|Home | About | Journals | Submit | Contact Us | Français|
Epithelial ovarian cancer (EOC) is the most commonly fatal gynecologic malignancy in developed countries. Most EOC patients are diagnosed at advanced stage when disease has spread beyond the ovary. While many patients initially respond to surgery and chemotherapy, the long term prognosis is generally unfavorable, with recurrence and development of drug resistant disease. There is a critical need to identify new therapeutic agents that prolong disease-free intervals and effectively manage recurrent disease. Murine models of ovarian carcinoma are excellent models to study tumor biology in the search for new treatments for EOC. Described in this unit are methods for establishing xenograft or allograft models of EOC using ovarian carcinoma cell lines, in vivo imaging strategies for detection and quantification of EOC in transgenic and in xenograft/allograft models, and procedures for necropsy and pathological evaluation of experimental animals.
The occurrence of spontaneous ovarian carcinoma in experimental animals is relatively rare with a notable exception being the aging hen, where the frequency is ~24% (Fredrickson, 1987). Differences in the reproductive tract anatomy and disease development between chickens and humans limit the use of the aging hen as an animal model for EOC. Many in vivo studies utilized s.c./i.p. (Freedman et al., 1978; Hamilton et al., 1984; Hamilton et al., 1983; Massazza et al., 1989; Ward et al., 1987) as well as orthotopically implanted (Bao et al., 2002; Fu & Hoffman, 1993; Kiguchi et al., 1998) xenografts of human ovarian carcinoma cell lines. Although s.c. tumor formation can be readily quantified with calipers, i.p. or orthotopic implantation of ovarian carcinoma cells provides a more relevant tumor microenvironment. Thus, tumor ‘take’ is less efficient with s.c. injected mice (Kiguchi et al., 1998). Models of ex vivo transformation of rodent, rabbit and human ovarian surface epithelium (OSE) have been described (Adams & Auersperg, 1981; Auersperg et al., 1995; Coppola et al., 1999; Godwin et al., 1992; Liu et al., 2004; Resnicoff et al., 1993; Roby et al., 2000; Testa et al., 1994). * Spontaneously transformed rodent, rabbit, or human OSE-derived cell lines can be grown as xenografts in immunocompromised Nod/SCID or athymic nude recipient mice (Adams & Auersperg, 1981; Coppola et al., 1999; Davies et al., 1998; Godwin et al., 1992; Liu et al., 2004; Roby et al., 2000; Testa et al., 1994) or as tumor allografts in syngeneic rats (Rose et al., 1996) or mice (Roby et al., 2000).
Despite available technology for making genetically engineered mouse (GEM) models, their use in studying of EOC has been limited by the lack of understanding of the epithelial precursor cell, as well as the critical molecular alterations that contribute to disease initiation and progression (see Background Information). A transgenic model of spontaneous EOC was created by expressing the early region of the Simian Virus 40 genome, including the small and large T antigen (TAg) genes, under transcriptional control of the Müllerian inhibiting substance type II receptor gene promoter (Connolly, 2003; Hensley et al., 2007). Female TgMISIIR-TAg-DR26 transgenic mice develop bilateral ovarian tumors with 100% penetrance by five to six months of age (Hensley et al., 2007) and The tumor histology resembles serous carcinomas, the most frequently diagnosed histologic subtype of clinical EOC. While TgMISIIR-TAg-DR26 mice (Hensley et al., 2007) and other conditional GEM models (Clark-Knowles et al., 2007; Dinulescu et al., 2005; Flesken-Nikitin et al., 2003; Wu et al., 2007) hold promise as preclinical models of human EOC, the time it takes for ovarian tumors to develop is variable and the use of tumor formation as an endpoint precludes the analysis of early lesions. In the absence of reliable methods for disease detection and evaluation of therapeutic response, experiments are limited to long-term studies involving large animal cohorts with a predetermined, and somewhat arbitrary, time point for euthanasia and necropsy. Technologies for in vivo longitudinal imaging of GEM models of ovarian cancer significantly facilitate therapeutic studies allowing for quantitative data acquisition, assessment of response (e.g., stable disease or regression) and minimization of the number of experimental animals required for each study.
