Development of A Novel Murine Model of Lymphatic Metastasis

Abstract

Current laboratory models of lymphatic metastasis generally require either genetically modified animals or are technically challenging. Herein, we have developed a robust protocol for the induction of intralymphatic metastasis in wild-type mice with reproducible outcomes. To determine an optimal injection quantity and timeline for tumorigenesis, C57Bl/6 mice were injected directly into the mesenteric lymph duct (MLD) with varying numbers of syngeneic murine colon cancer cells (MC38) or gastric cancer cells (YTN16) expressing GFP/luciferase and monitored over 2–4 weeks. Tumor growth was tracked via whole-animal in vivo bioluminescence imaging (IVIS). Our data indicate that the injection of tumor cells into the MLD is a viable model for lymphatic metastasis as necropsies revealed large tumor burdens and metastasis in regional lymph nodes. This protocol enables a closer study of the role of lymphatics in cancer metastasis and opens a window for the development of novel approaches for treatment of metastatic diseases.

Introduction

Distant metastasis is a complex process in which cancer cells leave the original tumor site, spread out in the body, and ultimately colonize other organs such as the liver, kidney, brain, and lung. It is believed that the metastatic cascade starts with local invasion of surrounding stroma, followed by intravasation, whereby the cancer cells invade the blood and lymphatic vessels and gain access to the systemic and lymphatic circulations. Extravasation of the tumor cells at secondary sites and outgrowth at those sites constitute the final steps [1]. An important question is whether lymph node metastasis is part of the cascade.

It is certainly well-established that lymph node metastasis is an ominous sign of cancer progression and a poor prognostic factor in the majority of cancers [2]. This is especially true in melanoma, breast, and gastric cancers, which have a propensity to spread to the lymphatic system and for which sentinel lymph node mapping and biopsy with possible resection of regional lymph nodes are part of the therapeutic protocol [3–5]. However, a subject of some debate is whether tumor cells traffic through lymph nodes en route to establishing distant organ metastases. Two recent studies in mouse models of various cancers suggested that tumor cells within lymph nodes have access to blood vessels within the nodes themselves, rather than via drainage into the thoracic duct, as a source of distant metastases [6,7]. In the case of human colorectal cancer, however, the role of lymph node metastasis is more controversial. Development of colorectal metastases is largely attributed to circulatory dispersal, and recent genomic evidence suggests that the majority of lymphatic metastases are seeded by tumor sub-clones distinct from those that seed liver metastases [8]. While some have proposed that the presence of lymph node metastases is merely evidence of general metastatic competence of a primary tumor [9] or even an indicator of survival [10], positive outcomes of lymphadenectomy in certain malignancies (renal, endometrial, colorectal) suggest that lymphatic involvement actively contributes to cancer progression although possibly by means other than the dispersal of systemic metastases. Animal models that can specifically address the contribution of lymphatic metastasis to the establishment of distant metastasis will be important to truly understand metastatic progression. Thus the goal of the research described here was to establish a robust model that can be used to address some of the questions regarding the specific role of the lymphatic system in metastasis.

Currently, the most widely used models for studying lymphatic metastasis involve the use of transgenic mice over-expressing lymphangiogenic factors [11–14] and/or primary tumors with lymphatic invasion [15,16]. In these cases, the impact of lymphatic metastases is confounded by the presence of a primary tumor that may seed additional metastases, and/or the presence of a transgene that may affect immune system development, resulting in an inaccurate systemic model. Here, we describe a novel approach to study the role of lymphatic vessels in cancer spread by means of an intralymphatic microinfusion technique in a murine model. While direct injection models have been previously described, they have either used direct injection into surrounding tissues of lymph nodes [17,18] or required specialized injector devices [19,7]. Our technique allows the implantation of tumor cells directly into the afferent lymphatic vessels of mice with a high success rate even with little practice. This technique uses a needled catheter to inject cells of interest into the mesenteric lymph duct (MLD) with minimal trauma to the vessel. The injection process is visible under a microscope, and lymph leakage is prevented by the application of surgical glue at the injection site. Notably, we chose to develop methodology for MLD injection as our interest is in the process of the colon and gastric cancer metastasis. Since lymphatics drain directly from the local tissue, the molecular constituents (proteins, cellular metabolites, etc) reflect the tissue environment [20], which may impact tumor cell behavior.

