Modeling Canine Hemangiosarcoma Progression Using Patient-Derived 2.5D Organoids and Orthotopic Xenografts Authors: Yishan Liu, Haru Yamamoto, Mohamed Elbadawy, Amira Abugomaa, Masahiro Kaneda, Yomogi Shiota, Tadashi Kondo, Tatsuya Usui, Kazuaki Sasaki Published: 2026 ABSTRACT Simple SummaryCanine hemangiosarcoma is a highly aggressive cancer that arises from blood vessel-forming cells and often spreads rapidly to other organs. Effective treatments for this disease remain limited, partly because the biological mechanisms that drive its progression are not fully understood. In this study, we developed a laboratory model using three-dimensional tumor-like structures grown from canine hemangiosarcoma tissues collected from affected dogs. These structures retained important characteristics of the original tumors and were used to investigate drug responses and changes in gene activity associated with the disease. Our analysis identified phospholipase A and acyltransferase 3 as a molecule that was more abundant in hemangiosarcoma than in non-cancerous tissue. Reducing the activity of this molecule, either by suppressing its production or by using a chemical inhibitor, markedly decreased the ability of cancer cells to invade surrounding areas, suggesting that it may contribute to tumor progression. The tumor model also reproduced tumor growth and spread when implanted into mice. These findings provide a useful experimental model for studying canine hemangiosarcoma and identify phospholipase A and acyltransferase 3 as a candidate molecule associated with the invasive phenotype of canine HSA. Because canine hemangiosarcoma shares important features with human angiosarcoma, this model may also provide a useful platform for future comparative studies of these diseases. 1. Introduction Canine hemangiosarcoma (HSA) is a malignant tumor, which accounts for a relatively low percentage (1.3–2.8%) of canine cancer cases [1]. As a highly invasive cancer, HSA often metastasizes to the liver and lungs through blood vessels [2,3]. Because HSA shares several pathological characteristics and molecular signatures with human angiosarcoma (AS) [4], it is considered a more relevant preclinical model than rodent models for advancing AS research. Conventional therapies for HSA include surgery and chemotherapy with drugs such as doxorubicin. However, the median survival time after chemotherapy is six months, whereas that after surgery alone is two to three months [5]. Therefore, to better understand disease progression and develop new treatment options, a more suitable animal model is required to uncover the transcriptional profiles associated with cancer progression. Clinically, splenic masses in dogs are mostly benign tumors, including nodular hyperplasia (NH). Splenic NH is defined as the overgrowth of normal splenic components, forming a well-restricted proliferative lesion [6]. During new target development, a common method includes the comparison of the transcriptional patterns between cancerous and healthy tissues. Several sequencing studies comparing HSA with normal healthy tissues have reported various mutated genes such as TP53 and PIK3CA, which are implicated in DNA repair, and those associated with the PI3K/AKT/mTOR and MAPK/ERK pathways, which are involved in cell proliferation [7,8,9]. However, a comparison of HSA with healthy tissues may not completely reveal the characteristics of cancerous proliferation. Fast-proliferating NH may provide a reference for identifying the cause of malignant metastasis. Although the propensity for metastasis has been reported to be highly associated with lysosomal [10] and membrane dysfunction [11], their roles in HSA remain unclear. Several currently established mouse models replicating the biological and molecular characteristics of HSA have been reported; however, these models are mainly based on the direct subcutaneous transplantation of patient tumor tissues into immunodeficient mice [12,13]. Therefore, these models cannot recapture the metastatic features of the original tumor within the liver or other organs. Given that high invasion and metastasis are key factors responsible for treatment failure and poor prognosis in dogs with HSA, a new reliable mouse model is urgently needed for mechanistic studies and drug development. In previous studies, we established a 2.5D organoid culture method using tumor tissues from dogs with bladder, breast, and lung cancers, which could recapitulate the characteristics of the original tumor tissues [14,15]. In this exploratory study, we