this protocol were approved by Mayo Clinic Institutional Animal Care and Use Committee (IACUC) and followed animal care guidelines.

1. Mate pair sequencing and analyses

NOTE: Either fresh or flash frozen tissue must be used for mate pair (MPseq) sequencing. Paraffin embedded material is not suitable because it contains fragmented DNA.

1. Isolate DNA from frozen tumor tissue. Use original human specimen obtained from surgical material or biopsy³⁶.
2. Use 1000 ng of DNA to make MPseq libraries and sequence as 2 samples per lane on the next generation sequencer (see Table of Materials)³⁶.
3. Analyze data using a set of algorithms to detect large chromosomal aberrations (deletions, insertions, amplifications, inversions and translocations) as described earlier^36,37.

2. Selection of therapeutic targets

1. Use the open access Panda tool (Pathway and Annotation) or an analogous tool to identify targetable alterations (http://bioinformaticstools.mayo.edu/Panda).
   1. Make a list of genes that are identified by MSeq as altered, as a simple tab-delimited file, using standard accepted gene symbols.
      NOTE: The included example features analysis of amplifications and gains.
   2. Add a “#” sign to the header line of the list to ensure that the table header is transferred to the pathway level view of the software.
   3. Upload the file by clicking the Upload Annotation Set navigation tab.
   4. Assign a single icon from the menu to represent the underlying data by clicking on the icon of choice and then clicking the Finalize tab.
   5. After the annotation files are uploaded, identify a column that displays the number of annotated genes per pathway. This is the last column on the right.
   6. Use Pathway Filter on the upper left of the main window to show pathways containing genes of interest.
   7. Identify pathways that have more annotated genes than is expected by chance. Use the function located under the Enrichment tab.
   8. Select a database to display potentially druggable genes from Preset Annotation by checking an appropriate icon on the left of the main window (e.g., DGIdb, PharmGKB).
   9. Select pathway for visualization by clicking on its name displayed in the Pathway Viewer page.
      NOTE: Icons representing each annotation set are shown next to the associated gene. Click on the gene of interest to open the corresponding GeneCards webpage.
   10. Select the pathways that showed the annotated genes of interest (i.e., altered in a given tumor) and “hits” for potential drugs for further analysis.
2. Use a database containing drugs approved for clinical application (https://clinicaltrials.gov/) to cross-reference the identified targets.
3. Prioritize targetable alterations for further testing in PDX models by performing a literature review (e.g., PubMed) to confirm relevance to the biology of a particular type of tumor.

3. Validation of genomic rearrangements by PCR and Sanger sequencing

1. Design primers using sequencing reads obtained from MPseq data.
   1. Select a junction of interest for the validation (i.e., potential therapeutic target) based on MPseq analyses.
   2. Design primers directionally such that the amplicon contains the junction. Design 2 primers on each side of the junction, for a total of 4, to increase the chances of amplifying the junction.
      NOTE: Name primers according to the case and chromosome location.
   3. Use standard PCR parameters for the primer design and a software of choice. Select melting temperature (60-62 ˚C) and the GC content (40-60%). Make sure that the primer sequence does not form primer dimers, palindromes or hairpin loops.
   4. Confirm that the primer sequence lacks homology to other areas of the human genome by checking it using BLAT (http://genome.ucsc.edu/cgi-bin/hgBlat?command=start).
2. Run a PCR to amplify the junction of interest.
   1. Dilute primers with water to 10 mM and combine 10 mL of each primer so that each forward primer is paired with each reverse primer into a primer mix.
      NOTE: Sequences for the primers for the selected example are shown in Table 1.
   2. Label 0.2 mL strip tubes as: C1, T1, C2, T2, C3, T3, C4 and T4 where C=control human genomic DNA (commercial), T=Tumor DNA, isolated from patient or PDX tumor, and the number indicates the primer mix.
   3. Add 1 mL of each primer mix into its labeled tubes.
   4. Prepare the Taq mastermix by combining reagents listed in Table 2, leaving the enzyme in the freezer until needed, adding it to the mix at the very end.
   5. Make 2 master mixes, one for each template DNA, control human DNA and tumor DNA. Add 24 mL of each Taq master mix to its respective strip tube. The total reaction volume is 25 mL. Vortex very briefly and then spin the strip tube down.
   6. Run a PCR in a thermocycler. Use the parameters shown in Table 3. Adjust the annealing temperature so that it is at least 1 °C colder than the melting temperature of the primers.
   7. Store the completed PCR product at -20 °C (long term) or refrigerated at 4 °C (short term) until needed.
   8. Perform electrophoresis at 1-5 V/cm to visualize PCR product using a 1.5% agarose gel. Leave 2 mL aliquot of the product to be used for Sanger sequencing.
3. Perform Sanger sequencing to confirm the junction and identify the exact breakpoint³⁸.
   1. Use the PCR product if the PCR generated a single product (band). Alternatively, cut the band out of gel, purify and submit for Sanger sequencing along with the primer used for amplification.
      NOTE: The tumors for which genomic analyses were performed then are used for implantation in mice.

