Investigating Lymphangiogenesis in a Sacrificially Bioprinted Volumetric Model of Breast Tumor Tissue

Abstract

Lymphatic vessels, as a means to metastasize, are frequently recruited by tumor tissues during their progression. However, reliable in vitro models to dissect the intricate crosstalk between lymphatic vessels and tumors are still in urgent demand. Here, we describe a tissue- engineering method based on sacrificial bioprinting, to develop an enabling model of the human breast tumor with encapsulated multiscale lymphatic vessels, which is compatible with existing microscopy to examine the processes of lymphatic vessel sprouting and breast tumor cell migration in a physiologically relevant volumetric microenvironment. This platform will potentially help shed light on the complex biology of the tumor microenvironment, tumor lymphangiogenesis, lymphatic metastasis, as well as tumor anti-lymphangiogenic therapy in the future. We further anticipate wide adoption of the method to the production of various tissues and their models with incorporation of lymphatics vessels towards relevant applications.

1. Introduction

Cancer is the leading cause of death worldwide according to the World Health Organization [1]. The lethality of cancer is caused primarily by metastasis, which accounts for majority (~90%) of cancer deaths and often occurs via the lymphatic vascular system [2–4]. The role of lymphatic vascular system in promoting cancer metastasis has received increasing attention in the past decades. Lymphangiogenesis, the process of the formation of new lymphatic vessels from pre-existing ones, involves several important steps: proliferation, migration, and tube formation by the lymphatic endothelial cells (LECs) [5, 6]. The interplay between LECs and their microenvironment is instrumental in lymphatic metastasis of tumor cells, and lymphatic vessels provide a potential therapeutic target to control cancer metastasis by modulating the various responses of the tumor microenvironment (TME) [5–7].

Two-dimensional (2D) cancer cell and LEC cultures can only provide planar environments leading to their inability to mimic the complex interactions between the cancer cells and the extracellular matrix (ECM) in vivo [8–10]. By contrast, three-dimensional (3D) models that host a co-culture of the different cell populations can possibly bridge this gap by supplying improved biomimicry of the ECM as well as the TME [10–12]. As discussed, lymphangiogenesis has attracted attention in a wide range of thrusts such as in cancer metastasis, and in providing a preferential recycling route to most administered antitumor drugs in our body [13, 14]. Yet, there are only few reports published so far relating to engineering of in vitro lymphatic vascular models that allow recapitulation of this important component of the TME. Kim et al. developed a microfluidic platform that reconstituted the lymphatic sprouting and demonstrated the effect of the microenvironmental contexts on lymphangiogenesis [15]. Furthermore, Gibot et al. developed a 3D lymphatic capillary network system containing co-cultured fibroblasts and LECs to evaluate lymphatic vascular permeability [16]. In another report, Osaki et al. engineered a microchannels-containing microdevice inducing lymphangiogenesis and vascular angiogenesis to evaluate its application in drug screening [17]. A properly designed 3D model for reproducing the lymphatic vessel-tumor interactions is critical in probing the efficacies of tumor therapies.

More recently, 3D bioprinting has found growing utility in the fabrication of volumetric tissues and their models. The 3D bioprinting technology is a reproducible and scalable fabrication strategy providing more precise spatial control of geometries in comparison to the conventional biofabrication methods [18]. In particular, the sacrificial 3D bioprinting technique has been widely adopted for the fabrication of vascularized tissue models include those of the cancer, which involves the deposition of a fugitive microfibrous structure that is subsequently removed leaving behind a hollow conduit in the surrounding hydrogel matrix [19]. For example, Kolesky et al. engineered tissue constructs featuring embedded vasculature and multiple types of cells using Pluronic F127 as the sacrificial ink [20]. Moreover, we designed an in vitro mammary ductal carcinoma model with the sacrificial bioprinting strategy, using agarose for generating the hollow microchannels within an ECM- like hydrogel construct [21]. However, there are very few prior reports on building an in vitro tumor model that incorporates lymphatic vasculature through bioprinting, in an effort to study their interactions and potential therapeutic effects [22].