aimed to establish and characterize HSA 2.5D organoids and compare them with NH organoids to investigate their biological and transcriptomic characteristics. Tumorigenesis ability of generated HSA 2.5D organoids was tested via splenic injection. HSA markers CD31, vWF, and VEGF were validated, and gene expression profiles between HSA and NH organoids were examined along with transcriptional differences, particularly those related to metastasis and invasion, such as lysosomal membrane-related pathways and genes such as phospholipase A and acyltransferase (PLAAT)3. Additionally, the effects of PLAAT3 knockdown and chemical inhibition in HSA cells on their invasion and proliferation were demonstrated. 2. Materials and Methods 2.1. Sample Collection and HSA 2.5D Organoid Culture All dog owners provided written informed consent for the present study, and all experimental procedures were carried out according to the guidelines of the Institutional Animal Care and Use Committee of Tokyo University of Agriculture and Technology (Approval number: 0020007; approval date: 6 May 2021). Sample information is presented in Table 1. Standard hepatectomy and/or splenectomy and mastectomy were performed with informed consent from the owner. Canine tissue samples used in this study were obtained from surplus tissue collected during routine surgical procedures performed for clinical diagnosis or treatment. No anesthesia or euthanasia was performed specifically for the purposes of this study. Tissue sections used for primary culture were obtained from the same positions that were used in pathological diagnosis and immunohistochemistry (IHC) staining. The samples were immediately transferred to a cooled shipping medium and transported to our laboratory. 2.5D organoids were generated using methods described in our previous study [14]. Briefly, after receipt from the clinic, the tumor tissue samples were minced and digested with Liberase™ TH (Roche Diagnostics, Mannheim, Germany) for 20 min at 37 °C and then centrifuged at 600× g for 3 min. The tissue pellets were trypsinized for 3 min, and the digested cells were collected using a cell strainer. Red blood cells were then removed using RBC lysis buffer. After washing three times with PBS, the primary cells were seeded in dishes containing our prepared 2.5D organoid medium. Culture conditions, medium composition, supplements, growth factors, cell handling, and passaging procedures were identical to those described in our previous study. The culture medium comprised advanced DMEM including 50% Wnt, R-spondin, and Noggin conditioned medium; 100 µg/mL Primocin; 10 mM nicotinamide; 1% GlutaMax; and 1 mM N-Acetyl-L-cysteine (Thermo Fisher Scientific, Waltham, MA, USA); 500 nM A83-01 (Adooq Bioscience, Irvine, CA, USA); and 50 ng/mL mouse EGF (PeproTech, Rocky Hill, NJ, USA). All representative phase-contrast images of the cultured cells were captured using a light microscope (BX-52; Olympus, Tokyo, Japan). 2.2. Hematoxylin and Eosin (H&E) Staining Paraffin sectioning and H&E staining were conducted according to standard procedures, as described previously [16]. All sections were stained with hematoxylin and eosin for 4 min and 30 s, respectively. Sections were observed and imaged using a light microscope (BX-52; Olympus, Tokyo, Japan). 2.3. Immunohistochemical (IHC) Staining IHC staining was performed on paraffin-embedded sections, 4 μm in thickness. The following antibodies were used: rabbit polyclonal CD31 antibody (GeneTex, catalog No. GTX130274; 1:500), mouse monoclonal PLA2G16 (PLAAT3) antibody (OriGene, catalog No. TA506908S; 1:200), rabbit polyclonal Von Willebrand factor antibody (abcam, ab6994; 1:200), and anti-VEGF polyclonal antibody (Bioss, catalog No. bs-1665R; 1:200). Sections were deparaffinized with xylene and subsequently rehydrated with ethanol and distilled water. For antigen retrieval, sections were autoclaved at 121 °C with sodium citrate buffer for 5 min, followed by cooling at room temperature (RT). After washing, sections were quenched with 1% hydrogen peroxide for 30 min. Sections were incubated overnight at 4 °C with primary antibodies (CD31, 1:500; PLAAT3, 1:200; vWF, 1:200; and VEGF, 1:200) diluted in 0.5% casein in a humid box. After washing with PBS, sections were incubated in an HRP system and stained using a Dako kit (K3468) following the manufacturer’s protocol. After DAB staining, sections were counterstained with hematoxylin, dehydrated using ethanol and xylene, mounted with Polymount, dried, and examined. The sections were observed and imaged using a light microscope (BX-52; Olympus, Tokyo, Japan). 