Intravital imaging of a spheroid-based orthotopic model of melanoma in the mouse ear skin

Abstract

Multiphoton microscopy is a powerful tool that enables the visualization of fluorescently tagged tumor cells and their stromal interactions within tissues in vivo. We have developed an orthotopic model of implanting multicellular melanoma tumor spheroids into the dermis of the mouse ear skin without the requirement for invasive surgery. Here, we demonstrate the utility of this approach to observe the primary tumor, single cell actin dynamics, and tumor-associated vasculature. These methods can be broadly applied to investigate an array of biological questions regarding tumor cell behavior in vivo.

Introduction

Tumor cell invasion is a critical step in the progression from a locally confined primary tumor toward metastatic disease. Live cell imaging in two-dimensional and three-dimensional cell culture systems has provided significant insight into how this process occurs. However, recent evidence suggests that the tumor microenvironment also plays a significant role in modulating the invasive behavior of tumor cells, and that tumor cells can influence the surrounding stromal cells.^1–3 Therefore, an understanding of cell behavior in in vivo models is required to more fully appreciate the complexities of invasive tumor cell migration. Genetically engineered mouse models with defined mutations, to some extent, can mimic those identified in human tumors and allow tumor development to be followed at various stages; however, generation of these models is costly and requires significant time investment. In contrast, orthotopic xenograft models allow more rapid analysis of tumor behavior but in an immunocomprised background.⁴ In recent years, multiphoton microscopy has emerged as a powerful approach to image tumor cells within tissues in vivo.^5–8 Furthermore, advances in intravital imaging techniques have provided a template to interrogate processes such as single cell and collective tumor cell invasion, intravasation into blood and lymphatic vessels, and associated innate and adaptive immune responses.^9–11

The mouse ear skin provides an attractive local milieu in which intravital imaging can address such biological questions regarding tumor cell invasion particularly in melanoma, a malignant lesion of melanocytic origin, which has the prominent clinical feature of early metastasis.^12,13 The ear skin comprises two major compartments: (1) the epidermis, which includes an avascular epithelial layer of keratinocytes, and (2) the underlying dermis, which is highly vascularized and is rich in structural matrix components such as collagen, fibronectin, and elastin. Human melanocytes typically reside at the basement membrane boundary between the epidermis and the dermis, whereas mouse melanocytes are rarely found at this junction and reside deep within the hair follicles or interfollicular dermis.¹⁴ Some notable exceptions are in the mouse ear, tail, and foot pad where the melanocyte distribution is similar to that of humans.¹⁵

Due to its similar melanocyte distribution, relative flatness, accessibility, and isolation from breathing motion and heartbeat, the mouse ear skin is a convenient location for intravital imaging.¹⁶ Previously, intravital imaging in the mouse ear has been used to study lymphatic drainage, vascular remodeling, and leukocyte migration.^17–19 We have developed an intravital imaging model that involves the orthotopic implantation of fluorescent multicellular tumor spheroids and allows tumor development to be followed longitudinally. Here, we describe methods for the immobilization of the mouse ear, visualization of subcellular actin dynamics and interactions with the collagen fiber matrix by second harmonic generation (SHG), and the use of contrast agents to observe the vasculature within the tumor. Our model is broadly applicable and will provide researchers with the valuable opportunity to visualize cell biological processes within an in vivo environment.

