Gene expression profiles of MPCs containing ESFT fusion proteins, FLI-1 and ERG-1 display limited similarity

Transcriptional changes were analysed 36 hours and 4 days following infection, when expression of all of the fusion proteins and that of the corresponding wild type ets family proteins was comparable. No significant difference in the gene expression profile induced by the various fusion genes at 36 hours and 4 days was observed (data not shown). The number of genes observed to be induced and repressed differed among cells expressing the three fusion proteins, with a higher number of genes being affected by EWS-FLI-1 and FUS-ERG than by EWS-ERG. Using a 5% false discovery rate, cells expressing EWS-ERG displayed 67 repressed and 144 induced genes, whereas EWS-FLI-1 and FUS-ERG expressing cells had 199 and 172 repressed and 250 and 208 induced transcripts, respectively (Table S1, Figure 5).

Open in a separate window

Figure 5

Venn diagrams indicating the extent of overlap of the transcription profiles of MPCs expressing the three ESFT- associated fusion proteins, ERG-1, FLI-1, the EWS-FLI-1 R340N mutant and the AML-associated FUS-ERG fusion protein.

A white asterisk marks the areas where Igf1 is commonly present.

Consistent with the notion that a predominantly gene expression inducing transcriptional activity is associated with ESFT fusions [13], [14], pairwise comparison revealed greater similarity among the induced than among the repressed transcript repertoires (Figure 5). Thus, of all of the genes induced in cells infected with EWS-FLI-1 and EWS-ERG retroviruses, ~40% were shared whereas only ~13% of those that were repressed were common. EWS-ERG and FUS-ERG expressing cells shared about 21% of the induced but less than 12% of repressed transcript repertoire, similar to EWS-FLI-1 and FUS-ERG expressing cells that shared 22% of their induced and 10% of their repressed genes (Table S1 and Figure 5). Fifty-seven induced (14% of the total) and 13 repressed (4% of the total) transcripts were common to all three fusion proteins (Figure 5 and Table S1). The same comparison revealed somewhat greater similarity between EWS-FLI-1 and EWS-ERG induced genes than between genes induced by either fusion and those induced by FUS-ERG, suggesting a potentially important role of the N-terminal portion of the fusions in target gene selection. In contrast to the limited but nevertheless significant overlap of induced and repressed transcripts in response to the three fusion proteins, ERG-1 and FLI-1 shared very few common targets in MPCs. Whereas ERG-1 induced 122 and repressed 329 transcripts, and FLI-1 induced 290 and repressed 169 transcripts, only 14 of the induced (3.5%) and 8 of the repressed transcripts (1.6%) were common to the two transcription factors (Figure 5).

Similarity between the effects of FLI-1 and ERG-1 and those of their corresponding fusion proteins was also limited to a restricted number of genes (Figure 5). Thus, about 20% of all of the genes induced and repressed by EWS-ERG were also induced and repressed by ERG-1 alone and less than 8% of genes induced and repressed by EWS-FLI-1 were induced and repressed by FLI-1 alone.

The role of the DNA binding domain in the transcriptional activity of EWS-FLI-1 fusion protein was addressed by comparing the transcriptional activity of wt EWS-FLI-1 with that of the DBDM. The EWS-FLI-1 DBDM induced 98 and repressed 167 transcripts in MPCs compared to, respectively, 250 induced and 199 repressed transcripts by wt EWS-FLI-1, suggesting that impairment of EWS-FLI-1 DNA binding affects the transcriptional activation capacity of the fusion more strongly than its transcriptional repression capacity. Pairwise comparison revealed that only 65 out of a total of 649 (~10%) genes that were induced and repressed were shared (Figure 5).

Finally, comparison between the effects on MPCs of FUS-ERG fusions found in ESFT and AML revealed a similar number of induced and repressed transcripts. Of the total induced transcripts by the two fusion proteins, about 34% were common whereas about 22% of all of the repressed transcripts were shared between cells expressing the two fusions (Figure 5, Table S1).

