Induction of Histiocytic Sarcoma in Mouse Skeletal Muscle

Myeloid sarcomas are extramedullary accumulations of immature myeloid cells that may present with or without evidence of pathologic involvement of the bone marrow or peripheral blood, and often coincide with or precede a diagnosis of acute myeloid leukemia (AML). A dearth of experimental models has hampered the study of myeloid sarcomas and led us to establish a new system in which tumor induction can be evaluated in an easily accessible non-hematopoietic tissue compartment. Using ex-vivo transduction of oncogenic Kras(G12V) into p16/p19−/− bone marrow cells, we generated transplantable leukemia-initiating cells that rapidly induced tumor formation in the skeletal muscle of immunocompromised NOD.SCID mice. In this model, murine histiocytic sarcomas, equivalent to human myeloid sarcomas, emerged at the injection site 30–50 days after cell implantation and consisted of tightly packed monotypic cells that were CD48+, CD47+ and Mac1+, with low or absent expression of other hematopoietic lineage markers. Tumor cells also infiltrated the bone marrow, spleen and other non-hematopoietic organs of tumor-bearing animals, leading to systemic illness (leukemia) within two weeks of tumor detection. P16/p19−/−; Kras(G12V) myeloid sarcomas were multi-clonal, with dominant clones selected during secondary transplantation. The systemic leukemic phenotypes exhibited by histiocytic sarcoma-bearing mice were nearly identical to those of animals in which leukemia was introduced by intravenous transplantation of the same donor cells. Moreover, murine histiocytic sarcoma could be similarly induced by intramuscular injection of MLL-AF9 leukemia cells. This study establishes a novel, transplantable model of murine histiocytic/myeloid sarcoma that recapitulates the natural progression of these malignancies to systemic disease and indicates a cell autonomous leukemogenic mechanism.


Introduction
Myeloid sarcomas (also known as chloromas) are extramedullary tumors composed of myeloid lineage cells. Myeloid sarcomas typically present in the setting of acute myeloid leukemia (AML) or in conjunction with transformation of a myelodysplastic syndrome (MDS) [1]. Myeloid sarcomas without bone marrow or peripheral blood involvement often precede the development of new or recurrent leukemia [2][3][4]. Myeloid sarcomas arise predominantly in the bone, soft tissue, lymph nodes, and skin, but essentially any part of the body can be affected [5][6][7][8]. Treatment of these malignancies generally follows the same therapeutic algorithms established for their systemic, leukemic counterparts and may additionally involve local radiation [2]. The prognostic significance of myeloid sarcoma at first diagnosis of AML remains somewhat unclear. An association with less favorable disease outcomes has been discussed [9,10], and a recent paper showed that orbital and CNS (central nervous system) myeloid sarcoma in children have a significantly better survival than myeloid sarcoma at other organ sites or AML without myeloid sarcoma [11].
The relative dearth of knowledge regarding the biology of myeloid malignancies arising in extramedullary tissues led us to comparatively evaluate myeloid tumors initiated in either skeletal muscle or in blood following introduction of identical oncogenetic lesions (i.e. oncogenic KrasG12(V) and loss of p16 Ink4A /p19 Arf ). Both oncogenic lesions are strongly associated with human and mouse hematopoietic malignancies [12][13][14][15][16]. Moreover, constitutively activated Kras(G12D) combined with p16/p19 deficiency induces aggressive cancers in a number of non-hematopoietic tissues and organs in mice [17][18][19][20][21][22][23][24]. We therefore introduced oncogenic Kras into p16p19 2/2 bone marrow cells by ex vivo gene transduction, and then transplanted these genetically altered cells to induce systemic leukemias (by retro-orbital injection), as well as to produce the first transplantable model of murine histiocytic sarcoma (by injection into the gastrocnemius muscles of NOD.S-CID mice). Irrespective of transplantation location, tumor cells shared similar morphological and phenotypic features, and histiocytic sarcomas initiated in mouse skeletal muscle seeded systemic disease within weeks of emergence, recapitulating the leukemic progression seen in humans. Finally, murine histiocytic sarcomas could be induced using genetic lesions distinct from p16/ p19 2/2 ; Kras, i.e. MLL-AF9, thereby suggesting that the murine histiocytic induction model described here can be applied to study a broad spectrum of hematopoietic malignancies. In summary, this work suggests that the phenotype of hematopoietic neoplasms is largely independent of the tissue environment in which they develop and provides a rapid and reproducible platform to generate and study the behavior of extramedullary hematopoietic tumors.