Here, we describe a method for the construction of a 3D in vitro model of breast tumor lymphangiogenesis in a gelatin methacryloyl (GelMA) matrix. We engineered the model by seeding human LECs in the sacrificially bioprinted GelMA-embedded microchannel, with MDA-MB-231 breast tumor cells encapsulated in the surrounding GelMA hydrogel. In addition to observations of tumor cell-induced lymphangiogenesis, our investigation further proved that vascular endothelial growth factor (VEGF)-C could promote LEC sprouting. We anticipate our model to find widespread applications in understanding the tumor-lymphatic vessel interactions as well as in developing therapeutics in the future.

2. Material and Methods

2.1. Preparation of GelMA

Materials used for fabrication of the GelMA constructs were obtained from the following sources: gelatin Type A from porcine skin and methacrylic anhydride were purchased from Sigma-Aldrich (USA), and lithium phenyl (2,4,6-trimethylbenzoyl) phosphinate (LAP) was obtained from Allevi (USA). GelMA was synthesized according to our previously published procedures [21]. Briefly, gelatin was mixed at 10 wt.% in phosphate-buffered saline (PBS) without calcium and magnesium and stirred at 50 °C until fully dissolved. Methacrylic anhydride was added gradually until a concentration of 5 mL in 100 mL of gelatin solution. The solution was stirred for 2 h at 50 °C. Following a 2× dilution with pre-heated PBS, the solution was dialyzed for 7 days against deionized water at 50 °C. The dialyzed GelMA solution was filtered and freeze-dried. The obtained GelMA foam was stored at −80 °C until further use.

2.2. Cell culture

MDA-MB-231 breast cancer cells (HTB-26, ATCC, USA) were cultured using Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1 vol.% antibiotic-antimycotic (anti-anti), all from Thermo Fisher (USA). Human dermal lymphatic microvascular endothelia cells (LECs, HMVEC-dLyAd, Lonza, USA) were cultured in endothelial basal medium-2 (EBM-2, Lonza) supplemented with the BulletKit (endothelial growth medium, EGM). Early passages of LECs were used in the study. The common medium (45 vol.% DMEM:45 vol.% EGM:10 vol.% FBS) was used for co-culture of MDA-MB-231 cells with LECs. All cells were cultured at 37 °C and 5% CO₂ in a humidified incubator.

2.3. Fabrication of the lymphatic vessels-impregnated breast cancer (LV-BC) model

Agarose (Sigma-Aldrich) was used as the sacrificial material to allow the formation of the microchannels in the GelMA constructs. An agarose solution (8 wt.% in PBS) heated at 80 °C was loaded into a glass capillary (500 μm in diameter) and then placed at 4 °C in PBS for 10 s to achieve gelation. To create fully surrounded 3D cylindrical microchannels in GelMA, GelMA hydrogel precursor (5 wt.% containing 0.3 wt.% LAP in PBS) was poured in a polydimethylsiloxane (PDMS) container with a dimension of 10 × 8 × 4 mm³ (L × W × H), and then semi-crosslinking for 10 s under UV light. An agarose microfiber was gently pushed out from the glass capillary using a stainless-steel piston and deposited onto this semi-crosslinked GelMA layer in a pre-defined pattern controlled by a bioprinter (in-house made), following which more GelMA solution mixed with 5 × 10⁶ cells mL⁻¹ of MDA-MB-231 cells was poured to fill the entire PDMS mold. After crosslinking for additional 40 s under UV light, the agarose microfiber was removed by manual extraction to form the hollow microchannel. Subsequently, an LEC suspension (1 × 10⁶ cell mL⁻¹ in EBM-2) was injected into the microchannel. After filling with the cell suspension, the construct was maintained in an incubator. To improve cell adhesion, the GelMA hydrogel construct was inverted several times over a 3-h period. After cell attachment onto the inner surface of the microchannel, the co-culture medium was gently added for the culture of the LV-BC model.

2.4. Treatment with VEFG-C

The LV-BC models were treated with or without 250 ng mL⁻¹ of VEGF-C (PeproTech, USA). In all cases, observations and measurements of LEC and MDA-MB-231 migration were performed on days 14, 17, and 20 of culture.