2.4. Immunofluorescence (IF) Staining Immunofluorescence staining of HSA 2.5D organoids was performed as previously described [14,15]. Immunofluorescent HSA 2.5D organoid cells (2 × 105/well) were seeded onto round cover glasses in a 6-well plate. After reaching 80% confluence, cells were fixed in 4% PFA for 1 h at RT. Subsequently, the cover glasses were washed three times with PBS and blocked with 1.5% normal goat serum (diluted in PBS) for 30 min at RT, and subsequently incubated in a humid box overnight at 4 °C with primary antibodies (CD31, 1:300; PLAAT3, 1:100; vWF, 1:200; and VEGF, 1:200) diluted in PBS. After washing with PBS, the slides were incubated with Hoechst (1:1000) and with a secondary antibody for 60 min at RT in the dark. Representative images were captured and analyzed. The secondary antibodies used were: Goat Anti-Rabbit IgG H&L (Alexa Fluor® 488) (ab150077) and Goat Anti-Mouse IgG H&L (Alexa Fluor® 488) (ab150113). All samples were observed under a fluorescence microscope (DAPI and FITC). Semi-quantitative analysis of fluorescence was performed using the ImageJ software, version 1.54r. 2.5. Cell Viability Screening Cytotoxicity activities were evaluated via the PrestoBlue™ Cell Viability assay. Cells (1 × 104 cells/well) were seeded in 96-well plates containing 2.5D medium. After 24 h, cells were treated with dimethyl sulfoxide (DMSO; vehicle), toceranib (Sigma Aldrich, America, MI, USA; at 2 μM, 4 μM, and 8 μM), carboplatin (FUJIFILM Wako Pure Chemical, Tokyo, Japan; at 1 μM, 10 μM, and 100 μM), cyclophosphamide (at 1 μM, 10 μM, and 100 μM), or doxorubicin (Cayman Chemical, Ann Arbor, MI, USA; at 1 μM, 10 μM, and 100 μM), and incubated at 37 °C for 72 h. Following incubation, 10 μM of PrestoBlue™ Cell Viability Reagent (Thermo Fisher Scientific) was added to each well, and the cells were incubated for an additional 3 h at 37 °C. Fluorescence (emission wavelength: 585 nm) was measured using a microplate reader (TECAN, Seestrasse, Switzerland). The chemical and company names were as follows: LEI110 (MCE, HY-125254), toceranib (Sigma, No. PZ0338), doxorubicin (Cayman Chemical, 25316-40-9), cyclophosphamide (TCI, 6055-19-2), and carboplatin (Fujifilm, 41575-94-4). IC50 values were determined by nonlinear regression analysis of the dose–response curves using GraphPad Prism. When 50% inhibition was not reached within the tested concentration range, the IC50 was reported as greater than the highest concentration tested. 2.6. RNA Isolation and Real-Time Polymerase Chain Reaction (PCR) Isolated RNA was converted to first-strand cDNA using the ReverTra Ace qPCR RT Kit (Toyobo Co., Ltd., Osaka, Japan) following the manufacturer’s protocol. Real-time PCR was performed using the QuantiTect SYBR I kit (Qiagen, Hilden, The Netherlands) and StepOnePlus Real-Time PCR system (Applied Biosystems, Waltham, MA, USA). The ΔΔCq method was used to quantify the data. Each cDNA sample was amplified with primers and run in triplicate. The expression level of each gene was normalized to GAPDH levels in each sample. Primers (FASMAC Co., Ltd., Kanagawa, Japan) used for canine genes MARVELD3, Huntingtin-interacting protein 1 (HIP1), PLAAT3, and GAPDH are listed in Table 2. 2.7. RNA Sequencing RNA sequencing and bioinformatic analyses were performed as previously described with minor modifications [17]. Ten biologically independent original tissue samples (HSA, n = 5; NH, n = 5) were collected from the clinic and used to establish 2.5D organoid lines for subsequent RNA sequencing. The sample IDs and corresponding sample information are provided in Supplementary Table S1. Total RNA was extracted from harvested primary cultured 2.5D organoids using the FavorPrep Tissue Total RNA Purification Mini Kit (Favorgen Biotech Corp., Pingtung, Taiwan) following the manufacturer’s protocols. One microgram of total RNA from each sample was used for cDNA library preparation following poly(A) mRNA enrichment. Libraries were sequenced on an Illumina NovaSeq 6000 platform using a 150 bp paired-end (PE) strategy. The reads were mapped to the reference genome (UU_Cfam_GSD_1.0; NCBI RefSeq assembly GCF_011100685.1) using Hisat2 (v2.2.1) [18]. Gene expression levels were quantified using HTSeq (v0.6.1) [19]. Differential expression analysis was performed using DESeq2 Bioconductor package [20]. Genes with an adjusted p value (Padj) ≤ 0.05 were considered differentially expressed. Principal component analysis (PCA) and hierarchical clustering were performed to assess the overall transcriptomic relationships among the HSA and NH samples. The PCA and hierarchical clustering results are shown in Supplementary Figures S1 and S2, respectively. Gene Ontology enrichment analysis was performed using GOSeq (v1.34.1) [21]. Gene set enrichment analysis (GSEA) was performed using GSEA software 4.4.0 developed by Broad Institute, Inc. [22,23] to identify significantly enriched biological pathways. Significantly enriched pathways were identified based on the normalized enrichment score (NES), nominal p value and false discovery rate (FDR). Gene sets with FDR q-values < 0.25 were considered significantly enriched. 