Tumor formation resulting from peritoneal or orthotopic implantation of ovarian carcinoma cells, or arising spontaneously in the ovary in GEM models, occurs relatively deeply within the peritoneal cavity making quantification of tumor burden difficult if not impossible. In vivo imaging technologies such as Magnetic Resonance Imaging (MRI), ultrasound, micro-positron emission tomography (micro-PET) and computed tomography (CT) may be used to study mice with spontaneous autochthonous ovarian tumor development (e.g., transgenic, conditional or inducible mouse models) in the absence of fluorescent or luminescent reporters. In particular, MRI is an excellent modality for non-invasive tumor detection, acquisition of high resolution images of anatomical structures and for accurate, serial measurement of ovarian tumors in mice over time (Hensley et al., 2007; Mabuchi et al., 2007). Bioluminescent imaging (BLI) or fluorescent imaging (FLI) may be employed for longitudinal in vivo imaging of tumor burden in GEM mice in which tumor formation is accompanied by expression of a light-emitting reporter such as firefly luciferase or green fluorescent protein (GFP). To use BLI or FLI for human or murine ovarian carcinoma cell lines to be used as tumor xenografts or allografts, cell lines can be engineered to express luciferase or GFP by standard methods of transfection or retroviral transduction. In general, BLI sensitivity is superior to FLI and BLI does not require the subtraction of the autofluorescence background signal from the animal (Choy et al., 2003).
Detailed in this unit are strategies for the uses of ovarian carcinoma cells engrafted in recipient mice (Basic Protocol 1) and a transgenic model of spontaneous ovarian cancer. Both of these models can be used for pre-clinical evaluation of therapeutic agents for treating this condition. Methods for longitudinal in vivo MRI and BLI are also described in detail to facilitate pre-clinical studies and the acquisition of quantitative data on tumor burden over time. Described in Basic Protocol 2 is the use of MRI to detect and quantify the size of spontaneous ovarian tumors arising in a GEM model of EOC (e.g., TgMISIIR-TAg-DR26 mice). The use of BLI to monitor tumor growth and progression in mice allografted with ovarian cancer cells transfected with firefly luciferase is detailed in Basic Protocol 3.
NOTE: Prior to commencing with any experiments involving the use of vertebrate animals, investigators are required to be trained in the proper use and care of experimental animals and to obtain Institutional Animal Care and Use Committee (IACUC) approval for a detailed protocol describing all experiments. All IACUC protocols must conform to governmental regulations regarding the humane use and care of laboratory animals.
Xenograft or allograft models have several advantages over other techniques. For example, ovarian carcinoma cells implanted s.c. allow for rapid, relatively high-throughput quantitative tumor formation studies to test the response to therapeutic agents. In this case tumor volume is measured with a caliper. To study tumor growth in a more relevant microenvironment, ovarian carcinoma cells may also be implanted i.p. or orthotopically by intrabursal injection. Like human EOC, most ovarian cancer cells grown following i.p. or orthotopic administration result in widespread peritoneal disease with tumor implants on serosal surfaces and the production of ascites., Metastases to distant sites, such as the lung or brain are uncommon. Although i.p. or orthotopic implantation of ovarian carcinoma cells may provide a more relevant tumor microenvironment, these approaches make quantitative assessments difficult. However, the ability to manipulate cells in vitro to introduce luminescent and/or fluorescent reporter genes overcomes this limitation by making it possible to use BLI or FLI (BLI is discussed in detail in Basic Protocol 3) to obtain quantitative data. Another advantage of xenograft/allograft EOC models is that target genes can be manipulated directlyin tumor cells in vitro, by constitutive or inducible gene expression or RNA interference, to study the in vivo effects of altered gene expression on EOC etiology or progression. Also, xenograft/allograft EOC models in which equal numbers of cells are implanted and with generally predictable disease latency allows for direct comparisons of the effects of gene expression alterations and/or therapeutic treatments in mice. In the protocol below, step by step methods are described for the preparation of ovarian carcinoma cells and for subcutaneous, intraperitoneal and orthotopic (intrabursal) injections..
