Hemangiosarcoma Cells Promote Conserved Host-derived Hematopoietic Expansion

Abstract Hemangiosarcoma and angiosarcoma are soft-tissue sarcomas of blood vessel–forming cells in dogs and humans, respectively. These vasoformative sarcomas are aggressive and highly metastatic, with disorganized, irregular blood-filled vascular spaces. Our objective was to define molecular programs which support the niche that enables progression of canine hemangiosarcoma and human angiosarcoma. Dog-in-mouse hemangiosarcoma xenografts recapitulated the vasoformative and highly angiogenic morphology and molecular characteristics of primary tumors. Blood vessels in the tumors were complex and disorganized, and they were lined by both donor and host cells. In a series of xenografts, we observed that the transplanted hemangiosarcoma cells created exuberant myeloid hyperplasia and gave rise to lymphoproliferative tumors of mouse origin. Our functional analyses indicate that hemangiosarcoma cells generate a microenvironment that supports expansion and differentiation of hematopoietic progenitor populations. Furthermore, gene expression profiling data revealed hemangiosarcoma cells expressed a repertoire of hematopoietic cytokines capable of regulating the surrounding stromal cells. We conclude that canine hemangiosarcomas, and possibly human angiosarcomas, maintain molecular properties that provide hematopoietic support and facilitate stromal reactions, suggesting their potential involvement in promoting the growth of hematopoietic tumors. Significance: We demonstrate that hemangiosarcomas regulate molecular programs supporting hematopoietic expansion and differentiation, providing insights into their potential roles in creating a permissive stromal-immune environment for tumor progression.


Introduction
Canine hemangiosarcoma, a vasoformative tumor originating from bone marrow (BM)-derived progenitor cells, is a common occurrence in dogs, unlike the rare human angiosarcoma (1)(2)(3)(4).Tumors from both species exhibit similar histology and natural history of disorganized, tortuous, and dilated blood vessels with high proliferative activity and metastatic potential (5,6).Molecularly, convergent transcriptional programs involving angiogenesis and inflammation are observed in both canine hemangiosarcoma and human angiosarcoma (7,8).While the tumor microenvironment (TME) plays a crucial role in tumor cell survival, disease progression, and metastasis through the niche it creates (9), its contribution to the molecular programs of hemangiosarcoma remains incompletely understood.Recent evidence suggests bidirectional interactions between cancer cells and niche cells, where cancer cells re-educate niche cells and vice versa, particularly in maintaining stemness and self-renewal of cancer stem cells in various tumors (9)(10)(11), including hematopoietic tumors (12)(13)(14).
The hematopoietic niche is composed of osteoblasts and various stromal cells including endosteal, endothelial cells, fibroblasts, nestin-positive mesenchymal stromal cells (MSC), leptin receptor-positive stromal cells, and CXCL12-abundant reticular cells.These cells support hematopoiesis by regulating the proliferation and differentiation of hematopoietic stem and progenitor cells (HSPC) into lineage-committed cells (15).Dysregulation of HSPC function can result in blood disorders (16,17) and hematopoietic malignancies such as leukemia and lymphoma (18)(19)(20).
In the current study, we aimed to elucidate the cellular and molecular properties of hemangiosarcoma.Our findings indicate that canine hemangiosarcoma cells not only have the capability to form vasoformative tumors but also create a niche for expansion and differentiation of blood cells, potentially contributing to the development of hematopoietic tumors.

Human Tissue Samples
The human angiosarcoma tissue samples used in this study were formalinfixed and paraffin-embedded, as described previously (8).The samples were obtained from two sources: the University of Minnesota Biological Materials Procurement Network and the Cooperative Human Tissue Network.All sample acquisitions were performed under standardized patient consent protocols.

Human and Mouse Cell Lines
Human bone marrow-derived mesenchymal stromal cells (hBM-MSC) were isolated from whole BM purchased from AllCells, as described previously (24)(25)(26)(27).M2-10B4 murine BM stromal cells were purchased from the ATCC and maintained following established protocols (25).Samples of human umbilical cord blood (hUCB) were obtained from the ClinImmune Stem Cell Laboratory at the University of Colorado Cord Blood Bank (28).

