A Novel in vivo Rat Mesentery Model for Studying Tumor Spheroid-Induced Microvascular Remodeling

The tumor microenvironment includes cancer cells, blood vessels, lymphatic vessels, extracellular matrix, and numerous other cell types. There continues to be growing interest in understanding and targeting this tumor stroma and in its interactions with cancer cells for the treatment of solid malignancies. The challenge to investigate cell dynamics within this complex environment has led to the development of in vitro and ex vivo biomimetic models. A limitation for most models, however, is their physiological relevance. Inspired by this challenge, our laboratory recently introduced an ex vivo biomimetic model that integrates tumor spheroids and cultured rat mesentery tissues [1]. The model enables real-time observation of cancer cells, blood vessels, and lymphatic vessels. The application of this model to study lymphatic/blood vessel remodeling led to the discoveries of tumor spheroid-associated lymphatic/blood vessel connections and mispatterning [1]. These discoveries exemplify the potential view that tissue engineering-inspired models can provide for guiding in vivo observations. While inherent differences in modeling prevent direct data comparisons, a first step toward determining the physiological relevance of vascular dynamics seen ex vivo is benchmarking against and developing new in vivo models that provide novel insights into tumor vasculature development.

Despite all of the benefits provided by maintaining the tissue as an intact system ex vivo, there are some key characteristics that ex vivo modeling cannot fully capture. Blood flow, for example, plays a significant role in cancer growth, metastasis, vessel assembly, and network formation [2, 3]. While endothelial cells in rat mesenteric vessels have been shown to maintain their phenotype and smooth muscle cells maintain their ability to constrict during culture [4], it is possible that the lack of perfusion, coupled with the introduction of tumor spheroids, may result in a gap in complexity that could obscure a complete view of tumor-induced vessel plasticity. This raises the question whether the ex vivo tumor spheroid-rat mesentery tissue culture model adequately mimics an in vivo system. The validation of pathophysiological relevance for any model necessitates comparative studies between in vitro and in vivo systems [3, 5]. The difficulty, in biomimetic model development research is designing studies to make those comparisons, as the complexity gap between in vitro and in vivo is usually too large to realize meaningful results. Motivated by this limitation, we attempt to address the challenge by replicating, in a new in vivo model, key experimental design details of the ex vivo tumor spheroid-rat mesentery model, including the tissue type, cancer cell type, the use of tumor spheroids, and timeframe.

The objective of this work was to develop and characterize an in vivo model to determine the effects of tumor spheroid transplantation on microvascular remodeling and lymphatic/blood vessel plasticity. In this study, we transplanted tumor spheroids formed from H1299 (lymph node metastasis-derived human non-small cell lung carcinoma) cells onto rat mesenteric tissues and harvested tissues 3–5 days post-transplantation. We found that during the 3–5-day in vivo time course tumor spheroids remained adhered to the mesentery and were observed near blood vasculature. Compared to tissues from sham animals, in which the surgery process was performed but sterile saline was applied rather than tumor spheroids, H1299 tumor spheroid-transplanted tissues exhibited increased vascular density and sprouting. Vessel tortuosity and diameter were also noted as well as altered junctional patterning of blood vasculature. We also report the discovery that in vivo transplantation of tumor spheroids can induce lymphatic/blood vessel plasticity in the tumor microenvironment and may play a role in the apparent disappearance of lymphatic vasculature seen by 5 days post-transplantation. Instances of plasticity include examples of connections and gaps in lymphatic patterning as well as new structures such as lymphatic marker-negative protrusions growing from lymphatic marker-positive vessels. Our results offer new insights about how cancer cells modulate microvascular structure and function in the tumor microenvironment.