Ovarian cancer cell line - There are a large number of human and murine ovarian carcinoma cell lines available from commercial and academic sources, with human cell lines available through ATCC, and human and murine cell lines from individual investigators under a Material Transfer Agreement. The choice of cell line most appropriate for a particular study depends on numerous factors including, but not limited to, the histology of the primary tumor from which the cells were isolated, the ability of cells to produce tumors in immunocompromised mice and relative drug sensitivity and/or resistance. There is an extensive literature regarding the use of human ovarian carcinoma cell lines that can be used to guide the design of individual experiments (Connolly, 2003; Dinulescu et al., 2005; DiSaia et al., 1975; Freedman et al., 1978; Hamilton et al., 1984; Hamilton et al., 1983; Ioachim et al., 1975; Kolfschoten et al., 2000; Mabuchi et al., 2007; Molpus et al., 1996; Ozols et al., 1987; Pieretti-Vanmarcke et al., 2006a; Pieretti-Vanmarcke et al., 2006b; Roby et al., 2000; Szotek et al., 2006; Ward et al., 1987; Woods et al., 1979; Xing & Orsulic, 2005a; Xing & Orsulic, 2005b; Xing & Orsulic, 2006; Zhang et al., 2002))
Cell Culture reagents - unless otherwise specified, standard cell culture reagents can be purchased from commercial sources such as Gibco/Invitrogen and Cellgro.
Cell culture medium – examples include DMEM and RPMI
Fetal calf serum (FCS)
100× Insulin/Transferrin/Selenium (Invitrogen or Cellgro)
Phosphate buffered saline (PBS), sterile
Ca++ and Mg++ free PBS (optional)
15 ml conical tubes (Fisher Scientific)
0.4% Trypan Blue
Hemacytometer (Fisher Scientific)
37°C water bath (Fisher Scientific)
37°C/5% CO2 incubator (Fisher Scientific)
Tissue culture flasks (75 or 175 cm2) (Fisher Scientific)
Immunocompromised (e.g., Nod/SCID) or syngeneic (e.g., C57Bl/6) recipient female mice, 8–10 weeks of age - available from standard commercial sources, including Taconic, Charles River, and Jackson Labs
Sterile 1 cc tuberculin syringes (Fisher Scientific)
26 G needles (Fisher Scientific)
29 ½ or 30 G needles (Fisher Scientific)
Electric clippers (Fisher Scientific or Roboz Surgical Instrument Co., Inc., Gaithersburg, MD)
Alcohol prep pads (Fisher Scientific)
Betadine (Fisher Scientific)
Sterile surgical instruments- includes scissors, forceps, tissue clips, and surgical wound staples (Roboz Surgical Instrument Co., Inc., Gaithersburg, MD)
6-0 Vicryl sutures (Roboz Surgical Instrument Co., Inc., Gaithersburg, MD)
Calipers for tumor measurement (Fisher Scientific)
Surgical tissue adhesive (Nexaband®, Abbot Laboratories, Chicago, IL)
Glass bead sterilizer (optional, Fisher Scientific)
Sterile physiologic saline
10 mg/mI ketamine hydrochloride and 1 mg/mI xylazine hydrochloride in sterile saline (Fisher Scientific)
Buprenorphine (Fisher Scientific)
Heating pad or infrared heat lamp (Fisher Scientific)
Liquid Infant Heel Warmer (optional, Fisher Scientific Cat. No. 22-024-646)
Scale (Fisher Scientific)
MRI is particularly well suited for imaging GEM models. Both normal mouse ovaries and primary tumors arising in the ovary are generally well circumscribed and can be identified easily using MRI scanners with a field strength of 7 Tesla or greater (Hensley et al., 2007; Mabuchi et al., 2007). Tumor formation arising from ovarian carcinoma cell lines implanted orthotopically by intrabursal injection may also be visualized and monitored using this instrumentation.. MRI is well suited for monitoring spontaneous ovarian tumor growth in vivo because of its superb soft tissue contrast and high spatial resolution
MRI is not as well suited to imaging diffuse peritoneal disease associated with ovarian metastasis or with tumor formation resulting from i.p. injection. For this purpose BLI (as described in Basic Protocol 3 below) should be used.