Mice and Xenotransplantation
We conducted multiple xenograft experiments using various cell lines and patient-derived tumor sections.First, we injected cultured tumor cells from three hemangiosarcoma cell lines (SB, Emma, and JHE) into NSG mice.Specifically, we injected 5 × 10 6 SB cells subcutaneously in four mice, and 5 or 10 × 10 6 SB cells intraperitoneally in 4 mice.In addition, we injected 2 × 10 6 Emma cells subcutaneously and intraperitoneally in 4 mice each, and 3 × 10 5 JHE cells subcutaneously in 4 mice.
We used sections of viable tumors from 4 dogs affected with hemangiosarcoma and implanted them into subcutaneous pockets of 4 mice for each dog.In addition, we implanted sections of non-hemangiosarcoma splenic tissues from 7 dogs into 18 mice as controls.Finally, after visible tumors developed in mice, we serially transplanted the tumors by inoculation of cultured tumor cells in 3 mice or by direct implantation of single-cell suspensions of the tumor in 8 mice.
We monitored the mice for tumor development and sacrificed them when they reached a tumor endpoint, including a mass measuring 1.5 cm in the longest diameter or at the end of a 16-week period after xenotransplantation.
Mice were euthanized by cervical dislocation according to the IACUC guidelines.All animal procedures were conducted in accordance with the Research Animal Resources husbandry and care protocols and reviewed and approved by the IACUC of the University of Minnesota (Minneapolis, MN; protocols 1006A84813, 1106A00649, 1306-30712A, and 1311-31104A).A total of 132 mice were used for xenotransplantation procedures, as described in Supplementary Table S1.

IHC
IHC staining was conducted at either the BioNet Histology Laboratory, Minnesota Veterinary Diagnostic Laboratory, or Comparative Pathology Shared Resource at the University of Minnesota (Minneapolis, MN).The following antibodies were used: CD3, PAX5, MAC387, CD163, CD204, Iba1, Ter-119, MPO, and CD45.Details of the antibodies used are provided in Supplementary Table S2.

RNA Sequencing for Gene Expression Profiling
To obtain transcriptomic profiles, we isolated total RNA from xenograft tumor samples using the TriPure Isolation Reagent (Roche Applied Science), followed by clean-up using the RNeasy Mini Kit (Qiagen).We generated RNA sequencing (RNA-seq) libraries with a targeted depth of 20 million paired-end reads (2 × 50 bp) at the UMGC, as described previously (4,8).We conducted bioinformatic analyses for gene expression profiling as previously reported (4,8,(29)(30)(31).RNA-seq data from human angiosarcoma tissues can be accessed via the Gene Expression Omnibus (GEO) under accession number GSE163359 (8).RNA-seq data from canine hemangiosarcoma tissues are available through the GEO (accession number GSE95183) and the NCBI Sequence Read Archive (accession number PRJNA562916; ref. 8).

Viral RNA Isolation and Library Generation
To detect RNA viruses, we isolated viral RNA from xenograft tumors using either a Qiagen viral RNA kit or TRIzol followed by the Qiagen viral RNA kit (Qiagen).The isolated RNA was then submitted to the University of Minnesota Genomics Center (UMGC) for viral RNA MiSeq.The Illumina MiSeq generated approximately 1 million paired-end reads (2 × 250) for each sample.

PathSeq Analysis
The PathSeq algorithm was used to determine transmissible agents in RNAseq data from dog and mouse samples.The PathSeq algorithm (32) was used to perform computational subtraction of mouse and dog genomes followed by alignment of residual reads to mouse and dog reference genomes and microbial reference genomes (including bacterial, viral, archaeal, and fungal sequences downloaded from NCBI).These alignments resulted in the identification of microbial reads in the data.
Human reads were subtracted by first mapping reads to a database of mouse (mm9) and dog reference genomes (canFam3.To identify microbial reads, the residual reads were aligned with MegaBLAST to a database of microbial and dog reference genomes.Raw read counts were calculated using the reads that were mapped to Epstein-Barr virus (EBV) and Helicobacter pylori with at least 90% identity and 90% query coverage.Using the raw read counts, the abundance metric or normalized read count of a given microbe in a sample was calculated as

