An Ex Vivo Organotypic Culture Platform for Functional Interrogation of Human Appendiceal Cancer Reveals a Prominent and Heterogenous Immunological Landscape

Epithelial neoplasms of the appendix are a rare form of gastrointestinal neoplasia that have increased in incidence over the last two decades. Most neoplasms are diagnosed incidentally during treatment for acute appendicitis and prognostic indicators of survival are highly dependent on the extent of disease at diagnosis, histological grade, and when feasible, the completeness of cytoreduction (1, 2). Although surgery is a highly effective therapy for disease localized to the appendix, peritoneal metastasis results in a condition called mucinous carcinomatosis peritonei (MCP) and often occurs before the detection of the primary tumor. Overall, patients with appendix cancer with MCP whose disease cannot be completely resected, or recurs post resection, generally have a poor prognosis as current chemotherapeutic regimens for appendix cancer are derived from the standards for colorectal cancer in part due to a lack of disease-specific models available for preclinical studies.

Although a small number of patient-derived xenograft (PDX) models of appendix cancer has been generated (3, 4), most grow slowly, have low engraftment rates, are resource intense, and must be studied in immunodeficient animals. Given this caveat, the majority of appendiceal cancer research has been focused on using human biospecimens for histological and molecular studies (5–7). Recently, an organoid platform using reaggregated cells was described for predicting drug response in appendiceal tumors. Although described as a feasibility study, these data show that organoids may be a means to determine patient-specific response predictions to therapeutic intervention (8). Because this is the only current preliminary model to study human appendiceal cancer to our knowledge, there is a clear need to develop additional models allowing for the investigation of appendiceal cancer, and in particular models that recapitulate the entirety of the tumor microenvironment.

To address this unmet need, we have developed an organotypic slice platform using tumor tissue obtained at surgical resection. The development of organotypic tumor slices in other gastrointestinal malignancies has offered a unique opportunity to study interactions in the tumor microenvironment that allows for minimal disruption to the cellular and extracellular tissue components of cancer tissue (9, 10). The current study revealed that tumor slices preserve cellular and acellular (extracellular matrix) heterogeneity reflected by original tumor samples and can be maintained for at least 7 days in culture. These results were validated by confocal microscopy of intact tissue slices, as well as flow cytometry analysis of dispersed single cells from living tissue slices that yielded identifiable immune, epithelial, and fibroblast populations. To our knowledge, this is the first study to provide a characterization of the immune microenvironment of MCP specimens. Furthermore, we show that experimental agents that affect immune cell populations can be studied using tumor slices, as we show that immune cells maintain functional calcium responses within in situ cyto-labeled leukocyte populations after 7 days of slice culture. Given the presence of a prominent immune cell infiltrate and that MCP specimens derived from appendiceal cancer are nearly all KRAS mutant tumors (5, 6, 11, 12), to test our model, we used the novel preclinical SUMOylation inhibitor TAK981, due to the known immunomodulatory effects of SUMOylation and its modulation of mutant Ras signaling (13). We determined that exposure to the small ubiquitin–related modifier (SUMO) inhibitor TAK981 selectively remodeled the tumor microenvironment by reducing the contribution of regulatory T cells (Treg) within tumor slices demonstrating the utility of this model to evaluate novel therapeutics in the preclinical space for this rare disease.

Human tissues from de-identified patient samples were collected under the auspices of a UC San Diego Moores Cancer Center Biorepository IRB-approved protocol. For mouse studies, all animal protocols were approved by relevant UC San Diego institutional regulatory committees.

Fresh tissue specimens were received by the pathology laboratory within 1 hour of surgical resection. Tissues were sectioned into fragments (approximately 5–10 mL cubed) and imbedded in 4% low-gelling temperature agarose dissolved in PBS. Tissue blocks were solidified (4°C) for 30 minutes. Living slices were then cutoff (150 μmol/L) on a vibroslicer (VT1000P; Leica Biosystems). Once cutoff, human appendix tumor slices were washed with PBS (Sigma-Aldrich, Cat# D8537). The slicing procedure was typically completed within 45 minutes to 1 hour. Slices were cutoff and subsequently cultured or fixed for immunostaining. Organotypic slices were cultured in medium on top of Corning Transwell polyester membrane cell culture inserts (Cat# CLS3450-24EA).

Initial experiments were cultured in complete DMEM, 1% penicillin–streptomycin amphotericin B solution (Sigma-Aldrich, Cat# A5955), and 1% Glutamax supplement (Invitrogen, Cat# 35050061) with 10% FBS heat-inactivated FBS (Invitrogen, Cat# 16140063). During culture, slices were removed at daily timepoints for viability, drug dosing, flow cytometry, and calcium-imaging studies. Media changes were made every 24 hours.