GEM models of EOC or mice with orthotopically implanted ovarian carcinoma cell lines (see Basic Protocol 1)
1× sterile filtered PBS
Gadolinium-diethylenepentaacetic acid (Gd-DTPA, Magnevist, Berlex Labs, Hamilton, NJ URL)
Access to a vertical or horizontal bore Magnetic Resonance Imaging scanner, field strength 7 Tesla or greater, with 25–30mm birdcage coil for imaging
Plastic tubing (2–3 mm inner diameter) for phantom to mark orientation of mouse
Isoflurane - oxygen based anesthesia system
Induction chamber for anesthetizing mice (Summit Medical Equipment, Molecular Imaging products, http://store.mipcompany.com/)
Electric clippers or lotion depilatory
1ml tuberculin syringe
30 G needle
Infrared heat lamp
Temperature regulation equipment for animal (if available)
A large variety of methods and software are available for image and volumetric analyses including software packages freely available on the internet. This allows for off-line analysis of images, and makes redundant measurements convenient so that measurements from different observers can be compared. The following instructions are specific to the analysis of image sets obtained on a Bruker MR imaging console, which are stored in the Paravision format. Images are converted to “analyze” format and volumetric analyses performed using the free shareware programs Bru2Analyzer and MRIcro (Rorden & Brett, 2000). We apply the standard planimetric technique of manually outlining regions of interest (ROIs) in contiguous slices and the volume is computed by adding the total number of pixels in the ROIs together and multiplying the sum by the pixel size and the slice thickness (Hensley et al., 2007).
MR datasets acquired from scanner
Shareware programs: Bru2analyzer and MRIcro (available online, see below)
To obtain quantitative tumor growth data in mice receiving i.p. or orthotopic injections of ovarian carcinoma cells, cells can be transfected or transduced with a firefly luciferase or fluorescent reporter gene and monitored by BLI or FLI. While both methods are suitable for obtaining quantitative data, the use of BLI may be somewhat more advantageous in that it generally provides superior sensitivity relative to FLI without complicating issues of autofluorescence (Choy et al., 2003). Described below are methods for verification of expression of luciferase in ovarian carcinoma cell lines and for BLI of in vivo tumor growth in mice.
Ovarian carcinoma cell lines expressing a luminescent reporter gene, such as firefly luciferase. Cell lines used include the human ovarian carcinoma cell line, SKOV3-luc-D3 Bioware®cell line, available from Caliperls (Hopkinton, MA). Alternatively, any ovarian cancer cell line of choice (as described in Basic Protocol 1) can be modified by standard transfection or viral transduction protocols to express the firefly luciferase reporter gene.
96-well cell culture dish
Ovarian tumor-bearing mice:
In vivo bioluminescent imaging system such as the IVIS Spectrum (Caliper LifeSciences)
1× sterile filtered PBS
Luciferin substrate (Caliper LifeSciences)
Isoflurane - oxygen based anesthesia system
Induction chamber for anesthetizing mice (Summit Medical Equipment, Molecular Imaging products, http://store.mipcompany.com/)
1ml tuberculin syringes
30 G needle
Infrared heat lamp
Perform image analysis using the software that accompanies the imaging system (e.g., Living Image Software for IVIS Spectrum).
To assure that complete assessments are made in the evaluation of potential therapeutics, it is recommended that a complete necropsy report be generated for each animal. In addition to quantitative assessments of gross tumor burden, collection and fixation of tissues and subsequent histopathologic evaluation may provide important information regarding the morphological and molecular characteristics of the tumors, particularly in response to test agents. Formalin fixed paraffin embedded (FFPE) tissue sections may subsequently be stained with various markers of proliferation, apoptosis, angiogenesis and tumor invasion. Additionally FFPE tissue sections may be stained for the expression of relevant signaling proteins that may be affected by chemotherapeutic agents. In some cases, such as RNA-based applications or immunohistochemical detection for some proteins, it may be desirable or necessary to obtain frozen tissue specimens. Therefore, it is recommended that frozen tissues be saved in addition to FFPE tissues.
CO2 for euthanasia
10% Neutral buffered formalin (NBF) (Thermo Scientific, Fisher)Sterile dissection scissors and forceps
Sterile transfer pipets
15- or 50-mL conical tubes
Necropsy report form
In the United States in 2008 it is estimated that approximately 22,000 new cases of EOC will be diagnosed and 15,520 deaths will occur (Jemal et al., 2008). Although the overall incidence of EOC is relatively low, it is the eighth most commonly diagnosed cancer and fifth most common cause of cancer death among American women. Globally, approximately 204,500 new cases are diagnosed and 125,000 deaths occur annually, making it the sixth most common cancer and the seventh most common cause of cancer death among women worldwide (Parkin et al., 2005). These statistics have changed relatively little over the past 30 years and, as there are currently no reliable means of early detection or prevention of EOC, and they are not predicted to improve in the near term.