4. Tumor engraftment and maintenance

1. Set up preparations to engraft tumors into PDX mouse models. Select for engraftment tumors for which genomic analyses will be or has been performed.
   1. Make sure that a supportive infrastructure is in place at the time of the start of the development of any PDX models, including dedicated laboratory and animal facilities, skilled technical staff and detailed standard operating procedures.
   2. Ensure the quick transportation and processing of specimens as speed is crucial for cell viability and successful engraftment.
   3. Use a sterile environment to reduce bacterial and fungal contamination for processing and engrafting specimens.
   4. Handle human specimens with caution, in accordance with institutional policies regarding potentially biohazardous materials, as they may harbor blood-borne pathogens.
   5. Prepare the tissue (0.5-0.7 cm³ in size) for engraftment by putting the surgical specimen into a pre-chilled 50 mL tube with 20 mL of tissue culture media.
      NOTE: The tumor tissue can be fresh or recovered from previously cryopreserved material⁵.
   6. Confirm the tumor content in the specimen by consulting a pathologist.
   7. Place the tumor tissue in a dish containing 10-15 mL of cold PBS, or tissue culture media such as RPMI 1640 or DMEM containing antibiotics (1% penicillin and streptomycin).
   8. Identify and isolate viable tumor material from adjacent normal and necrotic tissue with the help from a pathologist. Use sterile forceps and scalpel to remove necrotic material pointed out by a pathologist.
      NOTE: The tumor can be implanted either intraperitoneally or subcutaneously into the mice. Follow step 4.2 to perform an intraperitoneal implantation or skip to step 4.3 to perform a subcutaneous engraftment. In this study, targeted therapies selected based on genomic analyses were tested in a series of PDX models for high grade serous ovarian cancer with intraperitoneal implantation.
2. Prepare the tissue for intraperitoneal (IP) implantation mincing the tissue with sterile forceps and a scalpel on ice to make pieces approximately 1-1.5 mm³ in size and mixing with cold culture medium. Inject 0.3–0.5 mL of tumor slurry using a 16-17 gauge needle.
   NOTE: All surgical procedures are performed by using aseptic techniques. Sterile gloves, sterile instruments, supplies, and implanted materials were used to reduce the likelihood of infection.
   1. Mix the pieces 1:1 with ice-cold culture medium and inject 100 mL intraperitoneally in at least three female SCID mice.
3. Cut the tumor tissue for subcutaneous engraftment using sterile forceps, scalpel, or surgical scissors into small fragments, roughly 2 x 2 x 2 mm in size, and transfer the fragmented tissues to a pre-chilled Petri dish on ice.
   1. Add the cold basement membrane matrix into the dish with the fragmented tissue (approximately 200 mL per 10 pieces of tissue), mix well, and let the tissue fragments soak in the cold basement membrane matrix for 10 min.
   2. Anesthetize 5 female NOD/SCID mice to prepare them for engraftment.
   3. Inject each mouse intraperitoneally with ketamine (150 mg/kg) and xylazine (10 mg/kg) combination.
   4. Confirm that mouse is properly anesthetized by pinching the tail tip with atraumatic forceps.
   5. Remove gently fully anesthetized mouse from the chamber and put on the mouse a nose cone which has input from vaporizer and output from waste gas scavenging system.
   6. Put vet ointment on mouse eyes to prevent dryness while under anesthesia. Prepare the area where the surgery will be performed. Use aseptic technique to perform the surgery.
   7. Ensure that the surface on which the surgery is going to take is non-porous, sealed and is disinfected prior to surgery.
   8. Start surgery with sterile (by autoclave, gas or chemical sterilization) instruments.
   9. Use gloves to handle instruments and maintain the sterility of the instrument tips throughout the procedure by submerging them in ethanol between surgery steps.
4. Sterilize the surgical site by applying 3 alternating scrubs of iodine and alcohol. Use sterile surgical scissors and forceps to make a 5-10 mm vertical skin incision on both flanks of a mouse.
   1. Insert straight forceps gently into the subcutaneous space to create a pocket large enough for a tumor fragment to be placed under the fat pad.
   2. Use the sterile straight forceps to insert tumor fragments into previously prepared pocket in each of 5 mice.
      NOTE: Place 3-4 pieces of tumor tissue in one pocket.
   3. Close the skin incisions using tissue glue.
   4. Intraperitoneally inject each mouse with 100 mL of Rituximab after implantation to inhibit lymphocyte proliferation.
5. Place the mouse into a cage under a heat lamp for approximately 20 min until recovered from anesthesia. Monitor the mouse vital signs and ensure its sufficient hydration.
6. Return the mouse to the company of other mice after it recovered from anesthesia and started to have food and water. Provide postoperative care and monitoring according to institutional guidelines. Check for signs of pain and distress daily for 3 consecutive days after surgery.
   NOTE: Criteria for pain or distress include necrosis or ulceration, weight loss/body condition scoring, behavioral signs such as activity level, motor function and posture.
7. Routinely check the mice for tumor formation bi-weekly until the tumors reach the size of 0.5 cm in diameter as measured by a caliper.
8. Assess the health score of each mouse as derived from appearance, behavior, and body condition³⁹. Use scores of ≤6 as criteria for moribund mice to be sacrificed by carbon dioxide inhalation.