Applications of the method

Historically, the lymphatic network has been considered as a conduit for transferring molecules and nutrients to the systemic circulation. But our knowledge of the role of lymphatics in the development and progression of various diseases has been rather limited. This deficiency can be attributed to inherent difficulties associated with studying the lymphatic system, especially in pre-clinical models. In contrast to blood vessels, the lymphatic vessels are thin, fragile, and often difficult to access in mice. Large lymphatic vessels such as the MLD are located deep in the posterior peritoneum, and the more accessible dermal lymphatic vessels are too small for any cannulation procedures.

The idea of direct utilization of the lymphatic vessels for research purposes or therapeutic modalities has been considered for years, although the majority of such techniques have focused on injection into lymph nodes.While such methods may be useful for Intralymphatic chemotherapy or immunotherapy [21–23], they are problematic for studying metastatic mechanisms as they can potentially induce damage to lymph nodes that may impact signaling and immune activation. Lymphatic research is still developing, and due to the challenges noted above, there are relatively few research groups focused on studying lymphatic biology in pre-clinical models. Hence, our novel technique offers an additional avenue for exploration of the role of the lymphatic network in the development and progression of cancer metastasis.

MATERIAL AND METHODS

Cell lines

The cell lines used in this study were murine lines syngeneic with C57Bl/6 mice: the colon cancer cell line MC38 [24] and the recently described murine gastric cancer line YTN16 [25]. Cells were grown in vitro in Dulbecco’s Modified Eagle’s medium (Corning Cellgro by Mediatech, Inc.) supplemented with 5–10%(v/v) fetal bovine serum (Atlanta Biologicals) and 50 μg/ml gentamicin (Corning Cellgro by Mediatech, Inc.). YTN16 cells were cultured on plates that had been pre-coated with type I collagen, as described [28]. The medium used for culturing YTN16 cells was also supplemented with 0.1% Corning^™ MITO+ Serum Extender (Corning). All cells were maintained in a 5% CO2-humidified incubator at 37℃ and were regularly checked to ensure that they were mycoplasma negative (PCR Mycoplasma Detection Kit, Applied Stem Cell). Cells were modified to express a combination of GFP and luciferase using the appropriate lentiviral particles purchased from GeneCopoeia and following the manufacturer’s instructions. Modified cell populations were selected and maintained in medium supplemented with puromycin (Corning Cellgro by Mediatech) at 10 μg/ml.

Animals

Male and female C57Bl/6J mice were initially purchased from The Jackson Laboratory and used to establish an in-house breeding colony. They were fed a diet containing 11% fat (PicoLab® High Energy Mouse Diet - 5LJ5) for a minimum of 2 weeks before the lymphatic injections. Mice of both sexes and aged 6–10 weeks were used. All experimental animal protocols were conducted only after review and approval by the local institutional animal care and use committee. The detailed surgical procedure is described in a stepwise fashion in the results section.

In vivo analysis

For 2–4 weeks following surgery (depending on the cell line), tumor growth was tracked via whole-animal in vivo bioluminescence imaging (PerkinElmar IVIS Spectrum bioluminescent imager). Twice weekly, mice were injected with 150 mg/kg luciferin (Gold Biotechnology Inc) via intraperitoneal (IP) injection 10 minutes before imaging for varying periods up to 3 minutes exposure, all under isoflurane anesthesia. Mice were sacrificed after 2–4 weeks.

Living Image software (Caliper Life Sciences) was used to quantify the luminescence intensity in a defined region-of-interest over time (Supp Fig. 1A). The luminescent scale of each reading was adjusted to encompass a range suitable for all time points, and the region-of-interest defined by the area of luminescent signal from the final timepoint image was applied to each image in a series (Supp Fig. 1B). Each reading was balanced by subtracting the background luminescence from that instance.