Appendix slices were incubated with the Live/Dead viability/cytotoxicity kit for mammalian cells. Calcein-AM was used to determine live cells and the ethidium homodimer-1 for dead cells. Addition of both reagents was done according to the manufacturer's recommendations (Invitrogen, Cat# L3224). Once stained, slices were washed three times with DPBS (Sigma-Aldrich, Cat# D8537). The cutoff surface of the tissue slice was excluded from viability analysis.

For [Ca²⁺]_i imaging experiments, slices were bathed in HEPES-buffered solution (125 mmol/L NaCl, 5.9 mmol/L KCl, 2.56 mmol/L CaCl₂, 1 mmol/L MgCl₂, 25 mmol/L HEPES, 0.1% BSA, pH 7.4) and solutions were exchanged using a peristaltic pump (Isamtec, Wertheim Germany, catalogue number ISM832C). To visualize macrophages in situ, we used fluorescence-conjugated antibodies for CD11b (1:200; M1/70, catalog number 12-0112-82; Thermo Fischer Scientific). CD11b macrophages co-labeled with Fluo-4, AM, cell permeant (Thermo Fisher Scientific/Life Technologies, catalog number F14201) were further analyzed for functional analysis. Images were acquired on a Nikon Ti microscope with integrated autofocus, automated XY and Z stage, A1R hybrid confocal scanner, including a high-resolution (4,096×4,096) scanner, LU4 four-laser AOTF unit with 405, 488, 561, and 647 lasers, Plan Apo 10 (NA 0.8), 20X (NA 0.9) dry objectives. CD11b⁺ cells were reconstructed in Z-stacks of 5–15 confocal images (step size of 5.0–10.0 μmol/L) and analyzed using ImageJ. Images were acquired every 8–12 seconds. During ATP stimulation, the nonhydrolyzable ATP agonist ATPγS (Tocris Bioscience) was used. [Ca2⁺]_i responses in CD11b⁺ cells were quantified as the AUC of individual traces of Fluo-4 fluorescence intensity during stimulation.

Tumor tissue blocks were fixed in 4% paraformaldehyde, cryoprotected (30% sucrose), and tissue sections (10 μmol/L) cutoff on a cryostat. After permeabilization (PBS–Triton X-100 0.3%), sections were incubated in blocking solution (permeabilization buffer with 1% donkey serum). Primary antibodies were diluted in blocking solution. Immunofluorescence images were acquired with confocal microscopy. To visualize macrophages, we used antibodies against CD45 to visualize leukocytes (1:100, catalog number 304011; BioLegend), PanCK to visualize epithelial cells (1:100, catalog number SC8018; Santa Cruz Biotechnology), and αSMA to visualize fibroblasts (1:200, catalog number C6198; Sigma-Aldrich). Cell nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI) 1:1,000 DAPI (Catalogue Number D1306; Thermo Fisher Scientific/Life Technologies). Slides were mounted with ProLong Antifade (Invitrogen). Horseradish peroxidase–stained slides were used for morphologic evaluation of tumors. For automated counting of stained cell types (PanCK, αSMA, CD45⁺), ImageJ was used to split individual channels and to create a cell mask. The mask outline was analyzed and counted for each cell type.

Slices were incubated at 37°C with rotation for 10–15 minutes in 1 mL of digestion buffer consisting on DMEM high glucose, 10% Gentle Collagenase/Hyaluronidase from STEMCELL (GCH), 10% FBS, and 10% DNaseI (1 mg/mL stock). Slices and the digestion media were then added on top of a 70-μm filter on top of a 50-mL conical tube. With the back of a syringe the slices were grinded into smaller pieces and further washed with 4 mL of PBS 2% FBS. Flow through was spun at 300 × g for 10 minutes. Supernatant was removed and the pellet was washed with 1 mL of PBS 2% FBS. Fifty μL containing human Fc-block (2.5 μL per sample in 2% PBS/FBS) was added and left at RT for 15 min. Extracellular staining using respective antibodies was subsequently performed in 50 μL of 2% PBS/FBS. The following antibodies were used at 1:100 dilution immunotyping of tumor specimens: CD4 PE-Cy7-clone A161A1 catalogue number 357410 CD3 PE-clone UCHT1 catalogue number 555333, CD45 FITC-clone HI30 catalogue number 304006, CD25-APC-Cy7-clone BC96 catalogue number 302613, CD8 APC-clone SK1 catalogue number 344721, FoxP3 BV421-clone 206D catalogue number 320123, CD163 APC-Cy7-clone GHI/61 catalogue number 333621, CD11b PerCp-Cy5.5-clone M1/70 catalogue number 101228, and CD206 BV605-clone 15-2 catalogue number 321140 (all prior mentioned antibodies listed were purchased from BioLegend). Epcam BUV737-clone EBA-1 catalogue number 748397 (BD Biosciences). LD-Blue-Live–dead fixable dye Invitrogen catalogue number L23105. For intracellular staining samples were fixed and permeabilized with FoxP3 fix buffer (ebiosciences).