Results
Loss of p16 Ink4A /p19 Arf cooperates with oncogenic Kras(G12V) to induce leukemia Bone marrow (BM) cells were isolated from p16p19 2/2 mice, infected with Kras(G12V) in a GFP-tagged pGIPZ lentivirus, and injected retro-orbitally into immunodeficient NOD.SCID mice. All recipient mice showed significant weight loss, anemia and splenomegaly, and were moribund 35-60 days post injection (32 mice evaluated in 4 independent experiments, Fig. 1A-C). In contrast, p16p19 2/2 BM cells infected with control (Ctrl) virus (GFP-tagged empty pGIPZ vector) failed to induce leukemia in 12 out of 14 recipients (2 of the 14 recipients died without clinically apparent tumors but could not be subjected to necropsy due to autolysis) (Fig. 1A). Likewise, wild-type (WT) C57BL/6 BM cells infected with Kras(G12V) induced leukemias in only 2 out of 10 injected NOD.SCID mice. Thus, consistent with previous reports [15,[25][26][27][28], the combination of oncogenic Kras and p16p19deficiency potently drives leukemogenesis in mouse BM cells, whereas either of these two lesions alone shows limited leukemogenic potential within an 8 to 10-week follow up time. Significantly, prior reports indicate that even with longer follow up (8-9 months), p16p19-deficiency or oncogenic Kras alone produces hematopoietic neoplasms (mostly B-or T-lymphomas and T cell leukemias) with relatively low efficiency [15,29].
The spleen and liver of leukemic mice originally injected with p16p19 2/2 ; Kras(G12V) BM cells, exhibited extramedullary hematopoiesis and massive infiltration by intermediate to large size cells with oval, irregularly folded nuclei, prominent nucleoli, and a moderate to large amount of eosinophilic cytoplasm most consistent with involvement of a non-lymphoid hematopoietic malignancy (Fig. 1D). In contrast, the bone marrow of these mice contained variable foci of immature intermediate to large size cells with round or oval, irregular nuclei, prominent nucleoli and moderate cytoplasm. Peripheral blood smears revealed polychromasia and marked reticulocytosis, suggestive of extramedullary hematopoiesis; however, no blasts were noted in the peripheral blood (data not shown). Taken together, this constellation of findings is consistent with the development of murine histiocytic leukemia, equivalent to human acute myeloid leukemia, after retro-orbital injection of p16p19 2 / 2 ; Kras bone marrow cells [30].
The immunophenotype of tumor-derived GFP+ cells in the bone marrow (containing 5.262.3% GFP+ cells; n = 10) and spleen (containing 10.763.5% GFP + cells; n = 10) of leukemic mice was evaluated by flow cytometry.  Fig. 1E) was low or absent. Variable levels of CD8 (68.8614.9%) and CD71 (58.6616.8%) were seen in a subset of mice (10 out of 32 animals examined). Although these data indicate substantial heterogeneity in the leukemic clones propagated in vivo, this antigen expression pattern is consistent overall with an acute leukemia of myeloid phenotype.
Microscopic examination of tumor sections revealed extensive infiltration of tumor cells into the muscle and surrounding adipose tissue, with nearly complete destruction of normal muscle architecture. Infiltrating cells were immature-appearing intermediate to large size cells with round or irregular nuclei, prominent nucleoli and moderate to large amounts of cytoplasm (Fig. 3). Scattered macrophages also were noted. Immunohistochemical staining of primary tumor samples demonstrated that the tumor cells were GFP + (consistent with flow cytometry analyses (Fig. 2D) and with derivation from cells infected by Kras(G12V)-GFP virus) and stained strongly for Mac2, consistent with a myeloid phenotype. Tumors lacked expression of CD34, B220 and Ter119 (Fig. 3), but showed clearly detectable expression of myeloperoxidase (MPO), which marks myeloid cells. Finally, consistent with their aggressive growth kinetics, tumors exhibited extensive reactivity for the cell proliferation marker Ki67 (detected in 51.968.8% of all hematoxylin counter-stained cells, n = 8 slides) (Fig. 3). Based on their morphology and staining pattern, the muscle tumors arising from p16p19 2/2 ; Kras(G12V) BM cells were classified as histiocytic sarcomas (equivalent of human myeloid sarcomas).

