Due to its relative ease of synthesis, a decellularized scaffold was selected as the design method of choice. Not only does the decellularized matrix still contain key protein structures necessary for cell attachment and growth, but it also meets the size and shape requirements necessary to be re-implanted back into the mouse neck to evaluate responsiveness in vivo. As such, several decellularization protocols were assessed for their efficacy at removing cells and their ability to retain an intact ECM (data not shown). After optimization, the following protocol was established: cervical lymph nodes were removed from mice (averaging 4-6 lymph nodes per mouse) and washed in ice cold PBS to remove debris. The nodes were then transferred into 0.075% SDS in PBS and placed into an incubator with a rotating tube rack, and incubated for 16 hours. Afterwards, the decellularized nodes were removed and washed for one hour with four wash changes in sterile ice cold PBS (Figure 1). From histology images stained with hematoxylin and eosin (H&E), it is clear that the decellularization process removed the vast majority of nucleated cells from the scaffold, as evidenced by the lack of stained nuclei and the loss of distinguishable follicles (regions of higher cell density where cell maturation and priming takes place) (Figure 2). This result is further supported by a quantitative assay, in which the amount of dsDNA (a metric for number of cells) present in the decellularized scaffold was significantly reduced compared to a native lymph node, and falls below the industry standard of 50 ng dsDNA/mg ECM for decellularized tissue. Further decellularizations of lymph nodes at a wide range of sizes shows that lymph nodes with a minimum feret diameter under 2mm (2000 um) were consistently completely decellularized (Figure 3A and 3B). Lymph nodes with a minimum feret diameter of approximately 2 mm (2000 um) or above retain a cellular core indicative of incomplete decellularization, as shown through histological analysis (Figure 3C).

Sulfated glycosaminoglycans (sGAGs) are ECM proteins that play an essential role in cell-ECM binding and chemical signaling. As such, a functional lymph node scaffold must retain these important proteins after decellularization. Alcian Blue staining for sulfated GAGs was completed for native and decellularized lymph node tissue under an initial wash protocol (Figure 4A and 4B) and for updated wash protocol (Figure 4C and 4D). Due to the use of a sulfate-containing detergent (SDS) for decellularization, the initial measurements were overly saturated, indicating incomplete washing of the detergent post-decellularization. However, after updates to the wash protocol were made, the normalized ratio of Alcian Blue staining between native and decellularized tissue was approximately 1, indicating that the SDS was properly removed from the tissue and that sGAG proteins were retained after decellularization (Figure 4E).

Together, these data suggest that the optimized decellularization and wash protocols allowed for the successful creation of a dECM scaffold that can be used in later experiments to create a functional artificial lymph node.

In order to successfully perform an antigen-specific immune response, the dECM scaffold must contain cells that are able to take up surrounding antigen and present it to antigen-specific immune cells. As such, several methods of seeding bone marrow derived dendritic cells onto the scaffold were assessed for their ability to successfully enter the scaffold. In brief, the following methods were assessed: injection, in which 1 million cells were resuspended in 10uL and injected 3 times in 3 separate places throughout the scaffold; centrifugation, in which 3 million cells were resuspended in 100uL and centrifuged alongside the scaffold at 2500rpm for 2 minutes repeated 5 times; the Prellis drip method, in which 3 million cells were resuspended in 20uL and dripped over the top of the dry scaffold; and the swirl method, in which 3 million cells in 2 mL were swirled around the scaffold (Figure 5). It was initially hypothesized that the swirl method would provide the least amount of shear forces and stress to the cells, but that the injection method would be most successful. However, based on the DNA quantification assay, the centrifugation method was selected as the most successful, likely due to the shearing of cells upon injection and the inability of other methods to fully access the interior of the scaffold. However, despite these successes, a notable loss of scaffold mass is present. For a non-seeded dECM scaffold in culture, approximately half of the protein scaffold mass is lost after 3 days. This is likely due to protein breakdown, as there are no longer cells in place to rebuild the proteins as they naturally degrade. However, it may pose a significant challenge to the efforts to seed a decellularized scaffold, allow for APC integration, and re-implant in vivo before the scaffold fully degrades. More work is needed to fully characterize this scaffold breakdown to ensure the device is able to function properly.