Viability for drug studies was quantified by live/dead confocal imaging or CellTiter-Glo (Promega). For live/dead imaging, see “viability assessment” Materials and Methods section. Cell titer glow was performed on sequentially cutoff matched slices from independent patient tumors, where drug and vehicle treatments were performed on serial cutoff slices, alternating between treatment conditions. Slices were treated daily for 5 days with drugs. A concentration of 100-nmol/L TAK981 daily was selected on the basis of considerations of tissue thickness and a similarly reported concentration of the compound's IC₅₀ value (14).

Spleens were collected from C57B6 mice and processed into single-cell suspension by mashing the spleen into a 70-μm filter with the back of a syringe and washing the filter with 10 mL of PBS 2% FBS. Splenocytes were spun at 300 × g for 10 minutes and used to purify CD4 naïve T cells by magnetic separation (STEMCELL). Then 600,000 T cells per well were plated in a 24-well plate previously coated with anti-CD3e antibody (625 ng/well) and cultured in RPMI media containing 10% FBS, 2 mmol/L l-Glutamine, 10 mmol/L HEPES, 1 mmol/L sodium pyruvate, 100 μmol/L MEM non-essential amino acids and 50 μmol/L B-Mercaptoethanol, plus anti-CD28 (0.5 μg/mL) and Treg supplement from STEMCELL (1:100 dil). Tak981 (100 nmol/L) or vehicle (DMSO) were added daily and cell counts and viability (measured by trypan blue) were assayed daily.

To analyze the differentiation of naïve CD4 T cells to Tregs, cells were stained with CD45, CD25, CD4, and FoxP3 antibodies at day 3 of culture with the TREG differentiation cytokines.

Statistical comparisons were performed using Student t test or one-way ANOVA followed by multiple-comparison procedures with the Tukey or Dunnett tests. Data are shown as mean ± SEM unless otherwise stated.

Further information and requests for resources, reagents, and data should be directed to and will be fulfilled by J. Weitz (jweitz@health.ucsd.edu).

Tumor samples were obtained at the time of cytoreductive surgery and only consisted of metastatic disease to the peritoneal surfaces and omentum. The tumor histology we investigated included MCP low grade and high grade, which were derived from neoplasms of appendiceal origin (Supplementary Table S1 and Supplementary Fig. S1). We found that cutting 150-μm sections was best suited for confocal imaging of cellular and acellular cytoarchitecture (matrix) of living tumor slices, where imaging resolution was lost in tissues thicker than 150–300 μm using confocal microscopy (Supplementary Fig. S2). These results are in line with tissue imaging standards for confocal microscopy (15). In addition, as a limited tissue resource, thin slices (150 μm) allowed for more experimental conditions compared with 500-μm sections. Thinner slices were not produced (i.e., less than 100 μm) given that they were difficult to reproducibly cut using a vibratome. Representative image of the tissue backscatter and live/dead staining show a dense network of cellular and ECM tissues derived from tissue resections (Fig. 1A). Cellular viability was tracked over a 7-day period in five patient tumors. Four of five patient tumors had initially high viability (tissue from Donor 0.28 was found to have lower initial viability; Fig. 1B). High-magnification images show that a large proportion of the live tumor remained viable for 7 days (Fig. 1C–H). The cutoff surface was excluded from analysis given that the physical process of cutting has been shown to damage the outermost surface from other organotypic slices’ preparations (16). We found that a viability of approximately 60% could be maintained and tracked during slice culture for at least one week (∼20% loss compared with initial viability; Fig. 1I). These results were comparable with cross-sectional area measurements taken from the organotypic tumor slices during a 7-day period, which has been shown to be a measure of tissue health in other organotypic slice preparations (17, 18). During analyses of viability, we noticed that in some tumor samples the viability was higher at days 7 than 4. This raised the question of whether cells were proliferating locally. We performed cellular proliferation analysis by adding EdU to the slice culture for 24-hour time points starting at day 0. Representative images at days 0–1 and 6–7 show changes of proliferating cells (green) within DAPI-positive nuclei (blue; Fig. 1J and K). We found that after the cutting procedure, cells proliferated the slowest. After 24 hours, cellular proliferation peaked at days 4–5 (Fig. 1L). The quantification of results reflect the proliferative index of previously published proliferative capacity (Ki67 index) of MCP tumor specimens, which was a median of 15%, with a hotspot index of 50% in epithelial rich regions (19). These results indicate that the appendix slice tumor model allows for the study of viable tumor tissue, as well as the ability to interrogate functional cellular readouts for a duration of a least one week.