Long-term Culture Initiating Cell and Hematopoietic Colony-forming Unit Assays
We isolated hUCB CD34 + cells from 2 patients using the Miltenyi CD34 Microbead Kit and MACS separation column (Miltenyi Biotec), following the manufacturer's instructions.The purity of isolated hUCB CD34 + cells was determined by flow cytometry using the following antibodies (all anti-human): CD34-PECy7 (eBioscience), CD43-APC (BD Biosciences), and CD45-APC (BD Biosciences).Sorted hUCB CD34 + cells with >92% CD34 + CD45 + were used for subsequent experiments.We seeded M2-10B4, hBM-MSCs, 1426, and Emma in gelatin-coated 24-well plates at a density of 10 5 cells/well overnight.After inactivation with 10 μg/mL Mitomycin C for 3 hours, 5,000 hUCB CD34 + cells were added to each well in Myelocult H5100 (StemCell Technologies) supplemented with 1 μmol/L dexamethasone (Sigma-Aldrich).Gelatin-coated wells without stroma were used as negative controls.Cells were conditioned for 6 weeks at 37°C and 5% CO 2 with one-half media exchanges every week.At week 5, nonadherent cells from selected wells were harvested, counted, and analyzed for hematopoietic-specific surface antigens using the following antibodies (all anti-human): CD34-PECy7, CD33-APC (BD Biosciences), CD43-APC, and CD45-APC.At the end of the 6-week culture, both nonadherent and adherent cells were collected using 0.05% trypsin supplemented with 2% chicken serum for 5 minutes.Cells were filtered using a 70-μm cell strainer to generate singlecell suspensions and counted.We prepared triplicate CFU assays by seeding 50,000 cells/1.5 mL of semisolid Methocult H4435 Enriched media (StemCell Technologies) in 35 mm culture dishes (Greiner).After 2 additional weeks, the plates were manually scored for total number and phenotype of the CFUs, as described previously (36).

Statistical Analysis
Pearson and Spearman correlation coefficient was calculated for correlation between two variables.Statistical analysis was performed using GraphPad Prism 10 (GraphPad Software, Inc.) or Microsoft Excel.

Data Availability
The data generated in this study are publicly available in GEO at GSE163359 and GSE95183, as well as in NCBI Sequence Read Archive (PRJNA562916).Other relevant data generated in this study are available upon request from the corresponding author.

Canine Hemangiosarcoma Cells Can Recapitulate the Disease In Vivo
We used in vivo xenografts to investigate the biological behavior of canine hemangiosarcoma (21,23,(37)(38)(39).Mice were inoculated with canine hemangiosarcoma cell lines or primary tissues, as detailed in Supplementary Table S1.
The engraftment efficiency of canine hemangiosarcoma xenografts was found to be low, consistent with findings from previous studies (37,38).Only cultured DHSA-1426 cells, when injected subcutaneously, consistently generated vasoformative tumors in immunodeficient BNX mice (Fig. 1A and B).The tumorigenic potential of this cell line was maintained over multiple passages, as demonstrated by serial passaging of cells cultured from tumor xenografts into new recipient mice (Supplementary Fig. S1).
To establish the contribution of stromal elements to the formation of canine hemangiosarcoma, we utilized FISH to quantify the presence of both canine and mouse cells in the tumors.Specifically, we employed probes for canine CXCL, as this gene is absent in the mouse genome, and a unique region of the mouse X chromosome, to differentiate between canine and mouse cells, respectively.Our data revealed that the hemangiosarcoma xenografts displayed a complex topological organization, with blood vessels lined by both donor and host cells (Fig. 1C; Supplementary Fig. S2).The hemangiosarcoma tumors were comprised of a mixture of 50%-70% malignant canine cells and 30%-50% mouse stromal cells (Fig. 1D).Furthermore, we sought to quantify the microenvironment components in original hemangiosarcoma tissues using RNA-seq data.
Our cell type enrichment analysis revealed a mean microenvironment score of 42%, ranging between 14% and 70% across all 76 tumors.The DHSA-1426 tumor, which was used for the hemangiosarcoma xenograft, contained 37.5% of the estimated microenvironment score (Fig. 1E).These findings highlight the utility of xenografts as a valuable tool for quantifying the contribution of stromal elements to the TME.
To gain a deeper understanding of the involvement of stromal elements in canine hemangiosarcoma xenografts, we analyzed RNA-seq data obtained from cells and xenografts.The RNA-seq data were mapped to a reference genome that included both canine and mouse cells (29).Notably, we observed that the RNA-seq data from the cell lines predominantly mapped to the canine genome, whereas the xenografts exhibited mapping to both the canine and mouse genomes, which was consistent with the results obtained from FISH (Supplementary Fig. S3; Supplementary Table S3).