2.8. Orthotopic Patient-Derived Xenograft Mouse Model Ten-week-old C.B-17/IcrHsd-Prkdc scid mice (SANKYO Laboratory, Tokyo, Japan) were anesthetized via isoflurane inhalation and positioned on their right side. The area above the spleen was shaved, and the skin was sterilized using 80% ethanol. A subcostal incision (approximately 2 cm) was made below the ribs to access the peritoneal cavity, exposing the spleen. Primary cultured cells (BS24028, 1 × 106 cells in 50 µL PBS per mouse) were gently resuspended and slowly injected into the spleen. Following injection, the site was gently compressed for 2 min using a cotton swab to prevent backflow. The spleen was then repositioned at its site in the abdominal cavity, and the peritoneum and skin layers were sutured. Ten weeks post-injection, mice were euthanized by intravenous administration of an overdose of pentobarbital sodium, and tumor-bearing organs were subsequently harvested and examined. All procedures were conducted with the approval of the Tokyo University of Agriculture and Technology Animal Care and Use Committee and Ethics Committee (approval number: R5-124). All procedures were done in accordance with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines. 2.9. Cell Proliferation Assay HSA 2.5D organoid cells were seeded in 96-well plates at 1000 cells/100 µL medium/well. The number of living cells was detected via the PrestoBlue™ Cell Viability assay on days 1, 4, and 7, and analyzed by measuring the fluorescent intensity in each well using a Tecan microplate reader, as previously described [14]. 2.10. Boyden Chamber Invasion Assay Matrigel was diluted with serum-free medium (advanced DMEM) at a ratio of 1:1. For invasion assays, 50 μL of diluted gel was dropped in each Transwell® insert (8 μm PET membrane, Corning 3464). The gel was then solidified at 37 °C in a CO2 incubator for 1 h. After fixation, the extra medium was aspirated into the insert. Cells were trypsinized and resuspended in serum-free DMEM. Next, 3 × 104 cells were resuspended in 300 μL serum-free media with treatment or vehicle and seeded into the upper chamber. Thereafter, 600 μL of culture medium with 10% FBS as a chemoattractant was added to the lower chambers. Cells were maintained overnight in an incubator at 37 °C with 5% CO2. After 24 h, the Transwell inserts were washed twice with PBS, and the cells were fixed with 4% PFA for 2 min and methanol for 20 min at RT. The cells on the upper side of the inserts were removed using cotton swabs, and the lower side of the membrane was stained with Giemsa solution for 15 min at RT. Invading and migrating cells were observed and imaged using an optical microscope (BX52, Olympus, Japan). 2.11. PLAAT3 siRNA Transfection HSA 2.5D organoid cells were seeded at a density of 105 cells/well in a 6-well plate. After 24 h of culture, cells were transfected with siRNA against PLAAT3 or negative control siRNA (Sequence (5′–3′): si 1, AGCACAUCACACAACAUUA and si 2, AAUGGUGGUUAUUCAGUUC) using Lipofectamine RNAiMAX (Invitrogen) following the manufacturer’s protocol. After 72 h, the cells were harvested, and evaluations were performed via cell proliferation assay, cell invasion assay, and quantitative RT-PCR. 2.12. Statistical Analysis All statistical analyses in this study were performed using GraphPad Prism (11.0.1). Data are shown as mean ± S.E.M. Comparisons between two groups were performed using unpaired Student’s t-test. For multiple group comparisons, one-way ANOVA followed by Dunnett’s multiple comparisons test was used. For experiments involving two independent variables, two-way ANOVA followed by Dunnett’s multiple comparisons test was applied. p values < 0.05 were considered statistically significant. 