We next sought to identify how distinct cellular populations are maintained in sequentially cutoff slices, as well as during long-term slice culture (7 days) compared with their initial composition. To identify whether sequentially cutoff slices were comparable within a tumor, we performed 150-μm sections, which were fixed using 4% PFA at the initial cutoff time (day 0). By performing immunostaining, we sought to identify epithelial, immune, and stromal components of sequential MCP tissue slices (Fig. 2A–E). Epithelial cells were labeled with PanCK (white), smooth muscle cells including fibroblasts were labeled with αSMA (red), and leukocytes for CD45⁺ (green). We found that different donors had distinct ratios of cell types (Fig. 2F). Notably, donor 339 had a high proportion of epithelial cells (PanCK⁺), whereas donor 22 had a low number of epithelial cells and a higher number of CD45 leukocytes. Although different donors had distinct ratios of cell types, the cell type percentage was not significantly different in matched regions from similar tissue regions from sequential slices of the same donors (Fig. 2G). We next sought to determine whether the proportions of these cell types changed during slice culture over a 7-day period, and whether there were any changes to the viability of epithelial, fibroblast, or immune cell populations during slice culture. To do so, we performed flow cytometry to quantitate changes in cellular populations for 1 week in matched tumor slices (Fig. 2H). We found that the slice culture process did not induce systematic changes in αSMA or CD45 cell populations across 5 different donors (Fig. 2I–N). We saw an increase in PanCK⁺ total cell percentage in day 7 compared with day 0 from 4 out of 5 samples, suggesting that epithelial cells proliferated in slices culture (Fig. 2K). Notably, we saw that only 1 sample (donor 339) had a high percentage of epithelial cells, which correlated with a high-grade histology (see Supplementary Table S1). Viability changes induced by the slice culture in the three populations (fibroblasts, epithelial, and immune) was unremarkable (Fig. 2L–N). Although this is the first report to our knowledge to characterize the immune microenvironment of these tumors, our findings regarding the percentage of cell types are in line with previous clinical reports of epithelial cell content from varying tumor grades (20–23). These results reveal heterogeneity of MCP specimens and indicate that the cellular contents are maintained for 7 days in slice culture.

The factors underpinning tumor sensitivity to therapeutics are numerous and complex but are clearly influenced by both cellular and acellular components of the tumor microenvironment (24). Given that the immune landscape within the tumor microenvironment is one factor known to correlate with response to immune checkpoint blockade, we sought to characterize the immune cells present within appendiceal cancer metastases. We prepared single-cell extracts from tumor tissue as well as tumor slices. Single-cell suspensions were then characterized by flow cytometry analysis (Fig. 3A). Initial viability was assessed using the fixable viability marker LD Blue. Leukocytes were identified by the protein tyrosine phosphatase receptor type-c (CD45) and further characterized into myeloid (CD11b⁺) and lymphocyte subsets (CD3⁺). Within the myeloid lineage we assessed, M2-type macrophages were identified by the presence of CD206 or CD163 staining. Among the lymphocytes we assessed (CD3⁺), lymphocytes were either classified as CD8 or CD4 T cells. In addition, CD4⁺ T cells were further classified as Tregs defined by expression of CD25 and FoxP3. The proportion of immune cells present in the different patient samples analyzed ranged from 21% up to 95%, illustrating the marked heterogeneity between patients. These data correspond with our findings from analogous flow cytometry quantifications (Fig. 2D–F), and matching histological assessment (see Supplementary Fig. S1C). In addition, we found that the percentage of CD45⁺ cells did not correlate with tumor grade (Supplementary Fig. S3). To further examine changes occurring in the immunological profile of differences distinct in histological grade tumors, pie charts were generated to depict the representative immune populations within patient tumor samples, which were divided into two categories: low-grade (bottom row) or moderate/high-grade (top row) MCP (Fig. 3B). Within low-grade tumor tissues, we found that the total proportions of immune cell subsets were similar [i.e., T cells and macrophages had a smaller variability (standard deviation shown in bar plots) between patients]. On the other hand, we found distinct, donor-specific differences in the immunological landscape of moderate- and high-grade tumors. Namely, although macrophages were the predominant population amongst 3 of the 4 tumors, donor 0.35 had almost no macrophages, as T cells constituted the bulk of the immune cells of this donor. In addition, donor 0.32 had nearly no T cells, demonstrating a marked difference among moderate- and high-grade tumors. In this study, we did not include markers for other immune cells such as B cells, dendritic cells, myeloid-derived suppressor cells, natural killer cells, or neutrophils, but we plan to include them for future studies; for the moment these fall under the “other” category.