Results

We have previously developed a model of orthotopic injections of suspensions of fluorescent human melanoma cell lines into the ear skin of immunodeficient mice.²⁰ This procedure involves injecting a cell suspension of up to as many as one million cells, which is similar to methods used in other systems.^21,22 However, since tumors often develop at sites offset from the site of injection, this cell suspension approach does not allow for optimal imaging of tumor development (Fig. S1). In order to provide better spatial control, multicellular tumor spheroids have been used as in vitro and in vivo models for tumors.^23,24 To improve upon our existing model, we modified established methods by Timmins and Nielsen to generate highly reproducible fluorescently-labeled tumor spheroids with the goal of implanting them into the dermis of the mouse ear.^25–27 Using cell lines derived from a recently developed mouse model of melanoma (Tyr-CreER^T2/LSL-Kras^G12D/Lkb1^L/L),²⁸ as well as other well-characterized metastatic mouse and human melanoma cell lines such as B16F10 and WM266-4, respectively, we can consistently prepare spheroids containing a small number of cells (~2,000–4,000) and of a well-defined size of approximately 200 μm in diameter (Fig. 1A). These tumor spheroids are visible by the unaided eye (Fig. 1B). A single tumor spheroid in a 3 μL volume of DMEM is drawn into a 10-μL Hamilton syringe equipped with custom 27G needle. For the intradermal injection, the mouse is anesthetized and the ear is taped onto a 14-mL conical tube with the cap removed, which is affixed onto a heating pad (Fig. 1C). The spheroid is then injected into the skin, which results in a transient visible bump after injection (Fig. 1D). The small injection volume minimizes the amount of tissue injury. By macroscopic epifluorescence imaging, one can confirm successful spheroid implantation immediately after injection (Fig. 1E and F). Over the course of several weeks, these spheroids develop into a highly localized primary tumor (Fig. 1G–I). Beyond providing tight control over tumor location, this method also allows for reproducible tumor take and growth kinetics. Unlike cell suspension injections, spheroid injections yield extremely high tumor take (> 98% from over 200 injections). In addition, tumors are palpable after 1–2 weeks, which is followed by the recruitment of highly irregular blood vessels.

Figure 1.

Procedure for spheroid implantation. (A) A tumor spheroid from mouse melanoma cells expressing GFP and Lifeact-tdTomato was generated by the hanging droplet method. Maximum intensity confocal projection shows overlay image of whole mount spheroid. Scale bar, 50 μm. (B) Photograph of single spheroid in 100 μL media in a 0.5-mL tube. Inset shows magnified image of spheroid. (C) Setup for intradermal injection of tumor spheroid into the mouse ear. A conical tube is taped onto a heating pad. While the mouse is anesthetized, the ear is taped to a conical tube by double-sided tape. The spheroid in media is drawn in a 2–3 μL volume of media into a 10-μL Hamilton glass syringe equipped with a custom 1-in 27G needle. The syringe is held vertically for 1 min to allow the spheroid to settle into the needle. Under a dissecting microscope, the needle is held at a 30° angle bevel up and the spheroid in media is injected into the dermis of the anesthetized mouse. (G) Brightfield macroscopic image of site of injection. A transient bump is visible immediately after injection. (H) GFP epifluorescence image and characteristic “stained glass” pattern shows spheroid beneath the surface of the skin. (I) tdTomato epifluorescence image of spheroid. All other scale bars, 1 mm.

In addition to the ability to track tumor progression from its site of initiation, we have established methods to perform intravital imaging of single cell behavior within the tumor by confocal and multiphoton microscopy. To accomplish this, we have developed a standardized protocol to immobilize the mouse ear. For multiphoton microscopy, we engineered a custom aluminum clamp (Fig. 2A; Fig. S2). The aluminum facilitates transfer of heat to the ear from an underlying heating plate/pad (Fig. 2B). The ear clamp consists of two halves: (1) a bottom platform on which the ear rests and (2) an upper piece equipped with a pair of thumbscrews to secure the ear (Fig. 2C). A 12-mm #1.5 coverslip is cut in half and affixed onto the upper half of the clamp with a clear nail polish to provide a flat surface to image the ear. For upright confocal microscopy, which uses a dry objective, we tape the underside of the ear to the lower half of the clamp with doubled-sided tape and do not use the upper half. We place the mouse on a heating pad equipped with temperature feedback to keep the animal warm during imaging. We apply a viscous optical coupling gel with a refractive index matched to water to minimize evaporation during imaging.²⁹ For applications where precise temperature control is required, it may be necessary to include an objective heater, as the objective does reduce the temperature of the ear by approximately 4°C lower than body temperature.

Figure 2.