3. Results 3.1. Pathological Comparison of Patient Tissues with HSA and NH Primary cultured 2.5D organoids were established from canine hemangiosarcoma (HSA) and nodular hyperplasia (NH) tissue following pathological diagnosis (Figure 1A,B). Histopathological examination of the original canine HSA samples revealed irregular vascular channels and atypical cells with enlarged hyperchromatic nuclei, whereas NH exhibited comparatively well-organized splenic structures with minimal atypia. IHC staining demonstrated expression of vWF, CD31 and VEGF in both HSA and NH tissues (Figure 1B). Stronger CD31 and vWF immunoreactivity was observed in HSA tissues than NH lesions. In HSA samples, CD31 staining was associated with irregular vascular-like structures and showed heterogeneous distribution within tumor regions. vWF staining varied among HSA cases, ranging from focal positive staining to weak or nearly absent signals in highly cellular areas. In contrast, NH lesions showed limited or weak staining of all three markers, primarily associated with residual vascular structures. VEGF immunoreactivity was observed in HSA samples, whereas NH tissues exhibited only weak or minimal staining. 3.2. Generation and Drug Sensitivity Test of Patient-Derived HSA 2.5D Organoids Primary cultured HSA 2.5D organoids showed expression patterns of the three markers similar to those observed in the original tissues (Figure 2A). Carboplatin, doxorubicin, toceranib, and cyclophosphamide are commonly used chemotherapies for HSA in pet clinics. Therefore, we conducted drug sensitivity tests using HSA 2.5D organoids to evaluate their in vitro responses to these clinically relevant drugs. We treated HSA organoid lines with toceranib, carboplatin, cyclophosphamide and doxorubicin (Figure 2B). HSA 2.5D organoids derived from different patients exhibited variable drug sensitivities with toceranib, carboplatin and doxorubicin, suggesting interpatient heterogeneity in drug response. 3.3. In Vivo Tumorigenicity of HSA 2.5D Organoids in an Orthotopic Xenograft Model Next, to assess whether HSA 2.5D organoids retained tumorigenic potential in vivo, an orthotopic xenograft model was established by splenic injection (Figure 3A). Ten weeks after splenic injection with HSA 2.5D organoid cells (BS24028) in C.B-17/IcrHsd-Prkdc SCID mice, orthotopic splenic HSA tumors developed in 9/9 mice. Among these mice, liver metastasis was observed in 4 of 9 mice, consistent with the metastatic lesion observed in the original canine patient, while pancreatic seeding was observed in 3 mice (Figure 3B). Histopathological examination using H&E staining revealed abnormal cleft-like structures in xenografted splenic tissues, which morphologically resembled those in the original canine HSA tumor tissue. Enlarged nuclei and extensive infiltrative proliferation of atypical cells were observed inside and outside the spleen lesions (Figure 3C). Additionally, IHC staining demonstrated expression of VEGF, vWF and CD31 in xenografted tumor tissues (Figure 3C). These findings support that the primary cultured 2.5D HSA organoids retained tumorigenic capacity and several histopathological features associated with canine HSA in vivo. 3.4. Comparison of Transcriptional Profiles Between HSA and NH 2.5D Organoids RNA sequencing was conducted to compare transcription profiles between NH and HSA 2.5D organoids (Figure 4A,B). PCA and hierarchical clustering were performed to visualize the overall transcriptomic relationships among the HSA and NH samples (Supplementary Figures S1 and S2). Gene set enrichment analysis of differentially expressed genes (DEGs) revealed enrichment of pathways associated with autophagy, autophagic membrane, lysosomes, and lysosomal membrane in HSA 2.5D organoids (Figure 4C). Genes related to the cellular membrane, lysosomal membrane, and autophagy, including HIP1, PLAAT3, and BNIP3, were found upregulated in HSA 2.5D organoids compared to NH organoids (Figure 4D). Additionally, quantitative real-time PCR analysis results further supported the increased expression of PLAAT3 in HSA 2.5D organoids. In addition, IF staining demonstrated increased PLAAT3 expression in HSA 2.5D organoids (Figure 4E). IHC staining and quantification of PLAAT3 in HSA and NH tissues are also shown in Figure 4F. Among the upregulated genes identified by transcriptomic analysis, PLAAT3 was selected for further investigation because of its consistently elevated expression across multiple analyses. Given its reported association with membrane remodeling, PLAAT3 was considered a potential candidate involved in HSA progression. 