All cell lines were purchased from American Type Culture Collection (Manassas, VA, USA). The human-derived non-small cell lung cancer cell line H1299 (CRL-5803) was cultured in RPMI 1640 (Corning, Corning, NY) supplemented with 10% v/v heat-inactivated fetal bovine serum (FBS; Gibco, Grand Island, NY), 1% v/v penicillin-streptomycin (Gibco, Grand Island, NY), and 1% v/v l-glutamine (Gibco, Grand Island, NY). Cells were maintained at 37°C and 5% CO₂. Tumor spheroids were formed using a hanging drop culture system as previously described [1, 6]. Cancer cells were labeled with 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI; Thermo Fisher Scientific, Waltham, MA) or lysine-fixable chloromethylbenzamido (CellTracker™ CM-DiI; Thermo Fisher Scientific, Waltham, MA) diluted in phosphate-buffered saline (PBS; Gibco, Grand Island, NY) for 30 min at 37°C and 5% CO₂. CM-DiI staining was used to identify cancer cells post-fixation and collect qualitative images concerning their positioning relative to blood vasculature. Tissues with CM-DiI-stained spheroids required paraformaldehyde (PFA) fixation. For all other metrics, spheroids were prepared with DiI labeling, which could be used to confirm cell adherence after harvesting but faded after fixation. The use of DiI rather than CM-DiI was useful due to the fact that (1) the fading of DiI after fixation allowed for staining of another biomarker that had the same emission as DiI and (2) DiI staining allowed for methanol fixation which produced clearer images when compared to PFA fixation. DiI-labeled cells were then washed 3 times in warmed PBS. Both DiI-labeled and unlabeled cancer cells were resuspended in warmed minimum essential media (MEM; Gibco, Grand Island, NY) supplemented with 10% FBS at 2.86 × 10⁵ cells/mL. A total of 25, 30-µL cell-laden droplets were pipetted on the lid of the petri dish, the lid was inverted, and PBS was added to the base of the petri dish to prevent evaporation. The hanging droplets were cultured at 37°C and 5% CO₂ for 4 days to form tumor spheroids before harvesting of rat mesenteric tissues.

After 4 days of culture, tumor spheroids were washed off the petri dish lids using warmed MEM and collected in a 50-mL conical tube. After all spheroids had been collected, tumor spheroids were washed 3 times using MEM to remove any remaining FBS. All animal experiments were approved by the University of Florida’s Institutional Animal Care and Use Committee. Animals were housed on corn-cob bedding. To seed tumor spheroid, adult male Wistar rats (370–390 g; Envigo, Indianapolis, IN) were anesthetized via an intraperitoneal injection of ketamine (80 mg/kg body weight) and xylazine (8 mg/kg body weight). The abdominal area was then shaved and sterilized with 70% isopropanol and iodine. Analgesic injections of Ethiqa XR (0.65 mg/kg) and bupivacaine (2 mg/kg) were administered subcutaneously on the dorsal side and near the incision site, respectively. After an incision was made on the linea alba, a random section of mesentery was exteriorized using cotton tip applicators and placed onto a plastic stage. Mesentery consists of 20–25 thin windows which are connected to the intestinal tract and form a line of tissues, each approximately 5–10 mm wide, spanning from the jejunum to the ileum. Five to seven neighboring tissues were randomly selected, and 7-0 monofilament polyamide sutures (Ethicon, Raritan, NJ) were used to mark the first and last tissue. Approximately 20 tumor spheroids (600 µm in diameter) were pipetted onto each tissue randomly and spread out so that they were as homogenously dispersed as possible across each tissue. Excess media was removed to ensure spheroid adherence, and tissues were left exteriorized for 20 min. After 20 min, tumor spheroid tissues were then placed back into the peritoneal cavity. The muscle layer was sutured back together using sterile 5-0 monofilament polydioxanone sutures (Ethicon, Raritan, NJ), and the skin layer was sutured back together using 3-0 monofilament polydioxanone sutures (Ethicon, Raritan, NJ). Rats were then given meloxicam (5 mg/kg) and 5.0 mL of fluids subcutaneously on their dorsal sides during postoperative care before being placed in a fresh cage on a heating pad for monitoring until consciousness was regained. Meloxicam injections were administered every 24 h for 72 h, and Ethiqa injections were administered every 72 h. In sham rats, this procedure was repeated, but sterile saline was deposited onto the mesenteric tissues instead of tumor spheroids.

Rat mesenteric tissues were harvested at 3 and 5 days according to previous laboratory protocols [7, 8]. To harvest rat mesenteric tissues, adult male Wistar rats (350–379 g) were anesthetized via an intraperitoneal injection of ketamine (80 mg/kg body weight) and xylazine (8 mg/kg body weight). The abdominal area was then shaved and sterilized with 70% isopropanol and iodine. After an incision was made on the linea alba, the cecum was exteriorized using cotton tip applicators and placed onto a plastic stage. The 7-0 sutures were located, and the continuous sequence of tissues bordered by the sutures was excised and mounted onto microscope slides. Post-tissue harvesting, all rats were euthanized with a 0.2-mL intracardiac injection of Beuthanasia.

To confirm junctional analysis, an additional dextran injection study was conducted. In 1 rat, perfusable vessels were identified in day 3 rats via the injection of 1 mL of lysine-fixable 500 kDa FITC-dextran (10 mg per 1 mL sterile saline; Invitrogen; Carlsbad, CA) into the femoral vein of an anesthetized rat. Following the injection of dextran, rats were euthanized with a 0.2-mL injection of Beuthanasia and tissues were fixed for 10 min by filling the peritoneal cavity with 4% PFA. Tissues were then excised, sealed, and preserved in a 50:50 glycerol-PBS solution.