On the basis of our findings of a diverse repertoire of leukocytes in appendiceal cancer, we next sought to identify whether immune cells remained functionally active after 7 days in slice culture. As such we first determined that culture conditions did not significantly alter the proportion of immune cell populations compared with the original tumor biopsies (Supplementary Fig. S4). Next, to investigate immune cell functional responses, we incubated living tissue slices with the calcium-imaging dye Fluo-4 and subsequently perfused Ca²⁺-imaging buffer over the tissue slices using a fluidics-based approach. The representative scheme is described (Fig. 4A and B). Intracellular calcium [Ca²⁺]_i signals are required for many immune cell–signaling processes, including cytokine secretion (25), clearance of apoptotic cells (26), and immune cell activation (27). To specifically locate in situ macrophage subsets within organotypic slices, we used a CD11b antibody conjugated to a fluorophore, which had previously validated for myeloid cell staining (Fig. 4C). Co-labeling of fluo-4 and myeloid cells with antibodies allowed for real-time imaging of [Ca²⁺]_i responses in these cell subsets (Fig. 4C and D’). Also, see additional Supplementary Videos (Supplementary Movie S1 and Supplementary S2).

Leukocyte Ca²⁺ responses occur as a result of their stimulation by a multitude of different signaling molecules, including neurotransmitters such as ATP, which play a critical role in cytokine production and “Find-Me” signal apoptotic cell clearance (26, 28). To test whether immune cells remain functionally active during slice culture, we tested whether ATP could activate calcium responses in tumor-associated macrophages within appendiceal slices. We found that after 7 days of culture, perfusion of human CD11b⁺ macrophages with exogenous ATP-induced increases in signaling in multiple cell types (Fig. 4E and F). Tracking of raw [Ca²⁺]_i traces of stimulated macrophages allowed for temporal reconstruction of imaging recording sessions (Fig. 4G), as well as tracking of dose-dependent responses to the perfusion of purinergic (ATP) agonists (Fig. 4H and I). On the basis of these findings, we conclude that tissue slices are a suitable platform to study physiologic responses and pharmacological investigation of immunotherapeutic drugs within tumor slices from appendix cancers for at least 7 days.