3.5. Effects of PLAAT3 Inhibition on Cell Proliferation and Invasion in HSA 2.5D Organoids To further investigate the functional role of PLAAT3 in HSA, pharmacological inhibition was performed using LEI110, and its effects on cell proliferation and invasion in HSA 2.5D organoids were evaluated. LEI110 exhibited varying inhibitory effects on several HSA 2.5D organoid lines (Figure 5A). In the invasion assay, high-dose treatment (10 and 20 μM) of LEI110 significantly reduced the number of invasive cells compared with vehicle-treated controls, suggesting PLAAT3 inhibition may be associated with reduced invasive capacity in HSA 2.5D organoids (Figure 5B). To further validate the effects of PLAAT3 inhibition, we subsequently performed genetic knockdown using siRNA (Figure 5C). Cell viability assay showed a modest inhibitory effect in knockdown cells on day 4 and a significant inhibitory effect on day 7 in each cell line (Figure 5D). In addition, PLAAT3 knockdown remarkably reduced invasive activity compared with negative control siRNA treatment (Figure 5E). 4. Discussion CD31, CD34, and vWF are core endothelial markers commonly investigated in AS and HSA research [24,25], and are valuable in AS diagnosis to confirm endothelial cell origin. Active VEGF expression is observed in canine vascular tumors [26] such as HSA, as well as in breast cancers [27,28]. Although weak staining was occasionally observed in NH lesions, CD31 and vWF immunoreactivity tended to be more heterogeneous and irregular in HSA tissues. The heterogeneous staining pattern observed in HSA may reflect variable differentiation states of neoplastic endothelial cells [29]. In particular, reduced vWF staining in highly cellular regions is consistent with previous reports describing decreased endothelial differentiation in proliferative vascular tumors [30]. 2.5D HSA organoids were successfully generated from patient HSA tissues. In this study, orthotopic splenic injection successfully generated xenograft tumors from canine HSA-derived 2.5D organoids. Histopathological examination demonstrated that the xenograft tumors retained morphological characteristics of the original canine HSA tissues, including vascular-like cleft formation and marked nuclear atypia. Furthermore, immunoreactivity of xenograft tumor tissues for endothelial-associated markers, including vWF [31] and CD31 [32], as well as the angiogenesis-related marker VEGF [33], showed distribution patterns comparable to those observed in the original tumor tissues. These findings suggest that the xenograft tumors preserved key endothelial and malignant vascular characteristics of canine HSA in vivo. Previous studies in human cancers, including colorectal cancer [34,35], renal cell carcinoma [36], and osteosarcoma [37], have demonstrated that orthotopic transplantation models reproduce clinically relevant metastatic behavior more closely than heterotopic transplanted models. In this study, liver metastases were observed in 4 out of 9 xenograft cases, suggesting partial reproduction of metastatic behavior in vivo. Compared with conventional subcutaneous xenograft models, orthotopic splenic implantation may provide a more anatomically relevant tumor environment and enable partial reproduction of metastatic behavior, including liver metastasis. However, metastatic formation was inconsistent among cases, which may be attributable to the limited observation period (10 weeks) or leakage from the injection site. Therefore, further optimization and validation are required to improve the robustness and reproducibility of this metastasis model. Several non-metastatic cases in this study showed pancreatic seeding, adjacent to the injection site, suggesting possible local tumor dissemination. Spontaneous HSA pathology is frequently associated with the coexistence of tumor cells and infiltrating inflammatory cells, which may increase the risk of contamination during direct genetic expression detection in clinical samples. Therefore, the primary cell culture method established in the present study may provide a useful tool to address this issue. Transcriptomic analysis revealed upregulation of lysosome-associated pathways in HSA samples, suggesting altered lysosomal activity. A previous study comparing canine hemangiosarcoma with normal granular cells showed lysosome-like ultrastructural proteinaceous accumulation within single-membrane vesicles [38], supporting lysosomal involvement in HSA. Consistent with these findings, lysosome-related genes associated with lysosomal trafficking and function, including GLMP and GAA [39,40], showed elevated expression trends, whereas PLAAT3 expression was significantly increased in HSA samples compared with