After tissues were mounted onto microscope slides, DiI-positive cell clusters and tissues were imaged before being fixed in 100% methanol for 30 min at −20°C and then washed 3 times for 10 min each in PBS + 0.1% saponin (Sigma‐Aldrich, St. Louis, MO). For tissues seeded with CM-DiI, tissues were whole mounted onto slides, fixed in 4% PFA for 10 min at room temperature, and then washed 3 times for 10 min each in PBS + 0.1% saponin (Sigma‐Aldrich, St. Louis, MO). Tissues were then labeled to identify platelet endothelial cell adhesion marker (PECAM; CD31) and lymphatic vessel endothelial hyaluronan receptor-1 (LYVE-1). All antibodies were prepared in a PBS solution supplemented with 0.1% saponin and 2% bovine serum albumin (Jackson ImmunoResearch Laboratories, West Grove, PA). Tissues were washed 3 times for 10 min each with PBS + 0.1% saponin between each 1-h-long incubation at room temperature. Following the final antibody incubation, tissues were washed 3 times for 10 min each with PBS + 0.1% saponin before being sealed and preserved in a 50:50 glycerol-PBS solution.

PECAM/LYVE-1: Tissues were incubated for 1 h with 1:200 biotinylated mouse anti-rat PECAM primary antibody (BD Pharmingen, San Diego, CA), 1:100 rabbit polyclonal LYVE-1 primary antibody (AngioBio, Del Mar, CA), and 5% normal goat serum (Jackson ImmunoResearch Laboratories, West Grove, PA). Following the primary stain, tissues were then incubated for 1 h with 1:500 Strep-CY2 secondary antibody, 1:100 GAR-CY3 secondary antibody (Jackson ImmunoResearch, West Grove, PA), and 5% normal goat serum (Jackson ImmunoResearch, West Grove, PA).

Images were acquired using 4× (dry, NA = 0.2), 10× (dry, NA = 0.45), 20× (dry, NA = 0.75), and 40× (oil, NA = 1.3) objectives on an inverted microscope (Nikon Eclipse Ti2) coupled with an Andor Zyla sCMOS camera. Confocal microscopy maximum intensity projection images of cancer cells localizing near vessels and lymphatic/blood vessel connections were captured using the 63× (oil, NA = 1.4) objective on a Zeiss LSM 880 coupled with an Axio Observer microscope.

Tissues without lymphatic vasculature were excluded from this study as not all mesenteric tissues have lymphatic vessels, and as we primarily care about interactions between cancer cells, blood vessel, and lymphatic vessels, we felt it was important to have all 3 structures in all conditions. This helped increase consistency in responses between tissues, as previous work in our laboratory has shown that the presence of lymphatics can attenuate blood capillary sprouting [9]. For the angiogenesis metrics, tissues were harvested 3 days post-surgery and divided into 2 groups: (a) sham tissues (n = 8 tissues from 4 rats) and (b) tissues seeded with H1299 tumor spheroids (n = 8 tissues from 4 rats). Day 3 was chosen as a timepoint to measure angiogenic growth due to the observation in previous in vivo rat mesentery exteriorization studies from our laboratory which show angiogenesis peaks between 3 and 5 days post-stimulation [10]. To quantify angiogenesis, vascularized tissue areas were imaged and then randomly selected for analysis. For both groups, roughly half of the images originated from internal areas within a vascularized network while the rest were taken from areas closer to the edge of a network, ensuring a full range of networks was represented in the random selection. The number of sprouts, defined as a blind-ended PECAM-positive endothelial cell extension off a blood vessel, and the number of segments, defined as the length of blood vessel between two nodes, was calculated per 4× field of view (FOV) and normalized by the total vascularized area in the FOV of each image.

To quantify venule tortuosity, junction length, and diameters, the largest venule on each tissue analyzed for angiogenesis quantification was imaged at the tissue edge. Eight images were randomly selected for analysis from each group. Venule tortuosity was measured for each segment in the FOV and then averaged per image in (a) sham tissues harvested 3 days post-surgery (n = 8 tissues from 3 rats) and (b) H1299 tissues harvested 3 days post-implantation (n = 8 tissues from 4 rats). Tortuosity was calculated by dividing the length of the curved segment of venule by the length of the straight-line distance from branchpoint to branchpoint for each segment.