2.5. Cell viability assay

Cell viability was measured every 3 days using the LIVE/DEAD® Viability/Cytotoxicity Kit (Thermo Fisher). In brief, the medium of each tissue construct was removed and then washed with PBS. Next, a working solution of 2 μL mL⁻¹ of ethidium homodimer-1 and 0.5 μL mL⁻¹ of calcein AM in PBS was made and added to each construct. Then, the constructs were incubated at 37 °C for 20 min. Finally, the constructs were washed with PBS again and resuspended in it for imaging. Image J (National Institutes of Health, USA) was used to count the numbers of live and dead cells.

Cell morphology was assessed via F-actin/nuclei staining. Briefly, the tissue constructs were fixed with 4 vol.% paraformaldehyde (PFA, Sigma-Aldrich) and permeated with 0.1 vol.% of Triton X-100 (Sigma-Aldrich). Nuclei and F-actin were stained using 4’,6- diamidino-2-phenylindole (DAPI, Thermo Fisher) and Alexa 488-phalloidin (Thermo Fisher), respectively. Alexa 488-phalloidin was diluted at a ratio 1:40 vol.% in 0.1% wt.% bovine serum albumin (BSA, Sigma-Aldrich) in PBS. In a similar way, the working solution of DAPI diluted at 1:1000 vol.% in PBS was prepared. Afterwards, the constructs were washed with PBS three times for 5 min each. Then, the prepared phalloidin staining solution was added to each of them and allowed to incubate for 45 min at 37 °C. Subsequently, the constructs were washed with PBS and then stained with working DAPI solution incubated at 37 °C for 5 min. The constructs were finally washed once with PBS and resuspended in it for storage and fluorescence imaging.

2.6. Immunofluorescence staining

CD31 and epidermal growth factor receptor (EGFR) were characterized for LECs and MDA-MB-231 cells, respectively, using immunostaining with anti-human CD31 antibody (Abcam, USA)/Alexa 594-conjugated secondary antibody (goat anti-mouse IgG, Thermo Fisher) and Alexa 488 anti-human EGFR antibody (BioLegend, USA), respectively. The tissues were washed three times with PBS, fixed with 4 vol.% PFA in PBS for 15 min at 37 °C, and followed by three PBS washes at room temperature and permeation with 0.1 vol.% Triton X-100 for 20 min at 37 °C. The samples were blocked using 5 vol.% BSA for 30 min at room temperature and then incubated for overnight (>16 h) at 4 °C with primary CD31 antibody or Alexa 488 anti-human EGFR antibody. This step was followed with three washes using PBS. Secondary antibody in 5 vol.% BSA was incubated for 2 h to visualize CD31. The stained samples were imaged at room temperature.

2.7. Image acquisition and processing

Brightfield and immunofluorescence images of the tissue constructs were acquired with a fluorescence microscope (Eclipse, Nikon, Japan). Confocal images were acquired with a Zeiss confocal microscope (LSM 880 with Airyscan, Carl Zeiss, Germany). ImageJ was used to merge channels and render 3D reconstruction for confocal images and stacks. To measure the distance of LEC migration also using ImageJ, first we set the scale for each image, and then a rectangular area was drawn to cover about 90% of LECs that migrated away the initial position. Then the width of the rectangle was measured using the analyze function of ImageJ and calculated against the scale, to determine the distance of LEC migration.

2.8. Statistical analyses

Statistical analyses were conducted by GraphPad Prism 8 (La Jolla, USA). Data of cell viability were presented as means ± standard deviations. Significance tests were performed using Students’ t‐test. Results were considered significant for p ≤ 0.05. All samples used for quantifications were three in all groups, and at least five randomly selected images were taken per sample for analysis.

3. Results

3.1. Construction of the sacrificially bioprinted constructs

The overall procedure for sacrificial bioprinting of the tissue-like constructs with embedded hollow microchannels is divided into several steps (Fig. 1A). In the first step, a PDMS mold (Fig. 1A–i) is filled with a layer of GelMA solution whose volume is half of the PDMS model (Fig. 1A–ii), and semi-crosslinked for 10 s to become a weak hydrogel (Fig. 1A–iii). An agarose microfiber is then extruded onto this GelMA layer using a modified bioprinter (Fig. 1A–iv), after which another half of GelMA solution containing MDA-MB-231 breast tumor cells was filled (Fig. 1A–v), and subsequently UV-crosslinked for 40 s (Fig. 1A–vi). In the final step, the agarose microfiber is gently removed from the GelMA block using physical extraction (Fig. 1A–vii), to realize the formation of the hollow microchannel (Fig. 1A–viii) that allows perfusion and LEC seeding (Fig. 1A–ix).