NH. PLAAT3, which belongs to the PLAAT family, is known to exhibit phospholipase A2 activity, mediating lipid reprogramming and glycerophospholipid remodeling in cancer cells; PLAAT3 is consistently reported to be highly expressed and activated in various cancers [41,42,43,44]. Previous studies have reported that PLAAT3 is associated with membrane remodeling, cathepsin dysregulation, and increased invasive behavior in several cancers [45]. Consistent with these reports, our PLAAT3 knockdown experiments using siRNA and pharmacological inhibition of PLAAT3 in HSA 2.5D organoids demonstrated significant inhibition of cell invasion. Given that PLAAT3 is related to metastasis and membrane permeability, and is associated with increased drug sensitivity in human osteosarcoma, we carried out a combination treatment cell viability test; a higher sensitivity to each drug was observed upon co-treatment with a PLAAT3 inhibitor. Moreover, lipid accumulation in endothelial cells is related to angiogenesis and angiogenic sprouting [46,47]. PLAAT3 is involved in lipid accumulation within liver cells via its PLA2 activity [48]. These previous findings raise the possibility that the increased PLAAT3 expression observed in HSA may also be associated with angiogenic processes; however, this possibility was not directly investigated in the present study. Although pharmacological inhibition and knockdown of PLAAT3 led to a mild reduction in cell proliferation, our findings primarily support an association of PLAAT3 with cell migration and invasion rather than establishing a direct role in angiogenesis. Therefore, PLAAT3 may contribute to the invasive phenotype of canine HSA and represents a candidate molecule warranting further investigation. However, the present study did not evaluate the association between PLAAT3 expression and clinical outcomes or the therapeutic efficacy of PLAAT3 targeting in vivo. Therefore, further studies incorporating clinical outcome analyses and in vivo PLAAT3-targeting experiments are required to determine the prognostic and therapeutic relevance of PLAAT3 in canine HSA. Our sequencing analysis also identified increased HIP1 expression in HSA compared with NH. HIP1, which is associated with cellular trafficking [49] and tumor-related signaling pathways [50,51], has been reported previously to be up-regulated in human prostate and colon cancers [52]. Although the biological significance of HIP1 in canine HSA remains unclear, the increased expression observed in this study suggests that further investigation may be warranted. Although transcriptome sequencing showed only a modest increase in MARVELD3 expression in HSA compared with NH, RT-PCR analysis showed a significant increase. MARVELD3 from TAMP family is associated with tight junction regulation and epithelial remodeling. It is found to be upregulated in several human cancers [53,54,55]. Although the precise biological roles of HIP1 and MARVELD3 in canine HSA is still unclear, their increased expression in HSA suggested potential involvement in HSA-associated cellular regulation. Somatic mutations in PIK3CA and TP53 have been reported in both HSA and AS; however, transcript expression levels of these genes did not show significant differences from our results. An important limitation of this study is the relatively small number of biologically independent samples. RNA-seq analysis was performed using five HSA and five NH organoid lines, and the functional and xenograft experiments were conducted using a restricted subset of the established HSA lines. Therefore, the present findings should be considered exploratory and may not fully represent the biological heterogeneity of canine HSA. Further studies using larger and independent cohorts are required to validate these findings and determine their generalizability. Another limitation is that, although canine HSA shares several pathological and molecular characteristics with human AS, a direct transcriptomic comparison with human AS datasets was not performed in the present study. Therefore, further cross-species transcriptomic analyses will be required to establish the translational relevance of the molecular findings identified here. In conclusion, we established patient-derived HSA 2.5D organoids that retained tumorigenic potential and identified PLAAT3 as a candidate molecule associated with HSA invasion. Although the underlying mechanisms and on-target effects require further validation, our findings provide proof-of-concept evidence supporting further investigation of PLAAT3 in canine HSA.