Dissemination of histiocytic sarcoma cells
Clinically, myeloid sarcomas often coincide with or precede the diagnosis of new or recurrent acute myeloid leukemia [31]. Review of peripheral blood cell morphology of tumor bearing mice revealed polychromasia and marked reticulocytosis consistent with extramedullary hematopoiesis, as well as scattered immature monocytic cells, consistent with monoblasts, in 2 animals (data not shown). We also examined non-muscle hematopoietic tissues for the presence of GFP + tumor cells. Nests of myeloid-appearing cells were noted in the bone marrow, spleen and liver of tumor-bearing animals (Fig. 4A). Enlarged spleens were observed in all tumorbearing mice (Fig. S1B). The immunophenotype of GFP + cells in the bone marrow and spleen of sarcoma-bearing mice was evaluated by flow cytometry. GFP + cells were CD48 hi (95.463.1% in bone marrow, 98.960.9% in spleen), CD47 hi (93.466.1% in bone marrow, 89.2611.7% in spleen), Mac1 mid/hi (79.1612.5% in bone marrow, 70.5617.9% in spleen), and B220 2/lo , CD4 2/lo , CD8 2/lo , Ter119 2/lo , and CD71 lo/mid (Fig. 4B, 4C). This immunophenotype is highly consistent with the phenotype of GFP+ cells isolated from primary histiocytic sarcomas induced by intramuscular injection of p16p19 2/2 ; Kras(G12V) BM cells (see Fig. 2D), as well as the phenotype of GFP + leukemia cells derived by retro-orbital injection of BM cells modified by the same oncogenetic lesions (see Fig. 1E). Thus, induction of histiocytic sarcomas in skeletal muscle ultimately results in systemic dissemination with involvement of bone marrow, spleen and liver, thereby recapitulating the natural  Table 1. Summary of intra-muscular injection in NOD.SCID mice.  course of myeloid sarcomas in humans preceding manifestation of systemic leukemias [30].

Histiocytic sarcoma is highly transplantable
The rapid growth of p16p19 2/2 ; Kras(G12V) histiocytic sarcoma suggested that the initial seeding cell population contained clones with leukemia-propagating ability. To test directly the tumorpropagating potential of p16p19 2/2 ; Kras(G12V) histiocytic sarcomas, GFP + cells from primary tumors induced in the muscles of primary NOD.SCID recipients were sorted using FACS and defined numbers of sorted cells were transplanted into the cardiotoxin pre-injured gastrocnemius muscle of secondary NOD.SCID recipients (Fig. 5A). The frequency of tumorpropagating cells within the GFP+ population was assessed by limiting dilution analysis (using 200 to 100 000 cells per injection). All mice receiving 100 000 (12/12 mice injected, 3 independent experiments) or 10 000 (10/10 mice injected, 2 independent experiments) GFP+ primary sarcoma cells developed tumors within 25-40 days from injection ( Table 1, Fig. S1C). These studies suggest that the latency of secondary tumor formation is markedly shorter than that of primary tumor formation (30-50 days for tumor induction with 56106 p16p192/2; Kras(G12V) BM cells, Table 1). The smallest number of GFP+ cells giving rise to a histiocytic sarcoma in this analysis was 2 000 cells (1/6 mice transplanted, 3 independent experiments), which yielded a tumor 50 days after transplant (Table 1). No tumors were detected in recipients receiving 200 GFP+ cells (0/2 mice transplanted, Table 1). Based on these data, the average frequency of tumorpropagating cells in p16p19 2 /2; Kras(G12V) histiocytic sarcomas is 1/3 765 (confidence choice 95%, confidence intervals 1 870-7 578).
Secondary tumors exhibited histology similar to that of primary tumors and were GFP + and Mac2 + by immunohistochemistry (Fig. 5B). Weak staining for MPO was detected among tumor samples (Fig. 5B). Similar to primary tumor-bearing mice, the livers and spleens of secondary recipients showed extensive involvement by infiltrating myeloid cells (Fig. 5C, Fig. 4A). Secondary tumors showed a high proliferative index (80616.4% Ki67 + of all hematoxylin counter-stained cells, n = 5), and peripheral blood smears lacked involvement by histiocytic sarcoma (data not shown). Thus, p16p19 2/2 ; Kras(G12V) induced histiocytic sarcomas are highly transplantable in vivo.