Tissue processing and histology

At the time of sacrifice, any visible tumors in the abdomen were resected and placed in 10% buffered formalin (Fisher) for overnight fixation. Lungs were examined in situ and then inflated using 1.5mls of 10% buffered formalin introduced via the trachea. The inflated lungs were removed and fixed overnight in 10mls of formalin. Following fixation, tissues were transferred to 50% ethanol before processing through graded ethanols, xylenes and paraffin wax before embedding. Five-micron sections were cut and stained with hematoxylin and eosin using standard procedures. Images of the H&E-stained slides were captured using a Zeiss microscope fitted with a camera. The images were processed to white-balance and increase contrast to make the distinction between different cell types clearer.

Results

The central purpose of the studies described here was to develop a robust and reproducible surgical method for introducing gastrointestinal tumor cells directly into relevant lymphatic vessels. Following is the optimized step-by-step method:

Preparation

All procedures must be performed under aseptic conditions with sterilized surgical instruments.

• 1Needled catheters should be prepared and gas sterilized pre-procedurally.
• 2To prepare these, hold a 30G needle with a needle holder (approximately 1cm distal to the tip) under a micro-surgical or dissecting microscope and file using a micro-file to create a brittle spot on the needle. Then, cut the needle on the fragile spot and insert the cut end into a 10cm poly-vinyl chloride (PVC) tube (0.2 mm ID, 0.5 mm OD) Micro-RenathaneⓇ Tubing MRE-025 (Braintree Scientific Inc.) (Fig. 1).
• 3Weigh the animals and remove the abdominal fur pre-operatively.
• 4Trypsinize the cells and wash three times with Dulbecco’s phosphate-buffered saline (DPBS calcium and magnesium-free; Corning Cellgro by Mediatech Inc) to ensure a single cell suspension. Count cells and resuspend in DPBS at 1–10 × 10⁷ /ml.
• 5Load a 1 ml syringe with prepared cells and attach a needled catheter. Prime the needled catheter, ensuring there are no air bubbles. Keep the syringe and tubing on the ice before the injections.
• 6For analgesia, pre-medicate the mice with subcutaneous injection of 0.05 mg/Kg bodyweight buprenorphine (Patterson Veterinary) and continue every 12h administration for 48hrs post-surgically.
• 7Anesthetize the animal using 2–3% isoflurane in an induction chamber. Check for complete anesthesia by pinching the lower limb or tail.
• 8Once proper anesthesia is achieved, place the animal on a heating pad in the dorsal recumbence position and continue to administer 2% isoflurane via a nose cone. Apply an ophthalmic gel to eyes to prevent corneal dryness.
• 9Cleanse and disinfect the incision site on the abdomen using three repeated applications of betadine and ethanol. Cover mouse with sterile surgical drape with an opening to the surgical site.

Figure 1. Injection catheter preparation.

Under a microscope, a brittle spot is created on a 30G needle using a micro-file [A]. The needle is then cut [B], and the cut end inserted into a piece of flexible tubing (0.2 mm ID, 0.5 mm OD) [C-D].

Surgery

• 10Create a midline incision on the abdomen. Place a retractor on the abdominal wall. Place the intestine to the animal’s left side and cover with wet gauze.
• 11Locate the superior mesenteric artery (SMA) and adjacent MLDs (Fig. 2). In mice, MLDs run parallel to and on either side of the SMA.
• 12Hold and lock the needled catheter with a needle holder (Fig. 3A). Gently, insert the needle into the MLD and advance for 3–4 mm. Slowly, inject 20 μl of the cell suspension (Fig. 3B–D). After the cells have been injected, keep the needle in place for about 30 seconds to ensure the injected cells have flowed away. Gently, take out the needle and apply one drop of cyanoacrylate glue (Fig. 3E). Wait until the glue has completely dried.
• 13Gently place the intestine back into the abdominal cavity and close the abdominal wall with interrupted 6–0 absorbable polyglactin suture. Close the skin with either interrupted 6–0 nylon suture or with murine skin staples.
• 14Place the animal in the cage and allow it to recover on a suitable heating pad. Monitor the animal for proper recovery.

Figure 2. The anatomical position of the MLDs and their adjacent structures in mice.

MLDs run parallel and on either side of the SMA in mice. In the example shown, they are bright white as the mouse was given olive oil to enable easier viualization of lymphatics. IVC, inferior vena cava; LRV, left renal vein; MLD, mesenteric lymph duct; SMA, superior mesenteric artery.