Canine Xenografts Generate Lymphomas of Mouse Origin
An unexpected series of findings shed light on the relationship between hemangiosarcoma and inflammation in the microenvironment.Previous studies, including our own, have demonstrated that canine hemangiosarcoma cells can successfully form hemangiosarcomas in athymic nude, NOD, and NSG mice (21,23,(37)(38)(39).However, the studies are all indicative of low tumor take rates.For instance, in one series of transplantations, only two of 35 hemangiosarcoma cases achieved tumor formation in immunodeficient mice (38).In the current study, we noted instances where NSG mice, following inoculation with SB hemangiosarcoma cells or another cell line named Emma-brain (EFB), and BNX mice, following inoculation with tumor fragments derived from sample DHSA-1426 (freshly obtained from a dog with spontaneous hemangiosarcoma), developed exuberant myeloid hyperplasia and/or round cell tumors.Notably, 4 out of 5 mice that received 5 × 10 6 SB hemangiosarcoma cells intraperitoneally and 1 out of 4 mice that received 2 × 10 6 EFB cells subcutaneously, succumbed acutely 2 weeks after inoculation, displaying signs of anemia and splenomegaly.Histologic examination revealed that the spleens of these mice were expanded by monomorphic populations of hematopoietic cells (Fig. 2A and B).Upon further analysis, the cells were determined to be of mouse origin and represented erythroid progenitors (Ter-119+), with a few canine hemangiosarcoma cells admixed in the population (Fig. 2C-F).However, we were unable to definitively establish whether these cells had undergone malignant transformation.
Three of 4 mice that received DHSA-1426 tumor fragments, representing firstgeneration canine patient-derived xenograft (CPDX), developed tumors in multiple organs, including spleen, lymph nodes, meninges, cerebrum, and mesentery, 12 weeks after implantation.However, the morphology of these tumors resembled that of round cell tumors (Supplementary Fig. S4A and S4B).IHC analysis showed that the tumor cells expressed CD45, B220, and Pax5, but did not express CD3, Ter-119, and MPO (Supplementary Fig. S4C-S4H), indicating a mouse B-cell origin.To further verify the mouse origin of these cells, flow cytometry was conducted, revealing positive staining for mouse CD45, but not for canine CD45 or human αVβ3-integrin (CD51/CD61), which exhibits cross-reactivity with canine, but not with mouse αVβ3-integrin (Supplementary Fig. S5).Furthermore, RNA-seq data from the mouse round cell tumors demonstrated a near-complete alignment with the mouse genome, consistent with the findings from flow cytometry (Supplementary Fig. S3), thereby indicating that the tumors were derived from mouse cells.
The malignant B cells demonstrated the ability to undergo serial passage and establish B-cell lymphomas in recipient BNX mice independently, without the requirement for canine hemangiosarcoma cell support (Supplementary Fig. S6).
Intriguingly, similar results were obtained when single-cell suspensions derived from fresh tumor fragments of canine hemangiosarcoma xenografts were inoculated into BNX recipients, representing second-generation CPDX derived from a first-generation cell line tumor.The resulting tumors of mouse B-cell origin displayed comparable morphology and were likewise amenable to serial passage in BNX mice.
The potential etiology of these tumors was perplexing.To investigate the possibility of a transmissible, infectious agent as the cause of these tumors, we sequenced RNA from the xenografts and used the PathSeq platform for analysis.However, no bacterial or viral sequences with tumorigenic potential were identified in the xenografts or in the primary or metastatic canine hemangiosarcoma tumor samples or cell lines.While a recent study reported an association between Bartonella spp.and canine hemangiosarcoma (40,41), only 10 of 24 dogs tested in our study had detectable Bartonella spp.sequences, and these were present in low abundance (Supplementary Table S4).Furthermore, four of the samples contained sequences from B. bacilliformis, B. grahamii, or B. tribocorum, which infect humans and rats, respectively, and are not known to infect dogs as primary or accidental hosts (42,43).Three of the remaining 6 dogs had sequences for B. clarridgeiae, and 3 had sequences for multiple Bartonella spp., including human and rodent-specific types.The low abundance and the presence of sequences from organisms that do not infect dogs suggest that the Bartonella spp.sequences were contaminants.Furthermore, we identified murine leukemia virus (MuLV) reads in mouse B-cell lymphomas that arose from the hemangiosarcoma xenografts through independent viral RNA-seq and PCR experiments (Supplementary Table S5).However, MuLV sequences were detected in normal mouse tissues (liver and spleen), as well as in two subcutaneous canine hemangiosarcoma xenograft tumors.MuLV was also found in SB hemangiosarcoma cells that had been previously passaged through mice as hemangiosarcomas.These findings indicate the transferability of MuLV sequences to xenografted tumors but are insufficient to support any conclusion regarding the role of a virus(es) as the transforming agent(s) responsible for the development of mouse lymphomas.