As an orphan disease, there are no FDA drug approvals specific to appendiceal cancer. Given reports by our group and others that the genomic landscape of appendix cancer may differ substantially from colorectal cancer (5, 6, 29, 30), preclinical studies focused specifically on appendix cancer are needed. Given the rich immune microenvironment we identified in appendiceal tumor slices, we sought to evaluate a novel therapeutic agent in this model. TAK981 inhibits the SUMO-activating enzymes (SAE1 and 2) and is currently under investigation in several clinical trials (14, 31). Inhibiting SUMOylation with TAK981 has been shown to have both antitumor immunomodulatory effects and to negatively regulate oncogenic KRAS signaling (32, 33). Given that mucinous appendiceal cancers have high rates (∼70%) of oncogenic mutations in KRAS (5, 6, 11, 12), we hypothesized that TAK981 may have activity in MCP specimens, directly through interference with KRAS signaling, or indirectly through immunomodulation. Moreover, immunomodulatory effects have been seen in other KRAS mutant tumors such as pancreatic cancer (34). As such, we examined the impact of TAK981 on tumor slices derived from appendix cancer metastases. Slices were incubated on Transwell culture inserts with drug or DMSO vehicle control for 5 days, changing media and drug daily (Fig. 5A). We observed proliferation during slice culture in multiple cell populations during slice culture (Supplementary Fig. S5). After 5 days, sequentially matched slices (Fig. 2B) were incubated in Cell Titer-Glo for 30 minutes and assessed for cell viability. We found that TAK981 reduced the viability of six patient tumors we tested, which was comparable with Mitomycin C, which is the most common agent used during hyperthermic intraperitoneal chemoperfusion (HIPEC) for patients with MCP. Interestingly, donor tissue slices from two donors who had a previous HIPEC procedure before the most recent surgical resection, responded poorly to Mitomycin C treatment (Fig. 5C and D). These results are in line with in vivo findings from human patients that have yet to show clear benefit in repeating HIPEC therapy after recurrence (35, 36). High concentration of the proteosome inhibitor, bortezimib (2 μmol/L), was used to control for induced cell death (Fig. 5B). We next sought to identify whether the changes we observed in our toxicity (CTG) assay were specific to tumor cells, or other cell types. As such we performed TUNEL, to check for apoptosis in tumor cells. We noticed that although bortezomib induced cell death in all cell types (control for cell killing) as compared with the control, TAK981 induced more cell death in non-tumor cell types (Fig. 5E–H). We next sought to determine whether leukocyte populations would be significantly altered due to the known effects of TAK981 as an immunomodulatory drug. As such, we performed immunotyping of control and TAK981-treated samples after one week of slice culture (Fig. 5I–K). We observed a trend in decrease of Tregs (CD4⁺, CD25⁺, and FOXP3⁺; 5 of 6 patient samples), and a small effect size, but a significant decrease in CD8⁺ cells (6 of 6 patient samples) after 1 week of culture compared with slice-matched controls from the same tumor region (Fig. 5L). We did not observe a significant trend in viability or change of polarization in macrophage populations. These results were confirmed by treatment from TAK981-treated bone marrow–derived macrophages polarized to M1 or M2 in vitro (Supplementary Fig. S6). These findings are similar to those from other groups, which show that TAK981 does not have an effect on macrophage viability (37). These findings demonstrate how slice cultures can be used to reveal the impact of immunotherapeutic agents on components of the tumor microenvironment.

Because we observed some reduction of Tregs in 5 out of 6 treated samples, we set out to investigate the possible mechanisms of TAK981’s effect on these cells. For this purpose, naïve CD4 T cells were purified from C57B6 spleens by magnetic separation (Fig. 6A) and put in culture with Treg differentiation cytokine mix (STEMCELL) in the presence of either vehicle (DMSO) or TAK981 (100 nmol/L). Drug was added daily with media changes, and cell counts with trypan blue staining were performed daily, to assess viability. TAK981-treated cells had significantly reduced viability based on both cell counts (Fig. 6B) and trypan blue staining (Fig. 6C) as early as day 1; the differences grew larger with time and by day 4 only 25% of the TAK981-treated cells were alive compared with 75% of the DMSO-treated cells. To further explore these findings, differentiation studies looking at Treg markers CD25 and FoxP3 were performed at day 3 of culture to have enough viable cells to make an accurate assessment because we could already observe a reduced viability in the TAK981-treated cells at this timepoint (15% compared with 69% for DMSO-treated; Fig. 6D). Interestingly, the percentage of live CD45⁺/CD4⁺ T cells also decreased from 93% in vehicle to 63% with TAK981 treatment. This raises the possibility that TAK981 may de-differentiate CD4 T cells into CD45⁻/CD4⁻ cells (Fig. 6E). When we looked at Treg markers, we saw a significant reduction in CD25⁺/FoxP3⁺ CD4 T cells (84% in DMSO- vs. 21% in TAK981-treated; Fig. 6F). Interestingly, the drug-treated cells still expressed high levels of CD25 but had lost FoxP3 expression. These data were concordant with our observations of reduced Tregs in patient-derived tumor slices treated with TAK981.

Officially listed by the National Organization of Rare Disorders, appendiceal cancer is listed as a rare or “orphan” disease. Here, we have developed and characterized the first preclinical model of appendiceal cancer that embodies the entirety of the tumor microenvironment, and thereby allows us to probe the biology and study therapeutics in the laboratory. The paucity of models in this space exist for several reasons: (i) there is rare access to clinical tissues; (ii) a majority of the neoplasms have mucinous histology and many are low grade, making them poorly suited for standard 2D and 3D culture; and (iii) mice do not have a human appendix equivalent, thus making studies on genetically modified mice not feasible, and not likely to recapitulate human disease (38). As preclinical platforms using organotypic slices have been developed to model disease in various cancer types, including pancreas, gastric, lung, and colon (9, 10, 39), we anticipate that this highly reproducible organotypic slice model will be fundamental for studying numerous aspects of appendiceal cancer biology.