Primary histiocytic sarcomas are oligo-clonal
The lentivirus-based strategy we employed to generate hematopoietic tumors has the inherent advantage that the transformed cells are uniquely marked by viral integration [32], allowing direct assessment of tumor clonality. We therefore asked if the p16p19 2/ 2 ; Kras(G12V) histiocytic sarcomas were initiated by a single clone or multiple clones of tumorigenic cells by analysis of proviral integration sites. Genomic DNA was extracted from both primary and secondary histiocytic sarcomas as well as from the BM of the same set of tumor-bearing mice and subjected to Ligation-Mediated PCR (LM-PCR) Assay [33]. Three PCR bands, on average, were observed in each of the primary tumor samples, indicating tumor oligoclonality (Fig. 6A). Sequencing of PCR bands identified distinct loci in the murine genome, thus validating that the LM-PCR products indicated bona fide proviral integration sites (Fig. 6A, Fig. S2). Of note, sequencing of LM-PCR products demonstrated that a single clone harboring an insertion in the proximity of Ribosomal Protein S29 (RPS29) was present in one set of matching primary and secondary transplanted tumors (bands 10, 17, 18, Fig. 6 and Fig. S2). These data suggest that a dominant clone of the tumorigenic cell population may have been selected during serial transplantation, explaining the shortened tumor latency in secondary transplant recipients. We also compared integration bands identified in tumors and bone marrow specimens from 7 individual mice. 7 out of 7 BM samples were polyclonal, and appeared to share subsets of the dominant clones with the muscle-resident tumors (Fig. 6B). However, we also detected additional novel bands in 4 out of 8 bone marrow samples (Fig. 6B, marked by white arrows), suggesting differences in selection forces on tumor cells in different microenvironments.

Histiocytic sarcoma can be induced using alternative leukemogenic strategies
The very high rate of induction of histiocytic sarcoma observed using p16p19 2/2 ; Kras(G12V) BM cells as a donor cell population prompted us to ask whether this might be an intrinsic property of this oncogenic combination, or whether histiocytic sarcomas can be introduced in NOD.SCID mice using other leukemogenic systems. We therefore acquired a leukemic cell sample in which hematopoietic malignancy was induced using the MLL-AF9 retroviral system [34]. MLL-AF9-expressing cells, marked by coexpression of Ds-Red, were harvested from secondary leukemic BM and transplanted into the pre-injured gastrocnemius muscles of NOD.SCID mice (Fig. S3A). All of these tertiary recipients developed tumors in the injected hindlimb within 16 days (15/15 mice injected, Fig. S1D and Table 1), while control mice injected with DS-Red labeled WT BM cells showed no tumor development (0/5 mice injected, Table 1).