Human and Canine Hemangiosarcoma Tissues Show Evidence of Immune Cells
We reanalyzed RNA-seq data obtained from both canine hemangiosarcoma and human angiosarcoma samples, with the aim of investigating the presence of immune cells within the TME.To identify coexpressed genes, we used an unbiased approach called GCESS (Gene Cluster Expression Summary Score; ref. 30).No-tably, the highest expression of these immune signature genes was observed in the inflammatory subtype of canine hemangiosarcoma (Supplementary Fig. S7).We next segregated human angiosarcomas into tumors with high and low immune scores ("immune-high" vs. "immune-low") and identified 461 upregulated genes (P adjusted < 0.05) in the immune-high group compared with immune-low group (Fig. 3A).Fifty-eight of these genes were also found in (Continued) with signaling pathways of immune cell functions between human and canine tumors.Representative photomicrographs of H&E and IHC staining showing histologic morphology and immune cell infiltration in canine hemangiosarcoma (D) and human angiosarcoma tissues (E) using anti-CD3, anti-PAX5, anti-MAC387, and anti-Iba1 (for canine) or anti-CD163 (for human) antibodies for detecting T cell, B cell, and macrophages.Horseradish peroxidase (counterstain = hematoxylin) or alkaline phosphate (for Iba1; counterstain = methylene blue).Bar = 50 μm.Scatter plots display correlation between transcriptional and IHC immune score in canine hemangiosarcoma (F) and human angiosarcoma (G).Spearman correlation coefficient (R) was calculated.
canine inflammatory hemangiosarcomas, and they were associated with Tand B-cell activation (Fig. 3B and C).This approach allowed us to identify a common immune signature in both human and canine tumors.
To confirm whether the inflammatory gene signatures were originating from the malignant cells themselves or from inflammatory cells within the tumors, we employed bioinformatics tools and immunostaining techniques.Tumor purity was assessed using ESTIMATE and immune scores were assigned to each tumor using xCell.Both tools yielded consistent scores (Supplementary Fig. S8A and S8B).As expected, we found a correlation between immune scores and the predominant transcriptional phenotype of the tumor in both canine and human samples (Supplementary Fig. S8C-S8F).Specifically, angiogenic tumors exhibited low immune scores, whereas inflammatory tumors exhibited high immune scores.
To further validate that the observed immune signatures were the result of immune and inflammatory cells present within the tumors, we performed IHC staining on sections from 11 canine hemangiosarcomas and 10 human angiosarcomas.Antibodies against T cells (CD3), B cells (Pax5), myeloid cells (Mac387), and macrophages (Iba1 for canine; CD163 for human) were used; CD3 + T cells, Pax5+ B cells, Mac387+ myeloid cells, and Iba1+ or CD163 + macrophages were present in both canine and human tumors (Fig. 3D and E).Myeloid cells were the most abundant, while T cells varied in frequency from rare to abundant, and B cells were infrequent.Importantly, the distribution of inflammatory cells was diffuse throughout the tumor tissue.Furthermore, there was a direct correlation between xCell immune scores and IHC scores for both canine (Spearman R = 0.38; P = 0.255) and human (Spearman R = 0.78; P = 0.011) samples that were examined (Fig. 3F and G).These findings provide evidence that the observed immune signatures in both canine and human tumors are likely attributed to the presence of immune and inflammatory cells within the TME.