5. Testing responses to genomically identified targets in PDX models

1. Start the selected targeted treatments when the tumors are palpable and reach 0.5 cm³ as measured by an ultrasound scan.
   1. Prior to performing an ultrasound scan of the abdomen remove the mouse abdominal fur and apply sterile jelly lubricant.
   2. Use an ultrasound machine with a transducer to obtain images with the tumor positioned in cross-section. Make 3 measurements per session for each animal and average the value for a more accurate assessment of tumor size.
   3. Analyze the images using available software⁴⁰.
2. Administer chemotherapy consisting of a mixture of carboplatin at 51 mg/kg and paclitaxel at 15 mg/kg intraperitoneally (IP) once a week for a total treatment duration of 4-6 weeks. Make sure that total volume for injection does not exceed 0.2 mL.
3. Make a MK-2206⁴¹ stock solution in 30% cyclodextrin (e.g., Captisol) and deliver via oral gavage at 120 mg/kg daily for 4 consecutive weeks.
4. Prepare the mouse for oral gavage by hold it by pinching the skin of the back and pinning it back, so that the head and limbs of the mouse are immobilized.
   1. Insert the gavage probe down the back of the throat of the mouse until the probe reaches the esophagus. Make sure that the probe is not inserted too far, as the lungs of the mouse may perforate, causing death.
5. Make a MK-8669⁴² stock solution in ethanol at 25 mg/mL. Dilute it in a vehicle containing 5.2% Tween 80, 5.2% PEG400 in sterile water for IP injections at 10 mg/kg for 5 days every other week, with a total duration of the treatment of 4 weeks.
   NOTE: Volume injected into mice should be 50-120 mL, depending on animal’s weight.
6. Use 7-8 mice per treatment group to have enough statistical power to detect differences^43,44.
7. Assess body weight and general condition of the mice in therapy daily. Withhold drugs if an animal’s weight drops 20% or more from their initial weight.
8. Assess the tumor size weekly by using ultrasound scanning. Make sure that the individual performing the monitoring of the tumor growth is blinded to the treatment to ensure the unbiased scoring of responses.
   NOTE: Smaller laboratories may employ 2 different people to administer treatment and ultrasound monitoring.