Figure 3. Step-by-step demonstration of injection into the MLD.

To augment lymph flow and enhance MLD appearance, commercially-available olive oil was administered 20min pre-procedurally. Also, methylene blue was used as the injection solution for a better demonstration of the procedure. White cotton bud applicators are visible in the images as they were used to push back intestines to expand the visible field of view. [A] The needled-catheter gripped by a needle holder. [B-D] The needle is gently inserted into the MLD and the contents slowly injected. [E] The needle is slowly retracted and one drop of cyanoacrylate glue is applied onto the injection site.

Key points for successful implementation of the protocol

• Visualization of the MLD: Traditionally, feeding with olive oil has been used for visualization of MLD [26,27]. However, olive oil gavage before the surgeries may have unknown effects on the study results. We recommend feeding the mice with a fat-containing diet for the entire study duration. This will allow for proper visualization of MLD without having negative effects on the data. For beginners and training purposes, however, an olive oil (17 μL/g body weight) gavage can be done 20–30 mins before the injection to make the lymphatic vessels clearly visible [28].

• Avoid blood collection before injections: In our experience, even minimal blood collection before the experiments significantly reduces the lymphatic flow making the injections difficult. If needed, blood can be collected post-injections.

• Procedure duration: With enough practice, one should be able to perform the entire procedure in less than 15 minutes. It is very important to keep the procedure duration under 30 minutes. According to our experience, longer operative time diminishes lymph flow

For the initial studies using either the MC38 or YTN16 cell lines, cohorts of mice were injected with different numbers of luciferase-labeled cells (2 mice each with 0.5, 1, 2, or 3 million cells) in order to determine optimal cell numbers for later studies. These procedures had a minimal surgical mortality rate (less than 5%) despite four different individuals performing the procedure at various times. Cell take and subsequent growth were monitored starting one day after the surgery and every 3–4 days thereafter using bioluminescence imaging. As seen in other studies [29], the initial signal drops over the first week as a proportion of injected cells fail to engraft and are cleared (Supp Fig. 1). Thereafter signal rises as the successfully colonizing cells proliferate. From these initial experiments, an inoculum of 1–2 × 10⁶ cells was selected as 0.5 ×10⁶ did not reliably result in tumor growth within the time frame examined, there was little difference between the 1 ×10⁶ and 2 ×10⁶ amounts, and use of 3×10⁶ cells often resulted in cachexia necessitating early euthanasia.

We then expanded the number of mice used for imaging studies to gain a better appreciation of tumor growth dynamics. In MC38-injected mice, the general trend was an increase in luminescent signal over time, suggesting the continued growth of surviving tumors into the end of the observation period (Fig. 4A,B). However, this pattern was not as clear in the YTN16-injected mice, which showed varying signals throughout the incubation period (Fig. 4A). Even in the case of the MC38 cell line, the correspondence between the luciferase signal and tumor burden was not consistent. As shown in Fig. 5, even mice with decreasing or low signal (Fig. 5A) often had large tumors within lymph nodes in the abdominal cavity regardless of endpoint signal (Fig. 5B). No tumors were present on serosal surfaces suggesting no leakage from the initial injection. There was mouse-to-mouse variation in terms of how many of the mesenteric lymph nodes were involved and their exact location, which is likely dependent on the exact site of injection along the MLDs. Histological assessment of the tumors confirmed the presence of tumor cells within the lymph nodes (Fig 5C). Surprisingly, despite the similar signal range from many (but not all) YTN16 -injected mice, tumors were minimal in all cases. Together, the data suggest that the luciferase marker is an unreliable indicator of tumor growth.

Figure 4. In vivo analysis of mice post-surgically.

(A) Representative images of luminescence signal overlaid on white-light photographs of mice at different days post-injection. (B) The luminescence signal integrated over the designated region-of-interest was assessed for each animal. These values are shown graphed over time for individual mice (left panels) or averaged for the group (right panels) for either MC38 (n=5, upper panels) or YTN16 (n=8, lower panels).