In order to evaluate the efficacy of the seeded lymph node, the antigen-specific priming and corresponding cytotoxic efficacy was evaluated in vitro. Initial efforts to achieve this response were unsuccessful (Figure 6A). Naive antigen presenting cells were seeded into the lymph node using the centrifugation method detailed above, before being activated with CpG (a TLR9 agonist) and pulsed with the model antigen ovalbumin (OVA) to present. These seeded nodes were placed into a 6 well plate and co-cultured with OT-1 splenocytes, which are genetically modified to contain TCRs that recognize the OVA antigen. If the OT-1 CD8+ T cells were primed successfully, they would induce apoptosis in the B16-OVA cells which display the OVA antigen, as measured through a decline in %RFU compared to the B16-OVA controls. However, this priming was unsuccessful, likely due to mass transfer challenges associated with activation of the APCs and trafficking of cytotoxic T cells into the scaffold to allow priming to occur. To counteract this, a second experiment was conducted, in which the dendritic cells were activated prior to seeding, and the nodes and OT-1 immune cells were co-cultured within a 96-well plate.

This adjusted method yielded exciting results (Figure 6B). The seeded lymph node successfully primed the CD8+ T cells to induce an antigen-specific kill response (p<0.0001). Not only did the seeded scaffold successfully facilitate this immune response, but it also did so with the same efficacy as plate-bound activated dendritic cells, indicating that the seeded node is as effective at priming immune cells as the standardized in vitro method.

Based on the visualizations alone, the wild type mouse model (Figure 7A) appears to exhibit the highest amount of surface area compared to the surgery (Figure 7B) and tumor-bearing models (Figure 7C). Its network of connected lymphatic vessels also demonstrate a stronger connectivity than both the cancerous model, whose lymph nodes are abnormally enlarged, and the surgical model, whose lymph nodes have been removed as part of the SOC procedure for HNSCC patients. Additionally, whereas the tumor-bearing model lacks much more lymphatic vasculature in the magnified region of interest, where we approximated to be behind the tongue, both the wild type and surgical models possess vasculature that appears to spread throughout this region and further into the neck cavity (Figure 8B).

From the visualizations, we hypothesized that the wild type model would have the largest volume of lymphatic vasculature overall, while the neck dissection could come second, and the tumor-bearing model would have the least amount of vasculature throughout the cavity. We expected this hypothesis to also align with each model’s corresponding region of interest. Interestingly, the vessel count, average volume per vessel, and total volume of stained vessels measured by IMARIS for the entire head and neck cavity (Figure 8A) contradict our initial observations, while the volumetric analysis performed on the normalized region of interest in each model supported the hypothesis (Figure 8B). As shown in Figure 8A, the tumor-bearing model possesses the largest average size of lymphatic vessels, while the neck dissection model exhibits a significantly smaller average volume per vessel compared to both the wild type and cancerous models. However, both the neck dissection model and tumor-bearing models contained higher volumes of stained vessels than the wild type. Furthermore, the neck dissected mouse surpassed both other models in total vessel count.

In contrast to the volumetric analysis performed on each model as a whole, the measurements extracted from the region of interest behind the tongue in each model are consistent with each other and our initial belief (Figure 8B). The wild type mouse exhibits the highest values in vessel count, average volume per vessel, and total volume of stained vessels. The neck dissected mouse trails behind relatively close, compared to the significant difference seen in the tumor-bearing model in each characteristic.

This lymph node scaffold was created by decellularizing a native lymph node with sodium dodecyl sulfate (SDS). These experiments demonstrated the efficacy of this method at retaining native ECM proteins (including sulfated GAGs) and providing an optimum environment for seeded cells, as demonstrated through histology analysis and DNA quantification (Figures 1, 2 4, and 5). However, this method will be extremely difficult to scale up, as it requires the use of harvested cervical lymph nodes to create the decellularized scaffold. In addition, there is evidence that the scaffold quickly breaks down while in culture (assessed by an over 50% reduction in scaffold mass after 3 days in culture). As such, further work must be done to create a scaffold that retains the desired properties of this method while improving longevity and ability to scale up for future manufacturing. Furthermore, testing has shown that lymph nodes with a minimum feret diameter of 2 mm or greater did not decellularize completely (Figure 3). This suggests that the size of the lymph node prevents the SDS from fully permeating the node and decellularizing the center. As such, only lymph nodes that fall under the 2 mm minimum feret diameter threshold should be decellularized using this protocol. In scaling up, larger lymph nodes may be used to meet demands, but the protocol for decellularizing them must be reoptimized to do so.