Results

Tissue from resected ovarian tumors at the time of debulking surgeries were collected in accordance with IRB guidance and used for 1) genomic characterization and 2) engraftment in immunocompromised mice (Figure 1). Mate-pair sequencing protocol^36,37 was used to identify structural alterations in DNA including losses, gains and amplifications. A representative genome plot illustrating a landscape of genomic changes in one tumor (desi...

Discussion

We describe the approach and protocols we used to conduct a “clinical trial” in PDX models that takes advantage of molecular characteristics of the tumor as obtained by genomic profiling to determine the best choice of drugs for testing. Multiple sequencing platforms are currently used for genomic characterization of primary tumors including whole genome sequencing, RNAseq and customized gene panels. For high grade serous ovarian carcinoma, MPseq to identify structural alterations, DNA rearrangements and copy...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
3M Vetbond	3M, Co.	1469SB
anti-AKT antibody	Cell Signaling Technologies, Inc.	9272
Anti-GAPDH antibody(G-9)	Santa Cruz Biotech. Inc.	sc-365062
Anti-MAPK antibody	Cell Signaling Technologies, Inc.	9926
Anti-phospho-AKT antibody	Cell Signaling Technologies, Inc.	9271
Anti-mTOR antibody	Cell Signaling Technologies, Inc.	2972
Anti-Phospho-mTOR antibody	Cell Signaling Technologies, Inc.	2971
Anti-Phospho-S6 antibody	Cell Signaling Technologies, Inc.	4858
Anti-Rictor antibody	Cell Signaling Technologies, Inc.	2114
Anti-S6 antibody	Cell Signaling Technologies, Inc.	2217
Captisol	ChemScene, Inc.	cs-0731
Carboplatin	NOVAPLUS, Inc.	61703-360-18
DMEM	Mediatech, Inc.	10-013-CV
Easy-A Hi-Fi PCR Cloning Enzyme	Agilent, Inc.	600404-51
Lubricant	Cardinal Healthcare	82-280
Matrigel	Corning, Inc.	356234
McCoy's media	Mediatech, Inc.	10-050-CV
MK-2206	ApexBio, Inc.	A3010
MK-8669	ARIAD Pharmaceuticals, Inc.	AP23573
Nair Sensitive Skin	Church & Dwight Co.	Nair Hair Remover Shower Power Sensitive
NOD/SCID mice	Charles River, Inc.	NOD.CB17-Prkdcscid/NCrCrl
Paclitaxel	NOVAPLUS, Inc.	55390-304-05
PEG400	Millipore Sigma, Inc.	88440-250ML-F
Perjeta	Genetech, Co.	Pertuzumab
Rituximab	Genetech, Co.	Rituxan
RPMI1640	Mediatech, Inc.	10-040-CV
SCID mice	Harlan Laboratories, Inc.	C.B.-17/IcrHsd-PrkdcscidLystbg
SLAx 13-6MHz linear transducer	FUJIFILM SonoSite, Inc	HFL38xp
SonoSite S-series Ultrasound machine	FUJIFILM SonoSite, Inc	SonoSite SII
Tween 80	Millipore Sigma, Inc.	P4780-100ML