FIG. 1.

Construction of sacrificially bioprinted hydrogel-based tissue models containing embedded microchannels. (A) The sacrificial bioprinting procedure of the LV-BC model: i) preparation of a PDMS mold; ii) dispersing a layer of GelMA solution (half of the PDMS mold) followed by iii) semi-crosslinking by UV for 10 s; iv) bioprinting the agarose microfiber on top of the semi-crosslinked GelMA base in the center; v) dispensing the GelMA solution containing MDA-MB-231 breast tumor cells to fill the rest of the mold and vi) further UV crosslinking for 40 s; vii) removing the microfiber to create the microchannel; viii) seeding LECs in the microchannel; and ix) subsequent culture. (B) Photographs showing i-ii) the different designs of molds and iii-v) the prefusion of microchannels with different configurations or diameters in the GelMA constructs.

To illustrate the versatility of generating the microchannels, a collection of photographs was obtained. As shown in Fig. 1B–i and ii, the PDMS molds could be fabricated for preparing the models containing one (Fig. 1B–iii) or two (Fig. 1B–iv) microchannels depending on the need. By using agarose microfibers deposited with different diameters, we could further produce microchannels that were either thinner (300 μm in diameter, Fig. 1B–iii and iv) or thicker (1.2 mm in diameter, Fig. 1B–v). These microchannels within the GelMA hydrogel constructs could be further seeded with cells to endow them with biological functions.

3.2. Preparation of the LV-BC model

To model the lymphatic vessel for studying lymphangiogenesis, we populated LECs on these perfusable and structurally stable GelMA hydrogel-embedded microchannels, where the GelMA matrices were encapsulated with MDA-MB-231 breast tumor cells (Fig. 2A). From time-lapse micrographs taken at different days of culture, we could observe the behaviors of the LECs on the surface of a microchannel (Fig. 2B). The LECs were almost rounded and crowded at day 2, where they were then able to spread and proliferate in the following days, eventually covering the full area of the microchannel. The growth of the MDA-MB-231 cells within the GelMA construct over different days was also microscopically demonstrated (Fig. 2C).

FIG. 2.

Characterizations of the sacrificially bioprinted LV-BC model. (A) Schematics showing the arrangement of the LECs and the MDA-MB-231 breast tumor cells in the model. (B) Bright-field time-lapse top-view micrographs showing the proliferation of LECs on the surface of the microchannel. (C) Fluorescence time-lapse top-view micrographs showing the proliferation of MDA-MB-231 breast tumor cells in the surrounding GelMA matrix. (D) Confocal reconstruction images showing the LECs (left) and MDA-MB-231 breast tumor cells at day 8 of culture. The inset is a high-magnification image showing the CD31 expressions, where CD31 was stained in purple and nuclei in blue. (E) Cross-section reconstruction views of LECs in the microchannel and MDA-MB231 breast tumor cells in matrix at day 8 of culture. (F) Viability quantifications of LECs and MDA-MB-231 breast tumor cells in the model for up to 21 days of culture.

By approximately day 8, the formation of the structurally intact cylindrical lymphatic vessel was clearly indicated by CD31 immunofluorescence staining (Fig. 2D, left). The homogeneously distributed and well-spread MDA-MB-231 cells were also detected by EGFR antibody (Fig. 2D, right). Further analysis revealed that the initial distance between the lymphatic vessel and breast tumor cells was approximately 100 μm as revealed by cross-sectional confocal reconstruction images (Fig. 2E), which was caused by the slight sinking of the deposited agarose microfiber into the semi-crosslinked bottom GelMA hydrogel before filling the second half of the MDA-MB-231 cell-containing GelMA solution. Yet, this separation possibly mimicked the early stages of the tumor progression where the two cell types are not yet in close proximity. Viability assays further indicated that LECs in the microchannels exceeded 80% of viable cells, and MDA-MB-231 breast tumor cells in the GelMA hydrogel matrices were also >80% viable throughout the entire co-culturing period of up to 21 days (Fig. 2F). There were no statistical differences between the viability values measured across the different days.