Although early in vitro functional tests yield exciting results, they highlight potential translational challenges for this device. Initial efforts to facilitate immune cell priming and a corresponding antigen-specific response were unsuccessful. We hypothesized this was due to mass transfer limitations, and altered the experimental design to better facilitate cell-cell contact (ie, priming DCs before seeding, co-culturing in 96 well instead of 6 well plate, and increasing density of lymphocytes), yielding successful results. From the data generated, it is clear that the seeded artificial lymph node facilitated an effective immune response (p<0.0001) (Figure 6). However, this mass transfer barrier will exist in vivo, and may be more challenging to overcome. This issue emphasizes the need for reconnection to the native lymphatic vasculature, which can provide flow through the scaffold to facilitate the desired response.

The creation of a medical device for use in patients requires exhaustive safety assurances to provide patients with the most effective care possible and limit the risk of potential adverse effects. It is of primary importance from an ethical standpoint to fully ensure the safety of a device before implementation in patients. As such, this necessarily constrains the speed at which a device can be developed, as rigorous safety checks must be performed. This requirement limits our characterization of the efficacy of this device, as we did not have the time or resources to conduct the safety tests required for movement into an in vivo model. Although a pressing health need exists for devices capable of immunomodulation, the safety of such a device must be carefully studied and fully understood before it can be translated into the clinic.

Improving our understanding of lymphatic vasculature in animal models is essential for such future experiments, as one way to measure the efficacy of lymph nodes could be to analyze its ability to reconnect to original lymphatic vasculature. The imaging software IMARIS has yielded unique results regarding the status of lymphatic vasculature when comparing a wild-type mouse to one that underwent surgery and one that bears a tumor. Although we did not have enough individuals to collect statistically significant data, there were some clear overall patterns between the three types of mice. For one, while there were more vessels in the neck dissection mouse compared to the other two mice, the average volume per vessel is clearly smaller than the other two mice (Figure 8A). This suggests that the simple removal of the lymph node would lead to a breakdown of the normal connectivity of the lymphatic vasculature in the middle, leaving smaller broken pieces of the original network behind. Also, it is clear that as a whole, there are larger vessels in the tumor-bearing mouse as compared to the wild-type when looking at the whole head and neck region (Figure 8A). This is interesting because qualitatively, the images for the tumor-bearing mouse seem to be lacking in lymphatic vasculature when compared to that of the wild-type and the neck dissection mice.

To explain this contradiction with our initial hypothesis when observing the whole volumes compared to the volumes of interest and comparing it to the quantitative data collected for the entire head and neck cavity, a subsequent literature search was performed. We discovered that tumor growth naturally necessitates lymphangiogenesis, or the formation of new lymphatic vessels, to improve circulation rates and promote metastasis, while simultaneously aiming to hinder normal lymphatic drainage in the lymph nodes.³⁷ Hence, there is a lack of lymphatic vasculature throughout the head and neck cavity and within the region of interest in the tumor-bearing model (Figure 7C). Furthermore, one study stated that lymphangiogenesis of new lymphatic vessels is generally a late-stage tumor phenotype as the tumor prepares to metastasize.³⁸ In other words, new lymphatic vessels leading to the tumor site for a budding tumor would introduce a host of new immune cells that can recognize its non-normal phenotype and target it for degradation, inhibiting its growth potential. Rather, what appears to be happening in these early tumor models is that the tumor is stimulating only local lymphangiogenesis around the lymph nodes they are growing in, while disturbing the lymphatic vasculature in surrounding tissues. This biological explanation then connects the quantitative data to the tumor-bearing model in that the abnormally high volumes of lymphatic tissue and average volume per vessel measured are concentrated around the mouse’s lymph nodes, which are much more distinct in this model compared to the other two mice.

Although there are discrepancies between the whole models and their corresponding measurements, these do not exist when comparing the images of the region of interest in each model to the respective volume analysis. By modeling this region behind the tongue, we were able to focus on the direct impact of each experimental condition (neck dissection and cancer) on the structural integrity of a strongly connected section of lymphatic vasculature identified in the wild type mouse. Additionally, this specific region of interest allowed us to normalize the figure analysis and we can focus on the important region of the neck behind the tongue where lymph nodes are normally removed for the standard-of-care treatment. As a general trend, the vessel count, size, and total volume decrease when going from the wild type to the neck dissection to the tumor-bearing mouse model. These data from the normalized regions of interest match our hypothesis based on the lymphatic models generated by IMARIS.