Figure 5. In vivo imaging and gross appearance at study end.

(a) Luminescence readings over time from a set of three mice injected with luciferase-expressing MC38 cells. (b) Gross appearance upon necropsy of the same three mice after final imaging was completed. Arrows point to large tumors (confined within lymph nodes). ​(c) Hematoxylin and eosin-stained histological sections from two representative tumors (i and ii) harvested from the mice at 2x and 4x magnifcation, respectively. Lower panels (i’ and ii’) are the same sections at higher (10X magnification).

In approximately 30% of the analyzed cases, clearly visible lung metastases were found in mice injected with the MC38 cell line (Supp Fig. 2A). However, this does not mean to imply that no metastatic dissemination to lungs occurred.Hematoxylin and eosin staining of lungs from MLD-injected mice showed the presence of micro-metastatic lesions in multiple, though not all, samples (Supp Fig 2B,D). Micro-metastatic lesions were also present in the livers of a subset of mice (Supp Fig 2C), although these were generally found in mice with a higher number of tumor cells (2 ×10⁶) injected, and only with the MC38 colon tumor line. Together these results indicate that distant metastasis to other organs, such as lungs and liver, can result directly from lymph node metastases.

Discussion

Despite advancements in the development of new chemotherapeutic drugs and protocols that effectively target primary tumors, the treatment of metastatic cancers is still challenging. One reason for these challenges is the lack of effective pre-clinical models that represent metastatic cancers in humans. Hence, the development of novel laboratory models of metastasis is pivotal for finding new therapeutic modalities for the treatment of metastatic cancers.

Several mouse models of metastasis have been reported in the past. These models include spontaneous metastasis from subcutaneous or orthotopically implanted tumors, experimental metastasis via intravenous (tail vein) injection, or use of genetically modified mice aiming for metastatic spread of spontaneous de novo tumors [16]. Each of these models has strengths and weaknesses. As Gomez et al. described, some of the weaknesses can include the presence of mouse rather than human micro-environment, applicability to limited numbers of cell lines, poor tropism, development of asynchronous metastasis, or use of immunodeficient mice which are not truly representative of cancers in human [16]. To address some of these weaknesses, other routes of injection have been investigated as a tool for mimicking aspects of the metastatic process. These routes include intra-cardiac injections for induction of brain and bone mets, intra-splenic for liver and lymph nodes mets, and intra-peritoneal injections for development of mesentery metastasis [16]. While all these experimental models have increased our current understanding of the development of metastasis through circulating tumor cells, they do not provide much information about the role of lymphatics in metastatic cascade.

Lymphatic spread is one of the most important routes for cancer metastasis, but due to challenges with tractable models, our knowledge of how cancer cells use the lymphatic system to grow and spread to other organs has been limited. Hence, an improved understanding of the biology of lymphatic metastasis is crucial for developing novel drugs for metastasized cancers. The methodology described here enables a simple, robust model for interrogating the specific sequelae from cancer cells gaining access to the lymphatic system. The method is analogous to the intravenous experimental metastasis model bypassing primary tumor growth and escape from the primary tumor, and instead focusing on the steps that occur once tumor cells gain access to the lymphatics. Because we use injection into the MLDs, our model as described is most appropriate for use with gastrointestinal cancers. In our experience thus far, four different researchers have performed the surgical procedure described with minimal difficulty. Moreover, there was reproducible formation of tumors within lymph nodes that did not leak into the peritoneal cavity as evidenced by lack of tumor foci on any organ or membrane surface.

In contrast to the reproducible nature of the tumor formation in the model, the luciferase imaging results were quite variable. As shown in Figure 5, large tumors were present in some mice that had little to no luminescent signal. Moreover, in the case of the YTN16 cells, a similar endpoint signal was seen as with MC38 cells (Fig. 4), yet tumors were considerably smaller upon necropsy. These results highlight a problem with the use of exogenous proteins such as luciferase in immunocompetent mice. As has been documented by Baklaushev et al. [30], there is a possibility of immune clearance of luciferase-expressing cells due to its antigenic nature. In our study, it appears that tumor cells – particularly of the MC38 line - that lost or silenced luciferase expression were not cleared and grew well. The YTN16 cells may have retained the luciferase expression, but could not form large tumors. Despite these problems with the labels, the ability to use immunocompetent mice for these studies is very important. An intact immune system provides a model that is far closer to the normal human pathophysiology of lymphatic metastasis.