Canine Hemangiosarcoma Cells Promote Hematopoiesis and Express Hematopoietic Cytokines
To determine whether hemangiosarcoma cells were functionally sufficient to expand hematopoietic progenitor cells (HPC) in vitro, we conducted long-term culture initiating cell assays using DHSA-1426 and EFB canine hemangiosarcoma cells, to assess their potential in promoting and maintaining hematopoiesis in CD34 + human umbilical cord blood HPCs.Mouse M2-10B4 and human BM-derived MSCs were used as positive controls, while HPCs cultured without feeder cells served as a negative control.Remarkably, DHSA-1426 cells were found to promote expansion of human CD34 + HPCs with at least comparable, if not superior, efficiency compared with conventional mouse or human feeder cells, and resulted in comparable proportions of hematopoietic cell differentiation in vitro across all lineages (Fig. 4; Supplementary Fig. S9).
Similar results were observed with EFB cells, albeit with slightly more limited expansion and differentiation.
To further explore the molecular properties that govern hematopoiesis, we analyze RNA-seq gene expression data from DHSA-1426 hemangiosarcoma cells compared with nonmalignant endothelial cells.Our analysis revealed 2,391 differentially expressed genes (DEG), with 1,034 genes upregulated and 1,357 genes downregulated in DHSA-1426 cells (P adjusted < 0.01; log 2 fold change >|2|; Fig. 5A).Gene Ontology enrichment analysis highlighted the involvement of these genes in key biological processes including blood vessel morphogenesis, angiogenesis, cell differentiation, and extracellular matrix organization (Fig. 5B).Notably, DHSA-1426 cells exhibited an enrichment of marker genes for endothelial progenitors, such as PECAM, TIE, and KDR (Supplementary Fig. S10).Moreover, we observed a significant increase in cytokine genes associated with the regulation of HSPCs including CSF, IL, and IL in DHSA-1426 cells (Fig. 5C).We then investigated whether stromal cells in xenograft tumors express genes encoding receptors for these hemangiosarcoma-secreting cytokines.Our RNA-seq data enabled the identification of canine-and murinespecific genes in the xenograft tissues (Supplementary Fig. S3; Fig. 5D).Using these data, we discovered that hemangiosarcoma xenografts significantly increased murine-specific receptor genes, Csfr, Ilra, and Cxcr, which bind to Csf, Il, and IL8 homologous genes (Cxcl, Cxcl, and Cxcl), compared with mouse lymphomas (Fig. 5E).Furthermore, the expression of receptor genes was higher than that of their respective cytokine genes in the hemangiosarcoma xenografts.This indicates strong intercellular interactions and potential signal transduction in close proximity between donor canine hemangiosarcoma cells and host mouse stromal cells within the tumors, supporting the findings in our FISH data (Fig. 1C; Supplementary Fig. S2).
Altogether, our data suggest that hemangiosarcoma cells regulate hematopoiesis and tumor stromal cells through the release of immune cytokines, potentially contributing to the creation of their distinctive vascular and immune niche.

Discussion
Vasoformative sarcomas are aggressive tumors with uncertain cellular origin, proposed to arise from a multipotent BM cell or lineage-committed endothelial progenitor cells in humans, dogs, and mice (1-4, 8).While human angiosarcomas and canine hemangiosarcomas have a limited shared mutational spectrum, primarily in visceral forms of the disease and human breast angiosarcomas, convergent transcriptional programs characterized by deregulation of PI3K pathways are activated in tumors of both species, as well as in zebrafish (8,(44)(45)(46)(47)(48)(49)(50).
The transcriptional landscape of human angiosarcoma and canine hemangiosarcoma is strongly proangiogenic.However, a subset of tumors in both species exhibit robust transcriptional immune and inflammatory signatures that correlate with the presence of T cells and macrophages.The presence of immune and inflammatory infiltrates might be associated with longer survival outcomes, but further research is needed to determine their clinical significance regarding disease progression and their association with the approximately 15% of exceptional survivors reported in canine hemangiosarcoma (51,52).
The findings from our xenograft experiments provide further evidence supporting the concept that the disordered vascular organization in these tumors is driven by the tumor cells (1,21,23).Our experiments reveal that the tumor vessels are comprised of both malignant tumor cells and nonmalignant host endothelial cells, a phenomenon that appears to be unique to hemangiosarcoma among the three types of xenografts we investigated.This observation underscores the ability of hemangiosarcoma cells to adopt endothelial functions and suggests that nonmalignant cells play a role in the formation of aberrant blood vessels in vasculogenic tumors.Interestingly, our experiments also demonstrate that the stromal cells in these xenografts contribute to the angiogenic and inflammatory transcriptional signatures.This highlights the possibility that the stromal cells themselves are also influenced or reprogrammed by the malignant cells and supports a complex interplay between the tumor and its microenvironment.These findings highlight a potential vulnerability in the formation of the hemangiosarcoma niche, where the interactions between the tumor cells and their microenvironment are tightly orchestrated.
The initial perplexity surrounding the development of exuberant myeloid and erythroid hyperplasia, as well as bona fide lymphomas, arising from mouse cells in animals with primary or secondary hemangiosarcoma xenografts has led us