3.3. Breast tumor cells induced lymphangiogenesis in the LV-BC model

Lymphangiogenesis is a key step in tumor metastasis. To evaluate the function of our LV-BC model to mimic the dynamic lymphangiogenesis process, we took advantage of VEGF-C to stimulate the sprouting of LECs. As a ligand, VEGF-C acts on LECs primarily via its receptor VEGFR-3 promoting survival, growth, and migration of these cells [23]. Consistently, we proved that LECs were able to well-sprout out of the walls of the microchannels after VEGF-C (250 ng mL⁻¹) treatment for up to 17 days (Fig. 3A). Further F-actin staining also indicated that a large quantity of LECs migrated from the microchannel into the surrounding GelMA hydrogel matrix (Fig. 3B).

FIG. 3.

Breast tumor cells induced lymphangiogenesis in the LV-BC model, which was further promoted by VEGF-C stimulation. (A) Optical micrographs showing the migration of LECs from the microchannel inward out. (B) F-actin staining and CD31 staining of the LECs revealing the sprouting at 20 day. (C) Comparisons of LEC sprouting (CD31 staining) from the microchannels in the absence and presence of VEGF-C (250 ng mL⁻¹) at days 14, 17, and 20 of culture. (D) Quantification of the migration distances of LECs from the microchannels under different conditions.

To further quantify LEC migration, we measured the distances of all samples at different time points. From the fluorescence images (Fig. 3C), we could find that with the increase of time in co-culture, the distance of the migrated LECs increased. In addition, the sprouting lengths of the LECs stimulated by VEGF-C were much greater than those untreated with VEGF-C but solely dependent on the induction by the breast tumor cells. Quantification analyses confirmed the significant differences in the sprouting distances between the LECs treated with and without VEGF-C (Fig. 3D).

3.4. The crosstalk of lymphatic vessels and breast cancer cells in the LV-BC model

The interplay between the breast tumor cells and lymphatic vessels over culture was observed in confocal reconstruction images obtained for the different samples. Besides that the sprouting of the LECs from the microchannel was visible at day 17, the MDA-MB-231 cells were observed to also migrate towards the microchannel bridging the initial gap in between the two cell types as indicated in Fig. 2E, where they also appeared to be more stretched than those in the surrounding matrix (Fig. 4A). Moreover, after VEFG-C stimulation, the migration of LECs was more pronounced (Fig. 4B). Interestingly, the growth of the MDA-MB-231 breast tumor cells seemed to have also been promoted than the ones without stimulation. At day 20, these trends became more visible (Fig. 4C and D).

FIG. 4.

The interactions of LECs and the MDA-MB-231 breast tumor cells in the LV-BC model. (A, B) Confocal reconstruction images showing crosstalk of the cells at day 17 of culture in the (A) absence and (B) presence of VEGF-C. (C, D) Confocal reconstruction images showing crosstalk of the cells at day 20 of culture in the (C) absence and (D) presence of VEGF-C. LECs were stained with CD31 (red) and MDA-MB-231 breast tumor cells with EGFR (green).

4. Discussions

In the sacrificial bioprinting strategy for engineering the lymphatic vessel-embedded breast tumor model, it was found that the agarose microfibers as the sacrificial ink material could be readily pulled out from the GelMA matrices. The use of the sacrificial agarose microfibers also prevented osmotic damage to the cells in the surrounding matrices such as in the use of carbohydrate glass microfibers, with the latter requiring proper coatings to isolate the dissolved sugar at high concentrations from the cells [24, 25]. Pluronic solution presents a sol-gel transition that is temperature-dependent, and therefore at room temperature the gel can be printed while dissolving away at low temperatures (e.g., 4 °C) [25–27]; however, a high concentration of Pluronic F127 has been shown with different levels of cytotoxic effects necessitating careful removal of all residues of this sacrificial ink. To this end, the extraction of agarose microfibers also does not leave any noticeable trace on the surface of the resulting microchannels, leaving minimum concerns regarding cytotoxicity. This sacrificial bioprinting method for generating microchannel-embedding constructs, and in this work, a breast tumor model containing a central lymphatic vessel, is effective. Nevertheless, tuning the parameters for generating an intact monolayer of a different cell type other than those reported before [21, 28–30], required iterative re-optimizations. Indeed, under our optimized conditions including the GelMA concentration, the cell seeding density, and the culture conditions, among others, high cell viability values (>80%) of the LECs were observed, as well as the decent expression of CD31 by the same cells.