Immobilization of the mouse ear for intravital imaging. (A) Aluminum clamp engineered to immobilize the mouse ear. The lower half is mounted onto a larger 50-mm × 25-mm × 10-mm aluminum block. The lower half of the clamp has a raised platform to match the curvature of the ear. A #1.5 coverslip is affixed onto the upper surface of the clamp to provide a flat surface for imaging the tumor. A pair of thumbscrews is used to secure the upper half to the lower half of the clamp. (B) Photograph of mouse ear immobilization setup for multiphoton imaging. (C) Side view schematic of ear immobilization. A heating pad is taped down onto a closed staged insert. The ear clamp is taped onto the heating pad. The ear is placed onto the lower half of the ear clamp. An optical coupling gel that matches the refractive index of water is placed on top of the ear. The upper half of the ear clamp is secured over the ear. Another layer of optical coupling gel is placed in between the coverslip and the objective. (D) Multiphoton imaging through 90-μm section of melanoma tumor expressing GFP and Lifeact-tdTomato. Maximum intensity projection image of GFP. Dark holes are hair follicles and additional areas devoid of fluorescence are blood vessels. (E) Lifeact-tdTomato image. (F) SHG image of collagen fibers. (G) Overlay image of GFP, tdTomato, and SHG. Scale bars, 50 μm.

The use of a single tunable laser source allows for high spatial resolution but low temporal resolution when switching excitation wavelengths. Therefore, it is necessary to choose a combination of fluorophores that can be excited by a single wavelength for optimal temporal resolution. Using a 910-nm excitation wavelength on our multiphoton system and nondescanned detectors with appropriate filter sets, we have simultaneously imaged tumor cells expressing both green fluorescent protein (GFP) and tdTomato. At the same excitation wavelength, by second harmonic generation (SHG), we have simultaneously visualized the dense network of collagen fibers in addition to GFP and tdTomato (Fig. 2D–G). Using the SHG image as a guide to assess the location of imaging within the dermis,³⁰ we can visualize the tumor cells in the dermis (depth range of 30 μm to ~200 μm), between the epidermis and the cartilage layer. We have observed structural changes in the organization of the surrounding collagen matrix as the tumor develops.^2,31–33 In addition, we can identify known structures such as hair follicles and negative regions devoid of fluorescence corresponding to blood vessels. To test for the compatibility of this imaging regime with the melanin pigment frequently found in melanoma cells, we generated pigmented tumors from B16F10 cells in the ear of unpigmented mice and found that multiphoton fluorescence imaging of GFP is unaffected by the presence of pigment in the cells (Fig. S3). However, similar to previous observations,¹¹ in pigmented mice high-intensity “speckling” may pose an issue. Therefore, we recommend the use of non-pigmented mice in this spheroid implantation model.

Not only can we visualize single cells, but we can also distinguish fluorescently labeled subcellular organelles with this method. One commonly used approach to monitor cell polarization in in vitro wound healing assays is to examine Golgi positioning relative to the nucleus.^34–36 Using melanoma cells simultaneously expressing a GFP fusion to an N-terminal fragment of β-galactosyltransferase³⁵ and a nuclear localization signal fused to tdTomato, we have observed polarized invasion from the tumor spheroid into the surrounding tissue (Fig. 3A–D). Another approach to examine subcellular activities using this method is to visualize actin dynamics in single cells in vivo (Fig. 3E–G; Vid. S1). Through a 100-μm section of a tumor at an imaging rate of 1.25 frames per second, we have visualized highly dynamic leading edge protrusions of melanoma cells expressing the actin marker Lifeact-GFP (Vid. S2). We detected the presence of bundled actin as well as actin retrograde flow within lamellipodia. We anticipate that these methods will allow the visualization of additional subcellular dynamics and facilitate future quantitative analysis.

Figure 3.

Intravital imaging of subcellular organelles. (A–D) Multiphoton imaging through 150-μm section of melanoma tumor expressing GFP-labeled Golgi and tdTomato-labeled nucleus. (A) Maximum intensity projection of GFP-labeled Golgi. (B) Nuclear localized tdTomato image. (C) SHG image of collagen fibers. (D) Overlay image shows polarization of Golgi in cells invading from the tumor into the surrounding tissue. (E–H) Multiphoton imaging through 100-μm section of tumor expressing GFP and Lifeact-tdTomato. (E) Maximum intensity projection image of GFP. (F) Lifeact-tdTomato image. (G) SHG image of collagen fibers. (H) Overlay image shows leading edge protrusions in a single cell. Scale bars, 20 μm.