Although ovarian tumors can arise from stromal tissue and germ cells, those arising in the epithelium account for 85–90% of all ovarian cancers. The initiating cell population for EOC remains somewhat controversial, with evidence suggesting it originates from either the ovarian surface epithelium (OSE), inclusion cysts lined by OSE (Auersperg et al., 1997; Scully, 1995) or alternatively, components of the secondary Müllerian system, including the epithelial cells of the rete ovarii, paraovarian/paratubal cysts, endosalpingiosis, endometriosis or endomucinosis (Crum et al., 2007; Dubeau, 1999). One reason for the lack of certainty regarding tumor origin is that, unlike epithelial cancers arising in other organs, a well-defined disease spectrum consisting of benign, invasive and metastatic lesions has not been identified for EOC. This is largely because the majority of cases are identified at an advanced stage. In addition, common epithelial tumors consist of several distinct histologic subtypes, including serous, endometrioid, mucinous, clear cell, Brenner and mixed histology tumors (Scully, 1995). EOCs are further categorized as benign, borderline (also referred to as low malignant potential tumors) or malignant. The relationship of borderline tumors to invasive cancers is complex. Molecular evidence suggests that for some histologic subtypes, such as mucinous, endometrioid, and clear cell, borderline tumors are likely related to invasive cancers while others, such as serous, are not (Shih & Kurman, 2005).
Early attempts to use transgenic or other genetic engineering approaches to produce murine EOC models largely resulted in strains that develop sex-cord stromal tumors (Dutertre et al., 2001; Kananen et al., 1995; Keri et al., 2000; Kumar et al., 1999; Rahman & Huhtaniemi, 2001; Risma et al., 1995). Later, a number of laboratories made GEM models of EOC by using ex vivo transformation (Orsulic et al., 2002), transgenic (Connolly, 2003; Hensley et al., 2007) and conditional gene expression strategies (Dinulescu et al., 2005; Flesken-Nikitin et al., 2003; Wu et al., 2007). In 2002, Orsulic et al. described ex vivo transformation of p53 null murine OSE in cultured ovarian explants by retroviral transduction with at least two additional oncogenes (e.g., K-Ras, Myc or Akt). Several laboratories used conditional Cre-LoxP-mediated strategies to establish GEM models of EOC by intrabursal administration of Adenovirus-Cre recombinase. Mice with conditional inactivation of both Rb and p53 in the OSE develop serous ovarian carcinomas (Flesken-Nikitin et al., 2003). Mice with conditional inactivation of Pten and conditional activation of mutant K-ras (Dinulescu et al., 2005) or conditional inactivation of both Pten and Apc (Wu et al., 2007) develop EOCs of the endometrioid subtype. As there is currently no existing mouse model that exhibits OSE-restricted expression of Cre-recombinase, tumor induction that relies on Cre-mediated excision of LoxP-flanked sequences requires intrabursal administration of Adenovirus-Cre (Clark-Knowles et al., 2007; Dinulescu et al., 2005; Flesken-Nikitin et al., 2003; Wu et al., 2007).
Orthotopic implantation of ovarian carcinoma cell lines by intrabursal injection requires aseptic micro-surgical technique to avoid infection and post-operative complications. Successful implantation of cells is highly dependent on careful and precise handling of the reproductive tract and injection through the infundibulum into the intrabursal space to ensure the bursa does not tear and thereby allow cells to leak into the bursal space. Bursal tears and/or leaking cells from the bursa will effectively result in an i.p. injection.
As for any procedure that involves survival surgery, administration of anesthesia and post-operative monitoring are essential for the well-being of the animals and a successful outcome.
The procedures outlined in this unit involve either spontaneous tumor development or tumor induction by implantation of ovarian carcinoma cell lines. In either case, mice must be monitored regularly to ensure their well being and to assess the need for euthanasia. Mice should be euthanized as soon as they exhibit any of the following symptoms: 1) loss of > 10% of body weight, 2) general signs of ill health (e.g., lethargy, fur scruffing), 3) ataxia, 4) significant ascites accompanied by weight gain of 2 g over several days, 5) anorexia, 6) dehydration, 7) vocalizations indicating pain or distress, 8) severe anemia indicated by pallor of skin on the feet or around the eyes and/or 9) open, bleeding, or infected sores or wounds.