This model provides a convenient platform to study the interactions of LECs on the surface of the microchannels with the breast tumor cells in the surrounding ECM. GelMA hydrogels function as a benign microenvironment for the survival, spreading, proliferation, and migration of cells [31, 32]. As a hydrolysis product of collagen, the major component of ECM in most tissues, gelatin contains intrinsic arginine-glycine-aspartic acid (RGD) motifs that promote cell attachment [33, 34]. Cells can be suspended in GelMA solution and crosslinked on-demand upon exposure to light to form volumetric hydrogel constructs with tunable mechanical properties [34]. The viability of MDA-MB-231 cells laden in GelMA constructs as a breast tumor tissue mimic was more than 80% over the 21-day culture time, which further showed strong EGFR expressions.

Our sacrificially bioprinted LV-BC model could also be effectively used as a system to investigate the interplay between LECs and tumor cells in a 3D TME-like setting, such as their proliferation and migration. VEGF-C has been shown to induce survival, proliferation, and migration of various endothelial cells including the LECs [35]. Our results confirmed that the sprouting of the LECs was promoted in the presence of VEFG-C, in comparison with the control where the migration was entirely stimulated by the MDA-MB-231 cells. Of importance, a number of recent studies have investigated using animal tumor models and human cancer, that the VEFG-C/VEGFR-3 and VEGF-D/VEGFR3 axes are considered a major driver of tumor lymphangiogenesis and metastatic spread, and the expressions of VEFG-C, VEFG-D, and VEGFR-3 positively correlate with lymphatic metastasis of tumors [36]. For example, prostaglandins can promote lymphangiogenesis in cancer through VEGF- C [37]. VEGF-C overexpression leads to a subsequent increase in lymphangiogenesis, and a higher rate of lymphovascular invasion has been shown to worsen breast cancer prognosis [38]. Moreover, it has been suggested that unbalanced VEGF production is important in breast carcinogenesis [39]. Interestingly, our results further revealed that, besides the fact that the sprouting of the LECs in the peritumoral microenvironment was increased with the VEGF-C treatment, the MDA-MB-231 breast tumor cells also became attracted to the vicinity of the LECs where they exhibited enhanced spreading, in the presence of VEGF-C stimulation. Therefore, with this sacrificial bioprinting strategy, a volumetric lymphatically vascularized cell-laden breast tumor tissue model could be conveniently constructed to allow multi-week-long culture, as well as to partially mimic the interactions between lymphangiogenesis and tumor progression.

5. Conclusion

The generation of new lymphatic vessels via lymphangiogenesis is believed to be a critical step for the cancer metastasis, and anti-lymphangiogenic approaches for cancer therapy are potentially effective [40]. In our study, we modeled a 3D in vitro platform of the LV-BC model through sacrificial bioprinting, for studying the interplay between the lymphatic vessels and the breast tumor. LECs and MDA-MB-231 breast tumor cells exhibited several characteristic behaviors potentially similar to those processes of lymphangiogenesis and tumor invasion in vivo. Moreover, we also reproduced the importance of VEFG-C in stimulating the sprouting of lymphatic vessels and tumor cell migration, consistent with previous studies [36–39]. While the report is still preliminary as we primarily focused on the use of the sacrificial bioprinting method for generating this LV- BC model, ultimately our technology is anticipated to become a unique platform for not only fundamental mechanistic studies but also therapeutics screening under physiologically relevant 3D culture conditions, with possible extension to a variety of tissue types beyond (breast) tumor. For example, a prototype higher-throughput platform hosting 28 of our LV- BC models is illustrated in Fig. S1.