To quantify junctional length, the area of the venule with the least distinct junctional patterning was selected per image. A box approximately 200 μm in length was drawn along the venule at this point, and the total length of observable junctions was calculated within this area. Junctional length/area was measured in (a) sham tissues harvested 3 days post-surgery (n = 8 tissues from 3 rats) and (b) H1299 tissues harvested 3 days post-implantation (n = 8 tissues from 4 rats).

For quantifying venule diameters, the largest diameter measurement for the venules in the randomly selected images was taken per image. Venule diameter was measured in (a) unstimulated tissues harvested on day 0 (n = 8 from 2 rats), (b) sham tissues harvested 3 days post-surgery (n = 8 tissues from 3 rats), and (c) H1299 tissues harvested 3 days post-implantation (n = 8 tissues from 4 rats).

For the quantification of the percentage of tissues with lymphatic vessels, (a) unstimulated tissues harvested on day 0 (n = 16 from 2 rats), (b) sham tissues harvested 5 days post-surgery (n = 11 from 2 rats), and (c) H1299 tissues harvested 5 days post-implantation (n = 17 from 2 rats) were all compared. Day 5 was chosen as a timepoint to measure metrics related to lymphatic vessels and lymphatic growth due to previous observations in our laboratory that lymphatic growth in rat mesentery tends to lag angiogenic growth in response to both VEGF-C and 48–80 stimulation [9, 11]. Presence of lymphatics was observed after lymphatic staining and recorded for each tissue.

Lymphatic protrusions were defined as multicellular structures bulging from LYVE-1-positive lymphatic vessel segments with distinct PECAM-positive junctional labeling and were also characterized by indistinct or absent lymphatic marker coverage. To quantify the number of lymphatic protrusions per tissue, sham and H1299 tissues harvested 3 days post-implantation were compared with sham and H1299 tissues harvested 5 days post-implantation. Lymphatic protrusions were found only on tissues containing lymphatic vessels, so tissues without lymphatic were excluded from further analysis. Among tissues with lymphatics, the same four groups were analyzed for number of protrusions: (a) sham tissues harvested 3 days post-surgery (n = 17 tissues from 4 rats), (b) H1299 tissues harvested 3 days post-implantation (n = 12 tissues from 4 rats), (c) sham tissues harvested 5 days post-surgery (n = 4 tissues from 2 rats), and (d) H1299 tissues harvested 5 days post-implantation (n = 9 tissues from 3 rats).

For all studies, analysis of statistical significance was determined using Prism software (GraphPad Software, La Jolla, CA). In the angiogenesis studies, comparison of the sprouts and segments per vascularized area at day 3 (t = 72 h) was made using a two-tailed, unpaired Welch’s t test. For tortuosity and junctional length/area, comparisons were made at day 3 (t = 72 h) using a two-tailed, unpaired Welch’s t test. For diameter measurements, comparison of the average venule diameter was made at day 3 (t = 72 h) using an ordinary one-way ANOVA followed by a Tukey’s post hoc test for multiple comparisons. For the quantification of percentage of tissues with lymphatic vessels, statistical significance was calculated using a Kruskal-Wallis test (one-way ANOVA on ranks) followed by a Dunn’s test for multiple comparisons. For the quantification of average number of lymphatic protrusions per tissues, the ROUT method (robust regression and outlier removal) was used to confirm the presence of an outlier in the data set (Q = 0.1%). Upon being identified as an outlier, 1 datum point from group d (H1299 tissues harvested 5 days post-implantation) was removed. Statistical significance was calculated using a mixed effects analysis followed by a Fisher’s LSD test. Values for all graphs are presented as mean ± standard error of the mean.

Imaging of tumor spheroid-seeded mesenteric tissues prefixation revealed that DiI-positive H1299 cancer cells can be observed on harvested rat mesenteric tissues 3 days after implantation (Fig. 1a). Cells formed high-density cell clusters as well as spread as individual cells throughout the tissues, suggesting migration from the initial spheroidal configuration had occurred (Fig. 1b). Cancer cells were also observed on tissues 5 days after implantation (data not shown). PECAM staining also confirmed the presence of DiI+ cancer cells in close proximity to native vasculature (Fig. 1c). Cancer cells were also seen in close proximity to vessels, as confirmed by submicron confocal imaging, which revealed the presence of cancer cells aligning along blood vasculature (Fig. 1d–f).

Evidence of cancer cell effects on angiogenesis was supported by increased sprouting, vascular density, and tortuosity in H1299 tissues harvested 3 days post-implantation compared to sham tissues harvested 3 days post-surgery (Fig. 2a, b). Quantification showed a significant increase in the number of sprouts and segments (per vascular area 3 days post-implantation) with tumor spheroids compared to sprouts and segments (per vascular area on sham tissues harvested 3 days post-surgery) (Fig. 2c, d). A significant increase in venule tortuosity ratio was also observed in tissues harvested 3 days post-implantation with tumor spheroids compared to the tortuosity ratio of sham tissues harvested 3 days post-surgery (Fig. 2e).