Listed on Table 14.1 are potential problems that may be encountered in generating xenografts or allografts. Tables 14.2 and 14.3 detail potential difficulties with MRI and BLI. Possible causes and recommended actions for overcoming and/or avoiding procedural problems are indicated in each Table.
Representative images of MRI scans in the oblique sagittal orientation of two wild type (top panels) and two TgMISIIR-TAg transgenic (bottom panels) female mice are shown (Figure 14.3). A single section is depicted unmasked (left) and masked (right) for each animal. The masked images highlight the ROI; that is the normal ovary in wild type mice (top right) or ovarian tumors in the TgMISIIR-TAg transgenic mice (bottom panels) for each animal. Note the presence of the phantom marking the left side of the mouse (right side of the image) in the top panels. Our laboratory has used this method to monitor tumor burden over time in previously described chemotherapeutic studies (Hensley et al., 2007; Mabuchi et al., 2007).
Comprehensive analyses of tumor growth rates and responsiveness to a standard cytotoxic chemotherapy regimen, such as cisplatin and paclitaxel, and a molecularly targeted agent, such as the mTOR inhibitor RAD001 (Everolimus) have been performed (Hensley et al., 2007; Mabuchi et al., 2007). While not curative, TgMISIIR-TAg transgenic mice treated with cisplatin and paclitaxel exhibited a significant delay in tumor progression (Hensley et al., 2007). RAD001 treatment resulted in markedly delayed tumor development, ascites production peritoneal dissemination of ovarian tumors in these mice (Hensley et al., 2007; Mabuchi et al., 2007). Collectively, these data allowed the use of statistical modeling to aid in the design of preclinical studies (Hensley et al., 2007; Mabuchi et al., 2007). Simulations from these analyses suggested that therapeutic studies utilizing TgMISIIR-TAg-DR26 transgenic mice in conjunction with MRI can be conducted with as few as 20 animals in each treatment or control group and that groups that differ by +25% in time to acquire a given tumor volume (e.g., 100 mm3), the acquisition time can be distinguished with 80% power and 5% type I error (Hensley et al., 2007).
For a cell line to be useful for in vivo studies, it must produce a sufficiently bright luminescent signal. Prior to initiating in vivo experiments, it is highly recommended that the luminescent signal be verified in vitro. As an example, serially dilute (1:2) MOVCAR 5009 cells transduced with a retroviral luciferase construct (MOVCAR-5009-Luc) and plate them in triplicate wells ranging from 20,000 to 78 cells/well. After plating, 30 μl of a 1 mg/ml stock solution of luciferin diluted in PBS is added to each well and the plate imaged using the IVIS Spectrum (Caliper LifeSciences). The results of this experiment will demonstrate the luminescent signal can be reliably detected with as few as 312 cells (Figure 14.4A), and verifies these cells are suitable for subsequent in vivo experiments. The IVIS Spectrum allows absolute calibration of signal strength (in photons/second) from a given region of interest, and the brightness of a cell line can be calculated by dividing the calibrated signal from a given well by the number of cells in the well (Figure 14.4B). A signal strength of 5 photons/(sec × cell) is sufficiently bright to be used in mouse models. This corresponds to being able to detect 300 cells in a well with a signal-to-noise ratio of approximately 2 for an exposure time of 2 min.