Discussion
The work described here establishes a novel in vivo lentivirusinduced histiocytic/myeloid sarcoma model in immuno-compromised mice, combining ablation of the tumor suppressor gene locus p16p19 and ectopic expression of constitutively active oncogenic Kras(G12V). P16p19 2/2 ; Kras(G12V) tumor cells exhibit typical features of histiocytic sarcoma, including a predominant lack of expression of lymphocyte markers, positive expression of histiocyte/macrophage markers, and round to oval shaped cells with abundant, eosinophilic cytoplasm and nuclear atypia. P16p19 2/2 ; Kras(G12V) induced murine histiocytic sarcomas are aggressive neoplasms, and all tumor-bearing mice ultimately succumb to progressive leukemic symptoms [36].
The events that drive the formation of murine histiocytic sarcomas versus murine histiocytic leukemias as the first manifestation of a histiocytic neoplasm remain unclear. Murine histiocytic tumors induced by P16p19 2/2 ; Kras(G12V) tumor cells show evidence of a monocytic origin. A recent report showed that mice with coincident loss of Dok-1, Dok-2 and Dok-3 genes develop highly invasive and transplantable histiocytic sarcoma endogenously, and Dok-1/2/3 2/2 macrophages demonstrate enhanced proliferation ability [37], suggesting origination of the disease in monocytic cells. Histiocytic sarcoma also has been observed sporadically in pEm-Ras transgenic mice [38], and p16p19 2/2 mice develop histiocytic sarcoma with homozygous loss of Pten [39]. Deficiency of Pten leads to activation of Akt, as well as ERK1 and ERK2 in the histiocytic sarcoma cells, indicating hyperactivation of the Kras-MAPK pathway [39]. The majority (75%) of Pten 2/  [39]. In this regard, it is intriguing that in our study, we observed variable expression of the lymphoid marker CD8 in a subset of leukemic mice transplanted intravenously with p16p19 2/2 ; Kras(G12V) cells, whereas CD8 was less frequently expressed in histiocytic sarcomas generated by intramuscular transplantation of the same donor cell population. These data reinforce the previously suggested association of lymphomatous disease with histiocytic sarcoma [39], and suggest that the anatomical location of tumor origination may influence the manifestation of this biphenotypic pattern.
P16P19 proteins typically act as tumor suppressors for T cell and B cell malignancies [28], while Kras mutation induces myeloid leukemias [13]. Of note, in both the p16p19 2/2 ; Kras(G12V) and MLL-AF9 sarcoma models studied here, BM cells were cultured with a cytokine cocktail briefly (3 hr) after isolation and prior to intramuscular injection (see Materials and Methods). We cannot rule out the possibility that selective pressure from the cytokines possibly confers a myeloid bias on the engrafting cells [40]. Nonetheless, given prevalent in vivo data from murine leukemia models endogenously expressing either of the oncogenic combinations employed here [25,[41][42][43], it is reasonable to expect that culture conditions had only a minor impact on the lineage decision of the histiocytic tumor-initiating cells.
These studies demonstrate clear similarities between murine histiocytic sarcoma and murine histiocytic leukemia cells estab-lished by injection of the same oncogenetically modified BM cells in distinct anatomical locations (muscle vs. blood). A growing number of studies indicate that hematopoietic and leukemic cells functionally interact with a number of different ''niche'' cells in the BM, e.g., osteolineage cells, mesenchymal cells, reticular cells, endothelial cells, and adipocytes, as reviewed in [44]. These bidirectional interactions are mediated by an array of molecular and cellular signaling pathways, such that perturbations in microenvironmental regulators can influence both normal hematopoiesis and leukemic progression [45]. In particular, Wei et al reported that the lineage fate of the human Mll-AF9 leukemia cells in mice could be altered by manipulating growth signals or recipient strain [46]. Furthermore, mice with AML induced by co-expression of BCR/ABL and the Nup98/HoxA9 fusion protein showed a loss of osteolineage cells in the marrow, which may have contributed to the underlying pancytopenia [47]. Finally, ''niche''-specific deletion of the microRNA processing enzyme Dicer (using conditional ablation in mouse osteolineage cells) was shown to be sufficient to drive the development of a hematopoietic malignancy that requires this altered microenvironment for its continued propagation [48]. Yet, despite clear evidence of functional cross-talk between leukemic and niche cells in the marrow, much remains to be discovered about the role of the ''niche'' in the initiation and progression of extramedullary hematopoietic malignancy in vivo. Interestingly, the murine histiocytic sarcoma/leukemia models described here indicate a dominant influence of cell-autonomous, as opposed to microenvironmental, signals in the development of these malignancies. Whether arising initially in the skeletal muscle or hematopoietic tissue, the histiocytic sarcoma and leukemia cells induced by p16p19 deletion and Kras activation share nearly identical morphological, phenotypic, and histopathological features. Moreover, both neoplasms progress to systemic disease with similar kinetics and dissemination patterns. These findings deemphasize possible cell-non-autonomous effects of the microenvironment in which tumors are initiated as critical determinants of disease phenotype or progression, and highlight instead the strong influence of a cell-autonomous oncogenic program in specifying the emergence of these hematopoietic tumors.
The histiocytic sarcomas and leukemias reported here uniformly express high levels of CD47. Also known as Integrin Associated Protein (IAP), CD47 acts as a ''don't eat me'' signal, such that cells with high surface expression of CD47 escape integrin-mediated phagocytosis and death [49]. As previously reported, circulating hematopoietic stem cells, human and mouse myeloid leukemia cells, human bladder tumor-initiating cells, and multiple myeloma cells all express CD47 at an elevated level, and antagonistic treatment with anti-CD47 antibody both in vivo and ex vivo can induce remission of these cancers in mouse and xenograft models [50][51][52][53], suggesting that strategies targeting CD47 may provide promising therapeutic avenues in a variety of malignancies, regardless of the underlying oncogenetic lesions. As the histiocytic (myeloid) sarcomas studied here included a majority population of CD47 + tumor cells, the detection and targeting of CD47 in these tumors may present a useful therapeutic target. Related to this, recent studies have identified .20 distinct amino acid and glycosylation differences in the CD47 ligand SIRP-a (signal regulatory protein-a) in NOD.SCID mice (as compared to other mouse strains, including C57BL/6) [54]. SIRP-a is expressed by macrophages and inhibits their phagocytic function when bound by CD47 [55]. Protein coding as well as post-translational polymorphisms in SIRP-a correlate directly with the higher engraftment capacity in NOD.SCID mice of transplanted human hematopoietic stem cells [54]. Based on these findings, it is tempting to speculate that differences in the capacity for or consequences of recognition of tumor cell-expressed CD47 by SIRP-a-expressing macrophages in NOD.SCID versus C57BL/6 mice may contribute to the ability of intra-muscularly injected p16p19 2/2 ; Kras(G12V) tumor cells to generate histiocytic sarcomas in NOD.SCID but not C57BL/6 recipients, as reported here.
In summary, this work indicates that study of histiocytic (myeloid) sarcoma need not be limited to the few spontaneously emergent models currently available [37][38][39]. Through direct ex vivo modification and transplantation of oncogene-modified hematopoietic lineage cells, we established a rapid and reproducible system for the generation of this extramedullary tumor, and showed that this model recapitulates the natural progression of the disease in leukemia patients. Moreover, the tumor cells share highly similar features irrespective of the microenvironment at the site of initiation. Future studies using this model to uncover the molecular mechanisms that drive the establishment and dissemination of these malignancies will aid in the rapid diagnosis and effective treatment of these high risk hematopoietic tumors.