Immunohistochemical staining of the intercellular junctions on blood vessels, particularly venules, also revealed transitional junctional morphology over the experimental time course. Junctions on venules in sham tissue harvested 3 days post-surgery (Fig. 3a) were clear and typical of normal rat mesenteric vasculature, while venules in tissue implanted with H1299 tumor spheroids exhibited were less distinct and, in some cases, completely absent (Fig. 3b). Quantification of the junctional length/area revealed that venules in tissues harvested 3 days post-implantation with tumor spheroids had significantly fewer junctions compared to venules in sham tissues harvested 3 days post-surgery (Fig. 3c). Perfusion data also support tumor-induced junctional degradation. Perfusion with FITC-conjugated lysine-fixable 500 kDa dextran of rat mesenteric tissues seeded with tumor spheroids revealed the presence of cancer cells in close proximity to patent, perfused vasculature. Perfusate was not only confined to blood vasculature but also observed to evenly spread across the tissues (Fig. 3d). Tissues neighboring the set of tissues implanted with tumor spheroids were observed to still have some cancer cell presence and characteristic of tumor vasculature, points at which leakage of dextran from vessels into the extracellular space was also observed, though the leakage was not as extensive as was seen in tumor spheroid tissues (Fig. 3e). Tissues harvested approximately 4–5 tissues down the mesenteric line from the set of tissues implanted with tumor spheroids exhibited very minimal to no cancer cell presence and no examples of leakage of the dextran perfusate from blood vasculature (Fig. 3f).

Examples of lymphatic/blood vessel plasticity were also seen in tissues seeded with tumor spheroids. Similar to the results seen in our previous ex vivo studies, connections between blood and lymphatic vessels were observed to form, noticeably by 5 days post-implantation with tumor spheroids (Fig. 4a). Often at connection points, blood vasculature changed morphology and was observed wrapping around lymphatic vasculature (Fig. 4a, b). Confocal microscopy revealed the colocalization of the PECAM blood vessel marker with the LYVE-1 lymphatic vessel marker (Fig. 4b, c). While connections were seen throughout tissues, the frequency with which they appeared was difficult to determine due to diminished presence of lymphatic vessels on tumor spheroid tissues by 5 days post-implantation. While the light PECAM-positive marker coverage was still observable on lymphatics at this time point (Fig. 4d, e), the LYVE-1 lymphatic marker was noticeably difficult to differentiate from dense interstitial cells. Both gaps in lymphatic patterning as well as full areas where the vessels were indistinguishable from the surrounding cells were commonly seen in tumor spheroid tissues 5 days post-implantation (Fig. 4f). Conversely, while lymphatic vessels in day 0 tissues do maintain their light PECAM coverage (Fig. 4g, h), the exhibited LYVE-1 marker coverage is clearly distinct from both the interstitial space and the surrounding interstitial cells (Fig. 4i). The pervasive nature of the LYVE-1 marker coverage was present in all groups that were harvested 3–5 days post-surgery regardless of the presence of cancer, implying that the wound healing response may be affecting LYVE-1 marker presence. Still, future studies are needed to determine whether cancer cells co-label for LYVE-1.

This loss of lymphatic phenotype and potential disappearance of lymphatic vessels was also supported by observations in sham tissues. Quantification of the number of tissues with lymphatic vessels revealed a significant decrease in the percentage of sham tissues harvested 5 days post-surgery compared to the percentage of day 0 tissues harvested immediately post-euthanization (Fig. 5a). This implies the loss of lymphatic vasculature may be, in part, a surgery-induced wound healing response. However, the presence of novel lymphatic structures supports that the cancer cells could be inducing lymphatic remodeling. Unlike in analogous ex vivo studies, multicellular structures connected to and bulging from normal, elongated, LYVE-1-positive lymphatic vessel segments with distinct PECAM-positive junctional labeling were shown to form in vivo post-implantation of tumor spheroids (Fig. 5b). Quantification of these lymphatic protrusions shows there were significantly more protrusions in tumor spheroid tissues harvested 5 days post-implantation compared to protrusions on sham tissues harvested 5 days post-surgery (Fig. 5c). Additionally, the number of protrusions significantly increased on tumor spheroid tissues harvested 5 days post-implantation compared to tumor spheroid tissues harvested 3 days post-implantation, highlighting the presence of these protrusions may be both time and cancer dependent (Fig. 5c). Notably, while PECAM labeling was present in every protrusion (Fig. 5d, e), lymphatic markers such as LYVE-1 were consistently less distinct and, in most cases, absent from bleb structure (Fig. 5e, f), highlighting the potential for lymphatic/blood vessel plasticity in the tumor microenvironment.