The MOVCAR-5009 cell line expresses and secretes high levels of VEGF protein into the culture medium and, after i.p. injection, rapidly produces bloody malignant ascites in mice (Connolly, unpublished observations). In a pilot study to test the effects of RNA interference of Vegf expression on tumor formation, a previously described (Dickins et al., 2005) retroviral shRNA expression vector MSCV-LTR-miR-30-PIG (MLP) was used to express Vegf-specific shRNA constructs. Several individual Vegf-specific shRNA constructs were PCR cloned and tested. Levels of Vegf RNA and protein expression in MOVCAR-5009 cells co-transduced with pWZL-Luciferase and retroviral shRNA constructs for Vegf or the MLP vector alone were determined by quantitative RT-PCR and ELISA. Using this method, approximately 80% stable inhibition of Vegf mRNA and protein was observed with several Vegf shRNA constructs (data not shown). In a pilot experiment to test the effects of RNAi-mediated knockdown of Vegf in MOVCAR-5009-Luc cells on tumor growth in vivo, MOVCAR-5009-Luc cells co-transduced with Vegf shRNA 593 were compared to the MLP vector. Mice (n=5 mice/group) were injected i.p. with 5 × 106 MOVCAR-5009-Luc + MLP or MOVCAR-5009-Luc + Vegf Sh593 cells (as described in Basic Protocol 1) and subjected to weekly BLI (as described in Basic Protocol 3) to monitor tumor burden over time. Representative images (Day 7 and Day 21) demonstrated a significant increase in signal intensity over two weeks (Figure 14.5A). The ROIs (indicated by the ovals in Figure 14.5A) were defined as the area of the mouse that emitted luminescent signal and was confined to the abdomen of i.p.-injected mice. Total tumor burden was estimated from the integrated photon flux in the ROI (Figure 14.5B). All image analyses were performed using Living Image software provided by the manufacturer. Mice allografted with MOVCAR-5009-Luc + MLP cells were euthanized 21 days post injection while mice allografted with MOVCAR-5009-Luc + Vegf Sh593 cells did not require euthanasia until 29 days post injection. Therefore, expression of the Vegf shRNA in MOVCAR-5009-Luc cells significantly delayed ovarian tumor progression and the accumulation of ascites as compared to the vector control (Figure 14.5C).
In conjunction with the Fox Chase Cancer Center Biostatistics Core, it was determined that when using BLI to determine growth rates, sample sizes of 20 animals/test condition (e.g., therapeutic agent, short hairpin RNA or vehicle) should be sufficient to obtain statistically significant results using the following approach. Regression analysis of log transformed tumor area versus time yields a single growth rate for each mouse. The resulting growth rates (n=40) can be compared using the Wilcoxon two-sample test. Since power cannot be determined in advance for this procedure, the Fisher’s exact test should be used. Thus, the 40 growth rates can be sorted and the resulting data submitted to this test, which is predicted to distinguish odds ratios of 5.3 and 1.0, with 80% power and 4.79% type I error.
The amount of time necessary to perform the implantation of ovarian carcinoma cell lines is dependent both on the injection method chosen and the experience of the investigator. For s.c. injections, shaving the flank and administering the injection takes 2–3 min per animal. For i.p. injections, a reasonably experienced investigator should be able to inject a cage of 5 mice in less than 5–10 min. For either s.c. or i.p injections, a large number of mice (n= 40–60) can be injected in a few hours’ time. Intrabursal injection is more complicated,, requiring survival surgery and great deal more technical skill to execute the injections correctly. A highly experienced individual, such as a person who has made transgenic animals, would likely be able to complete the surgery and injection of one mouse in 20–30 min. It is essential for this procedure that another individual assist with preparation of animals for surgery, administration of anesthesia, application of wound clips and post-surgery monitoring. With assistance, intrabursal injections can be performed on 15–20 mice in one day.
The time required for tumor formation is largely cell line-dependent. For well studied cell lines, information regarding the recommended number of cells for injection and resulting latency of disease formation can be obtained from the literature or individual investigators that routinely utilize the cells. In the absence of such information, it is recommended that these parameters be determined empirically in pilot studies. For example, latency of disease formation after i.p. injection of 5 × 106 MOVCAR cells can vary from 4–16 weeks depending on the individual MOVCAR cell line used.
For studies aimed at assessing the effectiveness of drug candidates or other therapeutic modalities, frequency and routes of administration, and doses should be determined in advance. All experiments should be planned in consultation with a statistician to ensure that the minimum number of experimental and control animals are used while still allowing acquisition of statistically significant data. The use of in vivo imaging for longitudinal quantification of tumor burden is advantageous and should significantly reduce the number of animals needed. The time required to execute a given study largely depends on the frequency, duration and route of administration of the potential drug.
Denise C. Connolly, Fox Chase Cancer Center, Philadelphia, PA, Phone: 215-728-1004, Fax: 215-728-2741, Email: ude.cccf@yllonnoC.esineD.
Harvey H. Hensley, Fox Chase Cancer Center, Philadelphia, PA, Phone: 215-728-3156, Fax: 215-728-3574, Email: ude.cccf@yelsneH.yevraH.