FIGURE 6
Potential mechanisms that organize malignant vessels and establish distinct immune phenotypes in hemangiosarcoma.Hypothetical models illustrate that hemangiosarcoma cells may orchestrate with nonmalignant stromal cells to create malignant vascular channels through close intercellular cross-talk.Malignant vessels may facilitate the recruitment of hematopoietic progenitors, delivering of signaling molecules that promote lineage commitment of immune cells.This process enables hemangiosarcoma to create a niche for hematopoietic expansion and inflammation, establishing a vascular malignancy with a distinct immune phenotype or potentially giving rise to a hematopoietic tumor.
to investigate the underlying etiology.Our results suggest that a transmissible etiology from dog to mouse is unlikely to be the cause of these expanded hematopoietic cell populations.Instead, our data indicate that canine hemangiosarcomas have the capability to support robust expansion and differentiation of HPCs in vitro, which may account for the expansion of myeloid cells in vivo in mice and in primary canine and human tumors.This expansion of nonmalignant HPCs and mature leukocytes may also explain why clonal mutations (46) and fusions (8) are found less often in the inflammatory subtype of canine hemangiosarcoma.These findings are consistent with a previous report of angiosarcoma in the BM of a human patient with tumor-associated myeloid AACRJournals.orgCancer Res Commun; 4(6) June 2024 proliferation and extramedullary hematopoiesis (53).In the case of lymphomas that occurred in animals, MuLV might have driven the transformation of residual lymphoid elements within a hyperproliferative environment created by the hemangiosarcoma cells, but the precise etiology needs further investigation.
Other unexpected tumors have been reported in xenograft experiments and preclinical models of stem cell transplantation.For example, transplantation of murine MSCs has been shown to induce tumor formation and tissue malformation, possibly due to genetic instability and/or cellular transformation (54)(55)(56).Similarly, patient-derived xenografts of human solid cancers, such as breast, colon, pancreatic cancer, and rhabdomyosarcoma, have been reported to induce lymphomagenesis or lymphocytic tumors in immunodeficient mice, but in these cases, the tumors were derived from human tumor-infiltrating lymphocytes transformed by EBV (57)(58)(59)(60).Importantly, these previously reported tumors were all of donor origin, while the tumors observed in our study originated from the mouse recipients and were distinct from the donor hemangiosarcomas.
We were unable to find any reports of hematopoietic tumors of recipient origin arising from xenotransplantation experiments using other types of canine cancers in the literature, and we have not observed such events in our own studies (29,61,62).In seeking feedback for our conclusions, we learned that Dr. Stuart Helfand had observed similar results when he transplanted the SB-HSA canine hemangiosarcoma cell line in NOD mice; however, he did not report these observations at the time (Dr.S. Helfand, personal communication).Thus, this finding appears to be unique to canine hemangiosarcoma and may be attributed to the ability of hemangiosarcoma cells to support hematopoietic expansion.
Our findings also suggest that the normal counterparts of canine hemangiosarcoma cells might contribute to the development of hematopoietic malignancies through the creation of a permissive niche.
In this context, it is noteworthy that a shared region of the canine genome has been found to be significantly associated with B-cell lymphomas, hemangiosarcomas, and other blood-derived tumors in various breeds of dogs belonging to distinct genetic clades (63)(64)(65).This finding highlights the intriguing possibility of a molecular connection between these seemingly distinct tumors, suggesting that disruptions in hematopoiesis and cellular reprogramming may contribute to their development (66).It also sets the origin of this potential genomic vulnerability prior to the derivation of modern dog breeds.
The capacity of the hematopoietic niche to accommodate BM transplants and adoptive cell therapies has revealed its resilience, with BM stromal cells exhibiting high resistance to chemotherapy and radiation.These intrinsic properties may explain the relatively poor long-term responses of human patients with angiosarcoma and dogs with hemangiosarcoma to cytotoxic therapies, and they could provide opportunities for the development of more effective treatments.
However, it is important to recognize that therapies targeting the hematopoietic niche may also carry the potential for high toxicity.
Our data present a novel model to elucidate the capability of canine hemangiosarcomas, and potentially human angiosarcomas, to regulate hematopoiesis.We propose that the malignant endothelial cells have the capability to create niches that promote angiogenic proliferation and hematopoietic expansion, contributing to distinct tumor immune and inflammatory phenotypes, as illustrated in Fig. 6.The robust inflammation observed in some of these tumors may therefore be intrinsic to the tumor itself, rather than solely due to extrinsic factors associated with tissue disruption.These paths of differentiation may also influence the biological behavior of the tumors, with those exhibiting strong angiogenic propensity displaying more aggressive behaviors (8).
In conclusion, our study adds to the growing body of evidence demonstrating the complex interactions between tumor cells and the host microenvironment.Further investigations are warranted to unravel the underlying mechanisms that drive these observations and ascertain their significance in the development of human cancer.

⎠
mapped to the microbe #reads in the sample Average #reads in the sample cohort * Genome size o f the microbe Average genome size of the microbes in that kingdom⎞Relative abundance in a given sample was calculated as abundance metric of taxa divided by the total abundance metric at kingdom level of the sample.