Materials and Methods
Mouse husbandry and breeding This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All animal experiments were approved by the Joslin Diabetes Center Institutional Animal Care and Use Committee (Protocol 04-01 to Amy Wagers). All surgeries were performed under anesthesia, and all efforts are made to minimize suffering. Animals were humanely sacrificed prior to tissue collection.

Preparation of lentivirus
The Kras(G12V)-IRES-GFP pGIPZ plasmid was a gift from Dr. Junhao Mao (University of Massachusetts, Worcester, MA). The Kras(G12V)-IRES-GFP pGIPZ plasmid (10 mg) or pGIPZ vector plasmid (10 mg, Open Biosystems, Rockford, IL), the HIV gag-pol-REV expression plasmid pCMV-dR8.91 (6.5 mg) and the envelope expression plasmid pMD2.VSV.G (3.5 mg) were co-transfected into 293 T cells with Fugene 6 (Roche Indianapolis, IN) on Day 1. HEK 293T cell line was a gift from former Gary D. Gilliland's lab [56]. Medium was changed every day for the next 3 days, and supernatant collected on Days 3 and 4. Supernatant was concentrated by ultracentrifugation at 20,000 rpm for 3 hours at 4uC (Beckman L7 ultracentrifuge). Lentivirus was titrated and then stored at 280uC for #3 weeks prior to use. Titration of lentivirus was performed using 293 T cells. The percentage of GFP+ cells at 72 hrs post infection was determined using flow cytometry.