The primary goal of this study was to develop and characterize an in vivo platform integrating tumor spheroids to study cancer-induced vascular remodeling and coordination of blood and lymphatic vasculature. We found that H1299 tumor spheroids transplanted into Wistar rats induced angiogenesis and increased tortuosity and vascular remodeling over a 3–5-day time course. We were also able to quantify a significant decrease in the total length of junctions per area in venules of tissues cultured with H1299 tumor spheroids compared to sham tissues. This when coupled with the observed permeability of tissues transplanted with spheroids implies a functional effect of tumor spheroid junctional degradation consistent with established knowledge that cancer vasculature exhibits increased leakage due to looser junctional connections and lack of vessel integrity [12]. Additionally, lymphatic vessel remodeling and vessel plasticity were also observed in vivo. However, this observation may be dependent on the stage of tumor growth. While the blood vessel marker common on lymphatics remained present, lymphatic vessels were observed to fade and lose their LYVE-1 lymphatic marker phenotype over the culture time course. Additionally, we observed protrusions with distinct PECAM-positive labeling, but indistinct or absent LYVE-1 lymphatic marker coverage connected LYVE-1-positive lymphatic vessels, a potential example indicative of plasticity not seen in previous ex vivo studies [1]. While these observations would be supported by future experiments focused on determining the functional significance and characterization of lymphatic protrusions, these novel observations emphasize the impact of biomimetic models for not only predicting in vivo responses, but also guiding the interpretation of results. For example, while interpretation of the in vivo-based observation of the irregularly shaped lymphatic vessel segments with enlarged diameters, multicellular bleb-like morphologies, and loss of LYVE-1 labeling alone might be perplexing, consideration of the mosaic lymphatic/blood vessel formations highlighted in our correlate ex vivo tumor spheroid-rat mesentery study, in which we first detailed the possibility of lymphatic/blood vessel connections and plasticity in the tumor microenvironment, provides context for what may be happening [1]. This and the previous discovery that tumor spheroids can induce gaps in lymphatic patterning also help contextualize the in vivo observations of disappearing or faded lymphatics and LYVE-1-negative, PECAM-positive protrusions as another, perhaps more dramatic, form of the lymphatic mispatterning and lymphatic/blood vessel plasticity.

In addition to lymphatic remodeling, tumor vasculature is commonly characterized by increased vessel leakiness, tortuosity, dilation, and angiogenesis [13]. This rapid growth of new vasculature, often without the aid of support mural cells, results in immature and abnormal vessel formation [14]. For expansive tumor growth, cancer cells must develop a supportive vascular network through the process of angiogenesis. However, unlike normal tissue blood vessels, the continuous angiogenesis stress exerted by cancer cells leads to tumor-associated blood vessels that are structurally abnormal and leaky: characteristics known to aid cancer cell invasion and metastasis. In this study, we qualitatively observed increased vessel diameters in the tumor spheroid tissues compared to the day 0 tissues. Analysis quantitatively compared large network feeding venules and showed a significant difference in average feeding venule diameter between the day 0 and day 3 tumor spheroid tissues (online suppl. Fig. 1a–c; for all online suppl. material, see https://doi.org/10.1159/000543011). The observed increase in venule diameters in the tumor spheroid tissues was also characteristic of sham tissues. Thus, the increased diameters could be attributed in association with sham-induced remodeling and additional studies are needed to evaluate the cancer cell-specific effects. The variability in microvascular network branching architecture per tissue was a challenge with this analysis. However, quantification of increased angiogenesis, tortuosity, junctional length, and the incorporation of the initial dextran perfusion study supports that our model can reproduce these expected characteristics in the tumor microenvironment. The original purpose of the perfusion study was also to characterize whether observed lymphatic/blood vessel connections were superficial or patent, though this was made difficult to determine due to the disappearance of junctions that may be causing leakage of dextran out of the blood vasculature across the entire tissue. A previous study done in our laboratory suggests that lymphatic/blood vessel connections are not patent [15]. This study, however, did not incorporate cancer cells motivating future studies aimed at discovering the functionality of lymphatic/blood vessel connections. While future experiments quantifying additional junctional markers and characterizing the functionality of lymphatic/blood vessel connections would be useful to further characterize cancer-induced vascular remodeling, confirmation of increased angiogenesis, tortuosity, and loss of junctional patterning in response to tumor spheroids helps verify that the in vivo tumor spheroid-rat mesentery model is a useful modeling approach and supports the novel observations of lymphatic remodeling in the tumor microenvironment.