FIGURE 1
FIGURE 1 Establishment of xenografts derived from canine hemangiosarcoma in immunodeficient mice.A, Schematic illustration depicts process of tumor xenografts in BNX mice.B, DHSA-1426 tumor cell line was established from a canine patient diagnosed with hemangiosarcoma (A).Cells from DHSA-1426 passages 5 (p5) and 14 (p14) formed tumors histologically classified as hemangiosarcoma (B and C).Cells cultured from xenograft-derived tumors developed histologically identical tumors after serial transplantation (D).Both the primary tumor (E) and the xenograft tumor (F) were positive for CD31 IHC staining.A-D, Hematoxylin and eosin (H&E) stain.E and F, IHC with an anti-CD31 antibody (alkaline phosphatase conjugates; counterstain = hematoxylin).Bar = 200 μm.C, FISH images using canine-specific (CXCL8, red) and mouse-specific (X chromosome, green) probes in a canine hemangiosarcoma xenograft transplanted into receptive immunodeficient female mouse hosts.Red and green arrows point to representative xenograft canine tumor cells and mouse stromal cells, respectively, to aid in identification.D, Individual points on graph represent relative quantity of donor (dog) and host (mouse) cells in each tumor type.10-12 fields of pictures at high magnification (400X) per slide were acquired.A total of approximately 1,000 cells in individual xenograft tumor was counted, and the percentages for each species-specific cells are presented.E, Dot plot shows microenvironment scores for 76 primary hemangiosarcoma tissues, estimated using RNA-seq data and xCell algorithm.The red dot indicates a DHSA-1426 tumor tissue used for xenograft.

FIGURE 2
FIGURE 2 Hematopoietic expansion derived from adoptive transplantation of canine hemangiosarcoma in immunodeficient mice.A and B, Xenotransplantation of canine hemangiosarcoma cell lines created exuberant myeloid hyperplasia in mouse spleens.Representative photomicrographs show histopathology by H&E staining of spleens from NSG mice transplanted with hemangiosarcoma cell lines SB (A) and EFB (B).C, Immunoreactivity of anti-mouse Ki-67 (Tec-3) antibody shows strongly positive signal in proliferating cells in the spleen.D, Immunostaining of anti-human (and canine cross-reactive) Ki-67 (MiB-1) antibody shows lack of positive staining among proliferating cells.E, The proliferating cells are immunoreactive with anti-mouse Ter-119 antibody.F, SB cells expressing Luciferase are detected in mouse spleen (arrows).Images shown in C, D, E, and F are from IHC staining done in mice inoculated with SB cells modified to express GFP and firefly Luciferase.Horseradish peroxidase (for Ki-67 stains) and alkaline phosphatase (for Ter-119 and Luciferase) conjugates were used.Counterstain = hematoxylin.A-D: Bar = 200 μm; E-F: Bar = 50 μm.

FIGURE 3
FIGURE 3 Immune cell infiltration and comparative immune signatures between canine hemangiosarcoma and human angiosarcoma.A,A total of 461 upregulated genes were identified in immune-high (N = 8) compared with immune-low (N = 5) groups in human angiosarcomas (FDR P value < 0.05).B, 567 immune gene signatures were identified among three molecular subtypes of canine hemangiosarcomas (N = 76; FDR P value < 0.001; fold change > 3).The heat maps show upregulated (red) and downregulated (green) genes by unsupervised hierarchical clustering (average linkage; mean-centered; log 2 transformed).C, Venn diagram shows 58 common genes associated (Continued on the following page.)

FIGURE 5
FIGURE 5 Gene expression analysis in DHSA-1426 hemangiosarcoma cells and xenograft tumors using RNA-seq data.A, The volcano plot visualizes 2,391 significant DEGs (1,034 upregulated and 1,357 downregulated) in DHSA-1426 cells (n = 3) compared with nonmalignant endothelial cells (n = 4; q-value < 0.01; log 2 fold change > |2|).The DEGs are indicated by red dots.NS = not significant; log 2 FC = log 2 fold change.B, Gene Ontology enrichment analysis depicts biological processes associated with significant DEGs.P, adjust = adjusted P-value.C, Bar graphs present normalized gene expression values of CSF3, IL6, and IL8 in RNA-seq data generated from canine nonmalignant endothelial cells (n = 4) and DHSA-1426 hemangiosarcoma cells (n = 3).D, Schematic illustration visualizes the experimental steps for species-specific gene expression analysis of xenograft tumors using RNA-seq data.E, Bar graphs show the expression of mouse-specific genes in RNA-seq data generated from xenograft tumor tissues of canine hemangiosarcoma (n = 4) and mouse lymphoma (n = 7).A two-way ANOVA test was conducted to compare the means between groups.****, P < 0.0001