Tumor induction
Lentivirally transduced cells were washed and resuspended in Hank's balanced salt solution (HBSS) containing 2% FBS. Cells were injected retro-orbitally using a 28G 1/2 insulin syringe, or into the gastrocnemius muscles of isofluorane anesthetized NOD.SCID mice using a transdermally inserted dental needle attached to a Hamilton syringe via polyethylene tubing. Recipient muscles were pre-injured 24 hours before cell implantation by injection of 25 ml of a 0.03 mg/ml solution of cardiotoxin (from Naja mossambica, Sigma, St. Louis, MO). Muscle pre-injury has been shown in previous studies [57] to enhance the engraftment of transplanted muscle precursor cells, although pre-injury was not required for induction of murine histiocytic sarcoma (see Table 1).
For secondary transplantation, tumor tissue was harvested from tumor-bearing mice, dissociated to generate single cell suspensions, and then GFP + sarcoma cells were isolated by Fluorescence Activated Cell Sorting (FACS) (see Cell Isolation Procedures for additional information). Recovered cells were suspended in HBSS with 2% FBS and injected into the cardiotoxin injured gastrocnemius muscles of secondary recipient mice, as described above.

Histology/Immunohistochemistry
Tissues were fixed in 4% paraformaldehyde for 3-5 hours and embedded in paraffin. Sections (4 micron) were stained with Hematoxylin & Eosin (H&E) or subjected to immunohistochemistry staining at the Specialized Histopathology Core Facility of the Dana-Farber/Harvard Cancer Center using the following antibodies: GFP (1:1500

Cell isolation procedures
Bone marrow cells were flushed from femurs and tibias with HBSS containing 2% FBS. Spleen cells were isolated by mechanical dissociation in HBSS supplemented with 2% FBS. Muscle tumors were digested in DMEM +0.2% collagenase type II (Invitrogen) for 90 minutes at 37uC in a shaking water bath, triturated to disrupt the remaining tumor pieces and filtered through a 70 mm cell strainer. Red blood cells were lysed from all cell preparations using ACK lysing buffer (Lonza, Hopkinton, MA). Cells were resuspended in HBSS with 2% FBS for subsequent procedure and analyses.

Analysis of Proviral integration sites
Proviral integration sites were isolated using ligation-mediated (LM)-PCR as described [60]. Briefly, 200 ng of genomic DNA was digested with Tsp509I (New England Biolabs, Ipswich, MA) and then subjected to linear amplification using the biotinylated primer Lenti LTRI (59-GAGCTCTCTGGCTAACTAGG-39). The labeled amplification product was then purified using streptavidin coated paramagnetic Dynabeads (Invitrogen, Carlsbad, CA) and was subsequently subjected to ligation with a blunt-ended linker cassette that had been synthesized by hybridization of the oligonucleotides OC1 (59-GACCCGGGAGATCTGAATT-CAGTGGCACAGCAGTTAGG- 39) and OC2 (59-CCTAACTGCTGTGCCACTGAATTCAGATC-39). The ligation product was amplified via two rounds of nested PCR, using the primers Lenti LTRII (59-AGCTTGCCTTGAGTGCTTCA-39) and OCI  in the first round of amplification; and Lenti LTRIII (59-AG-TAGTGTGTGCCCGTCTGT-39) and OCII (59-AGTGGCA-CAGCAGTTAGG-39) in the second round of amplification. Reaction conditions were as described [60]. Amplification products were analyzed by agarose gel electrophoresis. Selected amplification products were isolated using Gel Extraction Kit (Qiagen), and cloned into the pCR2.1-TOPO vector (TOPO TA cloning kit, Invitrogen). Proviral integration sites were subsequently sequenced using M13 forward and reverse primers and the results were aligned against the mouse genome using Basic Local Alignment Search Tool (BLAST) program (http://blast.ncbi.nlm. nih.gov/Blast.cgi).