As cancer immunotherapy has emerged at the forefront of clinical oncology, organotypic tumor slices from patient-derived tumors have offered a unique platform to study local paracrine signaling between immune and tumor cells. Our results show that the preparation of living appendiceal cancer slices allows for immunocompetent studies derived from ex vivo preparations of patient tumors. To our knowledge, the immunological landscape of appendiceal tumors has not been previously characterized. As such, we found that immune cells (CD45⁺) represented a large proportion of the cellular mass in some appendiceal tumors, ranging from 21% to 95%. This population of cells included myeloid cells and lymphocytes, including CD4 and CD8 cells, as well as a smaller population of CD4⁺ Tregs. Tumor composition was very heterogeneous, especially among the high-grade tumors with some being macrophage rich (patient sample 32) and others T-cell enriched (patient sample 0.35). These findings raise the potential of grouping tumors not only based on high or low grade but based on similarities in immune cell composition and more targeted therapies could be developed for these patients, for example, targeting M2 macrophages in patients like sample 32 or using immune checkpoint therapy to take advantage of the large amount of T cells in patients like sample 35. Our next steps will be to further characterize the immune microenvironment and immune checkpoint status and be able to classify patients based on these parameters.

Given that appendiceal tumors had a large cellular component containing immune cells, we sought to identify whether appendiceal slice culture would be a platform that will allow for long-term studies of the tumor-immune microenvironment. To identify whether immune cells remained functional in culture at 1 week, we performed calcium imaging. We found that immune cells had increased [Ca²⁺]_i activity during activation in chamber perfusion of biological stimuli, such as neurotransmitters (ATP). Interestingly, we observed uncoordinated spontaneous Ca²⁺ response in their basal (unstimulated) state. These results suggest that endogenous paracrine signals released in tumor slices likely contribute to immune function. Given the propensity of appendiceal cancer to metastasize to the omentum (7), which contains large amounts of adipose tissue, as well as “milky spots” rich in immune cells, it will be interesting to study adipokine and chemokine signaling effects on appendiceal tumor slices given the known importance of these signaling effects in the etiology of metastasis in other cancers (40–43).

Notwithstanding, substantial technical limitations still exist. Although we observed that cellular viability and functionality remained for at least 7 days (greater than ∼60% for 4/5 samples tested), it will be important to continue to improve upon culture conditions for studies requiring longer culture (weeks to months). In addition, we successfully performed tumor slice culture on both high-grade tumors and low-grade mucinous neoplasms; however, we observed that mucinous, low cellularity tumor specimens (not containing nodules) were difficult to cutoff. As such, a recent study performed using appendiceal organoids showed success with highly mucinous tumors, and thus, this alternative approach may be important for studying these clinical samples (8). Another limitation of this study is based on the amount of tissue collected upon resection. We found that we could produce 20–50 slices on average depending on the tumor cytoarchitectural integrity (containing low or high amounts of mucin), as well as the amount of tissue received (1,000 mg on average). As such, given the limited resource, we did not have sufficient tissue slices to perform multiple experiments (i.e., immunotyping, viability, CTG, TUNEL, EdU) from all donors. Although this technique is not a high-throughput approach, there are currently no other available techniques to investigate cell pathophysiology without extensive disruption of the tumor microenvironment or to perform rapid bench-to-bedside testing for patients with MCP.

The organotypic tissue slice platform offers an alternative approach to current existing ex vivo models to study appendiceal cancer biology. Complementary research using patient-derived organoids have recently been described (44). Here, authors used an approach by deconstructing patient tumors, lymph nodes, and blood to recreate a bio-fabricated organoid that was used for drug screening. This approach was able to identify a personalized immunotherapy response profile for patients' tumor samples. However, tissue limitations existed for this type of sample preparation, because the authors did not expand organoids for use as a scalable established 3D cell line. Importantly, we reconfirm here that cancerous tumor cells only represent a small fraction of the cellular population of the tumor in low-grade tumors. This is in-line with molecular profiling of tumors with varying tumor grades (45), which show high-grade tumors have enriched expression profiles of cancer cells, including EPCAM. As such direct immunofluorescence using colocalization studies will be needed in future studies examining these tissues to determine whether therapies have therapeutic effects on tumor cells, or rather offer immunotherapeutic benefits in the local tumor microenvironment (i.e., activate local immune cells, or ablate immunosuppressive cell types). Using a different ex vivo model, the authors used explant models from patient-derived appendiceal tumors to show that extracellular mucin imparted tumor chemoresistance to therapy and drug treatment modeling using PDXs (46, 47). These findings may help with novel strategies to get higher drug concentrations to tumor tissues, which may produce better patient outcomes. As previously described, it is clear that organoid- and explant-based approaches are important for uncovering the biology of this rare tumor type. In addition to these highlights there have been numerous reports of PDX models (3, 48, 49). Although PDX models are also useful, they however are labor intensive, rarely engraft, and grow extremely slowly in vivo. The addition of the established organotypic slice platform described here will provide researchers with an additional tool to study this notoriously difficult disease to model.