The specific structures of, and changes to, lymphatic vessels in the tumor microenvironment remain under-characterized, highlighting the importance of better understanding of tumor-induced lymphatic vessel remodeling. Tumor cells are known to secrete angiogenic and lymphangiogenic growth factors, and depending on the study, lymphatic vessel remodeling is reported for both intra-tumoral and peri-tumoral regions [16]. Intra-tumoral vessels are characterized as nonfunctional and collapsed [17], and the extent of initial lymphatic vessel structural remodeling including network connectivity and patterning with blood vessels is unclear. Interestingly, increased lymphatic vessel density might not correlate with tumor progression [18] and some tumors do not even show evidence of lymphangiogenesis [19]. This variability can make studying vascular remodeling challenging, especially when visualization of intact lymphatic vasculature in the tumor microenvironment is not always easy, as the seemingly common approach is to immunolabel tumor tissue cross sections for lymphatic markers [20]. Tumor effect on lymphatic vessel function can also be identified using dye injections and tracking uptake, an approach which, like cross-sectional labeling, does not allow for a complete view of vessel morphology [21]. This model’s inclusion of intact lymphatic vasculature allows for a unique view of lymphatic vascular remodeling as shown by the presence of LYVE-1-negative protrusions on LYVE-1-positive lymphatic vessels, which cross-sectional labeling may not have been able to truly capture as a lymphatic remodeling phenomenon. Decreased lymphatic presence and apparent loss of lymphatic phenotype could also have interesting implications, especially when considering cases where sectioning data report is limited to no lymphatic vessel presence [19]. The loss of lymphatic phenotype and the mispatterning of remodeled structures such as protrusions reported in this study are indicative of lymphatic remodeling and potential lymphatic/blood vessel plasticity. The observations of gaps in LYVE-1 labeling along PECAM-positive vessels are consistent with our previous characterization of lymphatic vessels undergoing phenotypic transition [1, 22]. In the context of this study’s finding that sham and tumor spheroid-stimulated tissues have decreased lymphatic vessel presence, the observations suggest that the LYVE-1 labeling gaps might be associated with lymphatic vessel loss and highlight the potential for recharacterization of certain cancers previously classified as non-lymphangiogenic or lacking lymphatics and underscore the benefit of observing tumor-induced remodeling in a system with intact vasculature.

An additional contribution of this study is the introduction of another in vivo method for using the rat mesentery for characterization of angiogenesis and lymphangiogenesis responses. Rat mesentery tissue is 20–40 μm thick, making it ideal for culturing and imaging [23]. Rat mesentery tissue has also been used to gain fundamental knowledge of angiogenesis, lymphatic physiology, and microvascular remodeling. In the microvascular research space, the rat mesentery tissue model is an established ex vivo modeling system that benefits from a reproducible and often dramatic response to stimulus [10, 24, 25]. Previous studies support the relevance of the rat mesentery for characterizing cancer cell-associated effects. For example, Costa et al. [26] injected cancer cell-conditioned media intraperitoneally and reported the formation of increased angiogenesis on mesenteric tissues 10 days later and Nagel et al. [27] transplanted small pieces of subcutaneous tumors onto mesenteries as a model for investigating tumor growth. In another study, Yan et al. [28] used the rat mesentery to intravitally investigate cancer cell adhesion to postcapillary venules. In comparison with these examples, our results suggest that cancer cell spheroid transplantation onto mesenteries can be used as method for probing cancer cell-associated microvascular remodeling responses. Additionally, the use of rat mesentery to study metastatic spread of both ectopic and orthotopic tumor cells has also been reported on Yamagata et al. [29], for example, demonstrated in an orthotopic model, that fluorescently labeled rat hepatoma cancer cells in the peritoneal cavity can spread to mesenteric tissues. Our previous ex vivo modeling work also showed that metastasis-derived H1299 tumor spheroids exhibited increased spread when compared to cell spread seen in non-metastasis-derived A549 tumor spheroids [1]. These studies, as well as the demonstrated ability and observed cells elongating along blood vasculature in this model, support the use of the in vivo model introduced in here to study cancer metastasis. Expanding the number of cancer types that can be incorporated into this model could elevate the scope of this model in characterizing cancer dynamics beyond the cancer vascular-specific reported in this study.

