Nuclear expression of FLT1 and its ligand PGF in FUS-DDIT3 carrying myxoid liposarcomas suggests the existence of an intracrine signaling loop

Myxoid/round-cell liposarcoma (MLS/RCLS) is one of the most common forms of liposarcoma and accounts for about 40% of all cases [1]. The tumor cells are characterized by the FET family [2] FUS-DDIT3 fusion oncogene (also called TLS-CHOP) present in more than 90% of cases [3‐5] or the EWS-DDIT3 found in a minority of cases [6]. The causative role of FUS-DDIT3 in the initiation of MLS/RCLS and its role for the MLS-specific tumor morphology have been demonstrated in transgenic mice, xenografts and with FUS-DDIT3 carrying mesenchymal stem cells [7‐9].

FUS-DDIT3 encodes a protein consisting of the N-terminal half of the FUS protein juxtaposed to the DNA-binding basic leucine zipper transcription factor DDIT3 (also known as CHOP or GADD153) [4, 5]. The FUS-DDIT3 protein acts as an abnormal transcription factor [10] and the development of myxoid liposarcomas is thus regarded as a consequence of deregulated FUS-DDIT3 target genes [8, 9, 11]. In this study, we have investigated the expression of the putative FUS-DDIT3 target gene FLT1 and its encoded receptor tyrosine kinase in MLS cells.

Inspection of previously collected microarray data [8] identified FLT1 as a candidate downstream target gene for FUS-DDIT3. FLT1 encodes the Fms-like tyrosine kinase 1 (also called vascular endothelial growth factor receptor 1, VEGFR1) reported to be involved in autocrine growth stimulatory loops in other sarcoma types [12]. Here we report that HT1080 human fibrosarcoma cells stably transfected with FUS-DDIT3 express FLT1 transcripts more than 20 times stronger than the original HT1080 cells (Figure 1). In addition, DDIT3-transfected HT1080 cells showed minor but significant (< 3 times) increase in FLT1 expression whereas HT1080 cells transfected with the 5' part of FUS showed slightly downregulated FLT1 expression. Quantitative PCR (QPCR) analysis also revealed that three tested MLS-derived cell lines expressed high levels of FLT1 transcripts, further indicating that FLT1 is expressed in FUS-DDIT3 carrying cells (Figure 1). However, transient transfection of HT1080 cells with FUS-DDIT3 failed to induce an increased FLT1 expression during a 48 hour observation period (Additional file 1). This indicates that transcription of FLT1 is not a direct target of FUS-DDIT3. Instead, the change in FLT1 expression could be an indirect effect of FUS-DDIT3 expression. The FUS-DDIT3 protein has the capacity to reprogram HT1080 cells to a liposarcoma-like phenotype by hijacking parts of normal adipose tissue development [8, 16]. The induction of FLT1 transcription in FUS-DDIT3 expressing cells may thus be a part of this process. This explanation is further supported by QPCR analyses showing prominent expression of FLT1 transcripts in isolated normal adipocytes (Additional file 2).

Immunohistochemistry (IHC) analysis of human MLS tissues from nine cases and an MLS cell line xenografted in SCID mice showed strong, predominantly nuclear expression of the FLT1 protein (Figure 2a). The FLT1 nuclear stained tumor cells ranged from 60-80% in the nine investigated cases. No correlation between FLT1 staining and diagnosis (MLS or RCLS) were found. Immunofluorescence analysis of in vitro cultured MLS-derived cell lines 402-91 (Figure 2b) and 2645-94 (not shown), also showed a nuclear FLT1 localization. IHC staining of angiosarcoma tissue showed a cytoplasmic FLT1 staining in agreement with earlier reports [17] (Figure 2a). Indistinguishable results were obtained with two different antibodies for FLT1.

The nuclear localization of FLT1 in MLS/RCLS was surprising since the protein is commonly reported a cytoplasmic or plasma membrane bound protein [18]. To further probe the cellular distribution of the FLT1 protein, we prepared nuclear and cytoplasmic extracts from the MLS-derived cell line MLS 402-91. Western blot analysis of these extracts with an FLT1 specific antibody showed that anti-FLT1 reactive proteins were enriched in the nuclear fraction (Figure 2c). The western blot results thus confirmed the nuclear localization of FLT1 protein in MLS cells. The migration difference detected between nuclear and cytoplasmic FLT1 (Figure 2c) could be due to different ionic strengths of the extraction buffers used for fractionation but may well be a result of post-translational modifications of nuclear FLT1 proteins.