After two weeks of tumor development in our model, we frequently observe neo-vascularization of the tumors. To visualize the vascularization within the tumor by epifluorescence and confocal microscopy, we performed tail vein injections of AlexaFluor 647-labeled BSA (Fig. 4A–D). We insert a tail vein catheter into the mouse and inject a 75 μL volume of 5mg/mL fluorescently labeled BSA. Within 30 sec, we can observe the vascular network within the tumor (Fig. 4E–H; Vid. S3). This allows us to directly visualize the morphology of tumor-associated blood vessels and to potentially visualize events such as tumor cell intravasation and tumor-associated enhanced permeability and retention (EPR).³⁷ We expect that similar methods can be easily applied to multiphoton imaging in which others have imaged the vasculature with Evans Blue dye, fluorescent dextrans and lectins, and quantum dots.^38,39

Figure 4.

Intravital imaging of tumor-associated vasculature. (A) Macroscopic brightfield image of tumor expressing tdTomato. (B) Epifluorescence tdTomato image. (C) AlexaFluor 647-BSA dye was injected by tail vein and is shown in blue. (D) Overlay image shows vasculature in the primary tumor and the surrounding tissue. (E) Confocal imaging of tumor expressing GFP and tdTomato. GFP image. (F) tdTomato image. (G) AlexaFluor 647-BSA dye was injected by tail vein and is shown in blue. (H) Overlay image shows tumor and associated vasculature. Scale bars, 500 μm.

Discussion

The mouse ear skin has recently gained popularity as a site for intravital imaging of leukocyte response to infection and injury, lymphatic drainage, and tumor cell behavior.^11,40–43 Our orthotopic model allows direct imaging of melanoma tumor progression in this tissue. One significant advantage of our orthotopic spheroid implantation model is the ability to reproducibly establish and directly image a tumor from its initiation from a few thousand cells. In addition, these methods are unique in that they do not require invasive animal surgery such as the implantation of imaging windows.^44–46 While the use of such windows facilitates intravital imaging of the tumor, one caveat is that the tumor is growing in a non-physiological environment. Our model provides the ability to perform longitudinal studies to track the early stages of tumor development as well as the ability to image melanoma cell behavior within a more physiologically relevant environment.

While we have demonstrated the utility of our tumor spheroid implantation approach by using immunodeficient athymic nude mice and recognize that this does not fully recapitulate the microenvironment in an immunocompetent animal, the ear immobilization and imaging strategies can be readily adaptable to a wide array of fluorescent transgenic mice or syngeneic mouse models. We anticipate being able to use our model to visualize tumor cell interactions with stromal fibroblasts, monocytes/macrophages, and blood vessels. In conjunction with these techniques, our laboratory has also established an in vivo nanoparticle clearance assay by imaging the vasculature within the ear.⁴⁷ We expect in the future to be able to combine these assays to directly image interactions between therapeutic nanoparticles and tumors in a physiologically relevant setting.

One limitation of our current protocol of using inhalable anesthetics is that we can only observe cellular events within the time course of a few hours. While this provides ample time to image protrusion dynamics, imaging of slower processes such as cell translocation is limited. However, this can be circumvented by using intravenous anesthetics of ketamine/xylazine for longer term imaging.¹⁶ Other groups have also used ear explants to extend the imaging time.^43,48

Together, the methods described here provide a potential opportunity to gain a greater understanding of the mechanisms of in vivo tumor cell invasion. One can also envision the use of our model in examining challenging aspects of tumor biology such as the effects of primary tumor resection on metastasis and the effects of therapeutics on tumor cell motility at different stages of tumor progression.

Materials and Methods

Ethics statement

All methods involving the use of animals in this study were done in accordance with the NIH Guide for the Care and Use of Laboratory Animals. Procedures were approved by the University of North Carolina, Chapel Hill Institutional Animal Care and Use Committee (IACUC) (Protocol #: 10-281.0). All tumor spheroid injections and animal imaging were performed under inhalable isoflurane anesthesia and efforts were made to minimize animal discomfort.

Cells and cell culture

Murine melanoma cell lines were generated as described previously²⁸ and were cultured in high glucose Dulbecco’s Modified Eagle’s medium (Gibco) supplemented with 10% FBS (HyClone), 1% penicillin and streptomycin (Gibco), and 1% GlutaMax (Gibco). All cell lines were tested for pathogens prior to injection. Lentivirus production and infection were performed as described previously.⁴⁹ Cell lines expressing fluorescent proteins were sorted by FACS.