Thus, a limitation of this current approach is the use of H1299 lung metastatic cells. Additionally, this current approach also incorporates human tumor cells onto rat mesenteric tissues. Rationale for the use of this cell type in this study is supported by the motivation to indirectly compare in vivo results to the results of the previous ex vivo modeling study [1]. However, we recognize that gastrointestinal or ovarian [30] cancers are more relevant for mesentery tissues. This organ and species mismatch raises the question: how relevant are the vascular dynamics we see in response to implantation of these tumor spheroids? Our previous ex vivo study was able to show blood vessels directly crawling around and enwrapping transplanted human tumor spheroids, and expected migratory dynamics were observed to still occur over the culture time course [1]. In this in vivo study, we have observed increased angiogenesis as well as instances of vessel dilation and tortuosity, key hallmarks of cancer vasculature. These findings suggest that this model can accurately replicate expected behaviors and dynamics, support the physiological relevance of the system, and may still be a useful model to discover novel dynamics in tumor-induced vascular growth and remodeling, especially in lymphatics, which are often understudied compared to blood vasculature. Like our model, xenograft models, in which human cells are often injected into rats or mice, are one of the most common in vivo approaches used to study human cancers. In these models, immunocompetent animals are used to avoid immune rejection, a potentially rapid process that can take effect within minutes or hours of implantation in the case of hyperacute rejection and between days and weeks in the case of acute rejection [31]. While use of immunosuppression is important, it is also possible that immunosuppressed animals may not be the ideal system to investigate vascular dynamics between blood and lymphatic vessels, seeing as the activated immune cells are believed to play a big role in vascular growth [32] and immunosuppression can affect lymphoid organ and vessel function [33, 34]. For the current study, we do not know the effect of using an immunocompetent rat. Based on our results and the model’s ability to mimic the tumor microvascular characteristics over a 3–5-day time course, we speculate that the related immune response is associated with later time points.

Our longer term goal is to establish the in vivo method as an approach for probing cancer cell effects and we are less focused on mesentery-specific responses. Rather, the mesentery represents an accessible host tissue upon which we can directly transplant cells, analogous to cell transplantation studies in an ear sponge model, the back-pack window chamber model, the chicken chorioallantoic membrane model, the cornea transplant model, or even model using omentum tissue [35‒39]. Common motivations for use of these models are the control of local cell delivery and imaging of intact vessels. Specific advantages for use of mesentery tissue include its thinness, which makes it beneficial to generate high fidelity images with single-cell resolution as well as its incorporation of numerous microvascular cell types and structures [3]. An additional advantage, supported by our characterization of remodeling effects, is that initial remodeling occurs over 3–5 days. This timeframe is consistent with the timeframes for the other models and more importantly is consistent with example in vivo tumor xenograft studies focused on understanding vascular patterning [40, 41].

In this study, we have developed an in vivo tumor spheroid rat mesentery model for the purpose of validating the discovery of lymphatic/blood vessel plasticity in the tumor microenvironment. Tumor spheroids implanted in vivo were shown to adhere to mesenteric tissues and induce angiogenic growth and remodeling. We have also confirmed the presence of cancer cells near perfused vasculature and highlighted how they can affect junctional marker phenotyping. This model allowed for the discovery of lymphatic/blood vessel plasticity and remodeling in the in vivo rat mesentery tumor microenvironment, including novel structures, such as LYVE-1-negative protrusions growing from LYVE-1-positive lymphatic vessels. Through the development of a new in vivo platform, this study provides novel insights into how cancer affects lymphatic/blood vessel plasticity and vessel coordination in the tumor microenvironment, introducing new avenues for further exploration in the field of cancer and vascular research.

We would also like to thank the members of Dr. Murfee’s Microvascular Dynamics Laboratory who helped make this work possible.

This study protocol was reviewed and approved by the University of Florida Animal Care and Use Committee (IACUC Protocol # 202110060).

The authors have no conflicts of interest to declare.

This project was supported by funding from the American Heart Association Grant 23TPA1142184 awarded to W.L.M. D.W.S. was supported in part by the National Cancer Institute of the National Institutes of Health under Award Number T32CA257923.

Arinola O. Lampejo and Walter L. Murfee were responsible for the transplantation experiments and postoperative care and contributed to method development and the interpretation of observations. Arinola O. Lampejo was also responsible for imaging, analysis, and writing of the results. Luciana Fonseca Perez and Miriam M. Girgis also contributed to the analysis of results. Walter L. Murfee oversaw the project and contributed to the method design, interpretation of the results, and writing. Dietmar W. Siemann and Blanka Sharma were involved in the writing and editing of the manuscript and also contributed to the conception of the work.

All data generated or analyzed during this study are included in this article and its online supplementary material files. Further inquiries can be directed to the corresponding author.