These results prompted a further investigation as to whether FLT1 has a nuclear localization in other tumor types and in normal tissues. IHC analysis of tissue arrays containing 59 normal and 60 tumor-derived tissues showed a strong nuclear FLT1 expression in more than 90% of the tumor cells in 1/1 pancreatic carcinoma and in 1/3 ovary carcinomas (Figure 3a). Nuclear staining was also observed in several normal cell types, indicating that this localization of FLT1 is not restricted to tumor cells (Figure 3a). Most notably, FLT1 antibodies stained the nuclei of adipocytes, suggesting that nuclear FLT1 expression is a normal feature of this cell type. In addition, FLT1 antibodies stained the nuclei of normal skin-derived cultured F470 fibroblasts but with a more prominent cytoplasmic staining compared to the MLS cells (Figure 3b). The observation that FLT1 has a nuclear localization in some tissues may indicate additional, yet unknown, functions for this protein. Nuclear expression of FLT1 has also been reported in endothelial cells and in these cells FLT1 activation was necessary for growth and survival [19]. Nuclear localization of other receptor tyrosine kinases and specific functions in this compartment has previously been reported [20‐23]. In MLS/RCLS and in normal adipocytes the nuclear expression may be important for maintenance of cell lineage/differentiation through phosphorylation/activation of specific chromatin components or transcription factors.

Further inspection of array data suggested that PGF, encoding placental growth factor (PGF), a ligand of FLT1, is transcribed in MLS cells. Quantitative RT-PCR analysis of MLS-derived cell lines showed that PGF expression was elevated in these cells compared with normal adipocytes (Figure 4a). The other FLT1 ligands VEGFA and VEGFB were expressed in cultured MLS cells but to a much lower degree than in adipocytes. In addition, the MLS cell lines showed PGF expression detected by immunofluorescence analysis (Figure 4b). In contrast to MLS cell lines, three MLS tumors showed a much more heterogenous expression of PGF, VEGFA and VEGFB but all tumors expressed PGF to some degree (Figure 4c). These results suggest differences of in vivo and in vitro ligand expression. Still, in light of our results, we hypothesize that autocrine circuits are operating in MLS cells. Such autocrine loops involving FLT1 have been reported to contribute to growth and survival of tumor cells [12, 24, 25] and to regulate differentiation of normal cell types [26, 27]. We tested for the presence of a possible FLT1 autocrine growth stimulatory loop in MLS cell line 402-91 by using an FLT1-blocking antibody reported to inhibit the growth of cells that are dependent on FLT1 signaling [12]. Incubation with this antibody failed however to affect growth or survival of the MLS cells. Though, receptor-targeted antibodies may be ineffective if the receptor and ligand interacts inside of the plasma membrane, which could be the case here. Such intracrine growth stimulatory circuits were recently reported for FLT1 in mammary carcinoma cells [28]. To test for an intracrine circuit in MLS/RCLS cells, knockdown experiments using siRNAs against FLT1 and PGF mRNA were attempted. As previously reported for knockdown experiments in MLS/RCLS cells [29], we were not able to achieve efficient knockdown of FLT1 or its ligand gene PGF. Instead we employed two specific inhibitors of FLT1 kinase activity (TKIs) in proliferation assays of MLS cell lines. None of the kinase inhibitors caused any decrease in cell numbers within the concentration range where they have shown specific inhibition of FLT1 activity [30, 31] (Figure 5). Thus, it is not likely that FLT1 and a probable autocrine/intracrine loop has a role in survival or proliferation in MLS cells but may contribute to the liposarcoma phenotype. Treatment with the TKIs gave no effect on the expression or localization of FLT1 in the MLS cells. Thus, the nuclear localization appears not to be related to the phosphorylation status of FLT1.

We conclude that upregulation of FLT1 in MLS/RCLS is an indirect downstream effect of the liposarcoma fusion oncogene FUS-DDIT3 and that FLT1 is expressed as a nuclear protein both in MLS/RCLS lipoblasts and in normal adipocytes. The nuclear expression of FLT1 in adipocytes and other normal cell types suggest that this receptor tyrosine kinase has cell type specific functions in the nuclear compartment. Most likely these involve maintenance of cell lineage through phosphorylation of target proteins residing in the nucleus. The expression of FLT1 ligand genes PGF, VEGFA and VEGFB in MLS/RCLS cells and in normal adipocytes further indicates the presence of possible autocrine or intracrine loops. The importance of FLT1, PGF, VEGFA and VEGFB expression and their functions in MLS/RCLS tumor development remain unclear but our results point out this receptor/ligand system as operational in MLS/RCLS and as such a potential target for intervention in this tumor type.