Hanging droplet spheroids

Hanging droplet spheroids were generated according to modified methods of those previously described.^25–27 Cells were trypsinized and resuspended in media at 0.5–1 × 10⁵ cells/mL. Twenty microliters of cell suspension was placed into each well of a Nunc 60-well minitray (ThermoFisher Scientific). The minitray was inverted and placed into a local humidification chamber. The humidification chamber was prepared by placing a 60-mm tissue culture dish (with lid removed) filled with 8 mL PBS inside a 150-mm tissue culture dish. Cells were incubated at 37°C 5% CO₂ for 2–6 d to allow spheroid formation at the medium-air interface.

Intradermal injection of tumor cell spheroids

Single spheroids were harvested and placed into 100 μL media in a 0.5 mL microcentrifuge tube. Single spheroids in the tube were verified visually under a brightfield microscope and were placed on ice until time of injection. Athymic nude mice were purchased from the UNC Animal Services Core. Mice were anesthetized by inhalation with 2% isoflurane. A 14-mL conical tube was taped onto a heating pad next to the isoflurane nose cone under an Olympus MVX10 Macroview macroscope. The MVX10 is equipped with a MVPLAPO 1× objective, total mag 6.3–63×, NA 0.25, W.D. 65 mm. The mouse ear was affixed onto the conical tube with double-sided tape. The tumor spheroid, which was visible to the unaided eye in the tube, was drawn into a 10-μL glass syringe (Hamilton) with a custom 1-inch 27G needle in a total volume of 2–3 μL media. The glass syringe was held vertically for 1 min to allow the tumor spheroid to settle into the needle. While observing the ear through the macroscope, the needle (bevel facing up) was pierced into the dermis of the mouse ear and the spheroid in media was injected, which resulted in a transient visible bump under the surface of the skin. Successful spheroid injection was verified immediately by epifluorescence imaging with the MVX10, which has Cy5, GFP, and Texas Red filters. Images were captured with a Hamamatsu ORCA 03G CCD (high resolution), 1,344 × 1,024 resolution, 6.5 μm pixels, 12 bit-up to 9 fps.

Intravital imaging

Mice were imaged using an Olympus IV100 or FV1000MPE microscope. The IV100 is an inverted laser scanning microscope equipped with four lasers (488/561/633/748 nm) and allows for simultaneous detection of up to three fluorescent channels. Mice were imaged on the IV100 using a 10× air, 0.4 NA, WD 3.1 mm; or 4× air, 0.16 NA, W.D. Thirteen millimeters objective. For imaging of the vasculature, a tail vein catheter was inserted to the mouse and a 75-μL volume of AlexaFluor 647-BSA (Invitrogen) was injected. For deeper tissue imaging by multiphoton, mice were anesthetized with 2% isoflurane and placed on a 5-in × 8-in heating pad with temperature feedback to a Model TC-1000 Temperature Controller (CWE Inc.). Multiphoton imaging was performed on an Olympus FV1000MPE mounted on an upright BX-61WI microscope, using a 25×, 1.05 N.A. (2 mm W.D.) water immersion objective. The laser unit is a MaiTai DeepSee tunable from 690–1040 nm with a pulse width < 100 fs. There are 4 Channel Non-descan Detectors, Ch1 (420–460 nm) BFP, Ch2 (495–540 nm) GFP, Ch4 (575–630 nm) RFP, Ch3 (380–560 nm) wide range BFP, CFP, GFP, YFP. We engineered an aluminum clamp to immobilize the ear during imaging (Fig. S1). Isofluorane was continuously administered by inhalation throughout the experiments for no longer than 4 h. To decrease the rate of evaporation during imaging, an optical coupling gel with a similar refractive index to water was prepared using a solution of 300 mM D-sorbitol (Sigma-Aldrich) and 0.5% Carbomer 940 (Spectrum Chemical) as a gelling agent, which was adjusted to pH 7.4.²⁹ To simultaneously image GFP, tdTomato, and collagen by SHG, we used 910-nm excitation wavelength. The average power at the sample was measured using a Coherent FieldMaster GS laser analyzer and determined to be approximately 15 mW. Images were despeckled in ImageJ to remove background noise and prepared in Adobe Illustrator. The ImageJ StackReg plugin was used to correct for drift during acquisition of time-lapse movies.

Abbreviations

GFP

green fluorescent protein

SHG

second harmonic generation

EPR

enhanced permeability and retention