Direct injection of various cell types and proteins into rodent lymph nodes has been reported previously. Johansen et al. injected 10μl of a peptide solution into mouse inguinal lymph nodes using a 29 gauge (G) needle [31]. Cai et al. investigated the outcomes of intra-lymphatic chemotherapy in rats by injecting 100 μl of chemotherapy solution into rat mammary fat pads using a 27G needle through a 5mm incision [21]. Although lymph node injection of cells or molecules is feasible in rodent models, it only allows for minimal injection volumes, and more importantly, it disrupts normal nodal anatomy. Also, there is always a risk of leakage from the injection site complicating the data interpretation. An important aspect of the model described here is that the cells are carefully injected, using a small needle, into the afferent lymph duct rather than into the sinus of the lymph node, thus minimizing damage to the duct and avoiding any trauma to the affected lymph node. Other reported methods that do not damage the lymph node either use lymphangiogenesis in the primary tumor or direct lymphatic injection. In the case of the lymphatic injection procedures described by Braun et al. [19] and used by Brown et al. [7], a custom-built micro-injection device was needed. This contrasts with our use of widely available flexible tubing and standard needles and syringes. The use of primary orthotopic tumors with naturally occurring lymphatic drainage certainly has physiological relevance; however, the presence of the primary tumor can complicate interpretation of the role of the lymphatics. Our model enables isolation of the lymphatic and tumor cell interaction to facilitate investigation of its potential effects in metastatic progression.

To our knowledge, there has not been any published study utilizing the MLD for intralymphatic injection purposes. The use of a needled catheter keeps the MLD intact and is less traumatic to the vessel. By applying glue post-injection, lymph leakage is prevented and assures reestablishment of normal lymph flow post-surgically. Also, the flow of cells is visible during the injection phase making this technique a reliable method of delivering cells or molecules of interest into the lymphatic vessels. We provide representative data to demonstrate how this technique can be used for investigating the role of the lymphatic network in the development of colorectal cancer metastasis. The limitation of our model is that we are inducing metastasis without the presence of any primary tumor, which is not what happens normally. However, as mentioned previously, this method is most analogous to intravenous experimental metastasis models and allows us to bypass the first steps in the metastatic cascade and thus concentrate on the potential role of lymphatic metastases. Future studies have been planned to utilize a similar lymphatic metastasis approach within a tumor-bearing model.

In summary, we have developed a robust protocol for the induction of lymphatic metastasis in a murine model. This method will enable our future research focused on establishing the consequences of lymphatic metastasis, and by sharing it here, we hope will facilitate investigation of novel questions from others also.

Supplementary Material

Supplemental Figure 1. Supplementary Figure 1. Example of luminescence signal and region of interest definition.

(a) Luminescent signal overlaid on white light photographs of a mouse that had been injected with 3 million GFP+Luciferase+MC38 cells taken at days 1, 5, 8,12,15, 17 and 19 post-injection. (b) The region of interest was defined by the largest area of luminescence detected at any time point and then applied to all images from all time points. The luminescent signal from this region was graphed against time post-injection.

Supplemental Figure 2. Supplementary Figure 2. Evidence of lung and liver metastases in some, but not all, injected mice.

(A) in situ lungs displaying numerous metastatic foci (arrows point to two examples). These lungs are from one of the mice shown in Figure 5. (B) Hematoxylin and eosin-stained histological sections of lungs from two different mice showing metastatic lesions. Both images taken at 2X magnfication. (C) Hematoxylin and eosin-stained histological section of liver from an MLD-injected mouse showing metastatic lesions (4X magnification). (D) Lung section from another MLD-injected mouse showing absence of any metastatic foci (2X magnfication).

Abbreviations:

DPBS

Dulbecco’s phosphate buffered saline

IVC

inferior vena cava

LRV

left renal vein

MLD

mesenteric lymph duct

PVC

polyvinyl chloride

SMA

superior mesenteric artery