Finally, we describe an organotypic tumor slice model as a drug discovery platform to test potential therapeutics. Interestingly, we found that two donor patient samples who had undergone a second resection after an HIPEC treatment 1 year prior did not respond to Mitomycin-C. These results indicate that our model may be predictive of chemoresistance in-vivo, as Mitomycin-C is used during HIPEC procedures. Future studies will focus on bench-to-bedside therapeutic investigations. In this current article, we examined a novel drug with immunotherapeutic potential (TAK981) as an investigational therapeutic candidate for MCP. We found that a modest reduction of tumor slice viability during TAK981 treatment (CTG analysis) was donor dependent and was accompanied by a loss of TREGs (5/6 patient samples). TUNEL analysis confirmed results from CTG experiments as we found that the results of TUNEL staining showed similar drug responses in therapeutic-treated conditions (TAK981, Mitomycin-C, 5-FU, and bortezomib). Our results were validated in an in vitro model of TREG differentiation, which showed a reduction in viability, as well as differentiation of naïve CD4⁺ T cells into Tregs. These results are in line with other studies, which show that TAK981 interferes with TREG signaling processes, which was recently reviewed (31). As Treg populations contribute to immunosuppression (50), TAK981 may prove to be an interesting candidate to combine with treatments that harness the immune system to target cancer cells.

J. Baumgartner reports expert testimony. Y. Chen reports equity ownership and Board of Director with Suvalent and Aravalent Therapeutics, as well as consulting fees from Suvalent Therapeutics, Inc. outside the submitted work. A.M. Lowy reports personal fees from Kinnate, Bluestar Genomics, Pfizer, Merck, and Rafael, as well as grants from Syros outside the submitted work. No disclosures were reported by the other authors.

J. Weitz: Conceptualization, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. T. Hurtado de Mendoza: Conceptualization, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. H. Tiriac: Conceptualization, supervision, visualization, writing–review and editing. J. Lee: Formal analysis, validation, investigation, methodology, writing–review and editing. S. Sun: Formal analysis, investigation, methodology, writing–review and editing. B. Garg: Investigation, writing–review and editing. J. Patel: Resources, investigation, methodology, writing–review and editing. K. Li: Resources, formal analysis, validation, investigation, methodology, writing–review and editing. J. Baumgartner: Resources, methodology, writing–review and editing. K.J. Kelly: Resources, investigation, methodology, writing–review and editing. J. Veerapong: Conceptualization, resources, supervision, methodology, writing–review and editing. M. Hosseini: Conceptualization, supervision, funding acquisition, investigation, methodology, writing–review and editing. Y. Chen: Conceptualization, supervision, investigation, writing–review and editing. A.M. Lowy: Conceptualization, supervision, funding acquisition, investigation, writing–review and editing.

This work was supported by grants from the National Organization for Rare Disorders R21CA273974 (to A.M. Lowy), 1F32CA265052-01 (to J. Weitz), and generous gifts from the estate of Elisabeth and Ad Creemers, the Euske Family Foundation, the Gastrointestinal Cancer Research Fund, and the Peritoneal Metastasis Research Fund. Confocal imaging and histology core was done with support from the UCSD Specialized Cancer Support Center P30 grant 2P30CA023100. LJI BD FACSymphony S6 purchase was partially funded by the Bill and Melinda Gates Foundation, United States (Shane Crotty) and LJI Institutional Funds. We would like to thank staff of the Department of Pathology and the biorepository staff at the Moore's Cancer Center at the University of California, San Diego for their efforts to obtain and distribute de-identified human donor material. We would also like to thank Dawn Jaquish, Edgar Esparza and Evangeline Mose for help with critical lab needs. We would also like to acknowledge Denise Hinz at the La Jolla Institute of Allergy and Immunology (LJI) Flow cytometry core, for her assistance and expertise with flow cytometry experiments and Kersi Pestonjamasp form the MCCT microscopy core. Finally, we would like to thank Alfredo Molinolo, Mason Kyle and Brian Wishart with the UCSD biorepository for the coordination of patient tumor samples.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.