Distinct and common candidate target genes of ESFT fusion proteins and their component ets family members

Expression of fifteen of the genes that were found to be significantly up- or down-regulated in MPC expressing EWS-FLI1, EWS-ERG and FUS-ERG was validated by quantitative real-time PCR in the corresponding cells as well as in FLI-1 and ERG-1 expressing counterparts (a complete list of up- and downregulated genes is provided in Table S2, whereas shared genes are listed in Table S3). DHH, PODXL, KRT119 and CDH5 were up-regulated in FLI-1 and ERG-1-expressing cells as well as in all three fusion protein expressing MPCs (Table 1). By contrast, CITED1, SFRP4, IGF1, IGFBP5, IGFBP3 and ENO3 were among genes that were upregulated in all three fusion protein containing cells but not in cells expressing FLI-1 and ERG-1 alone (Table 1). Several other genes, including DCN, MT2, DKK3, DKK2 and MMP3 displayed movement in one or two fusion protein containing cells but not in FLI-1 and ERG-1 retrovirus-infected cells.

Comparison of the expression of these same genes in cells expressing EWS-FLI-1 and the EWS-FLI-1 DBDM revealed that IGFBP3, PODXL, KRT119 and CDH5 were induced in both cell types, albeit substantially more strongly in wt EWS-FLI-1 expressors for all except IGFBP3 (Table 2). By contrast, DHH, IGF1, IGFBP5 and ENO3 were upregulated in wt EWS-FLI-1 expressing cells only. CITED1 was upregulated in both but more than 150 times more strongly in EWS-FLI-1 expressing cells (Table 2).

When the effects of the two FUS-ERG fusions in MPCs were compared, all of the genes tested displayed movement in the same direction. The only difference was that the degree of induction of each gene was 1.25 (DHH) to more than 10 (IGFBP3) times greater in cells expressing the AML-associated fusion (Table 3). It is conceivable that the stronger induction associated with the AML than with the ESFT fusion is a reflection of the higher expression level of the AML fusion protein that appears to be tolerated in MPCs.

Among the other putative gene targets, CITED1, which is a major discriminator of human ESFT with respect to other sarcomas [15] was strongly induced by the fusion proteins but only weakly by the DBDM. DHH and SFRP4 were both induced by wt fusion proteins but not by the DBDM, whereas PODXL, KRT119 and CDH5 were induced both by FLI-1 and ERG-1 alone and by the EWS-FLI-1 DBDM. These observations suggest that the DBDM conserves some transcriptional or transcription co-factor activity, consistent with the report that it maintains a degree of transforming ability in NIH3T3 cells [12].

EWS-FLI-1 recognizes the IGF1 promoter

To address the interaction between EWS-FLI-1 and the IGF1 promoter in vivo, chromatin immunoprecipitates (ChIP) from lysates of mouse MPCs expressing EWS-FLI-1, the DBDM and the empty vector were subjected to quantitative RT-PCR using primers specific for the region stretching from –2754 nucleotides to –2683 nucleotides upstream of the exon 1 start codon (ENSMUSE00000369489) of the murine IGF1 promoter (Figure 6A). This 70 bp stretch corresponds to the sequence annotated as –408 to –338 of the murine IGF1 promoter in a recent report by Alfieri et al. [16] and is included in the putative murine IGF1 promoter region GXP_41004 defined by the Genomatix (http://www.genomatix.de) genome annotation tool “Eldorado”. This region shows the highest similarity (96%) to the human IGF1 promoter region GXP_79580 and, by Genomatix analysis, is suggested to contain several potential trancription factor binding sites including ets binding sites. Relative to empty vector-expressing cells, a 5 fold increase in IGF1 promoter occupancy by the fusion protein was observed in ChIP assays directed against the anti-V5 epitope on the EWS/FLI-1 fusion protein.

Open in a separate window

Figure 6

In vivo interaction of EWS-FLI-1 with the region from –2754 to –2683 of the murine IGF1 promoter.

A) Quantitative RT-PCR analysis was performed on chromatin immunoprecipitates obtained from mouse MPCs expressing EWS-FLI-1, EWS-FLI-1 R340N mutant or an empty pMSCV vector. Primers specific for murine GAPDH DNA were used as input DNA control. Error bars reflect results of triplicate experiments. (B) IGF1 mRNA induction by EWS-FLI-1 and EWS-FLI-1 R340N in the MPCs used for the ChIP experiments. Quantitative RT-PCR was performed using a murine IGF1 TaqMAN probe (applied biosystems).

EWS-FLI-1 expressing cells displayed a 9 fold increase in IGF1 message as measured by quantitative real time PCR (Figure 6B). No IGF1 promoter occupancy was observed in chromatin immunoprecipitates from the DBDM expressing cells, which also displayed no induction of IGF1 transcripts (Figure 6). Similarly, no enrichment was observed using a second set of primers amplifying a region located about 14000 bp upstream of the IGF1 exon 1 start codon containing no putative ets binding sites (data not shown).

Cell cycle analysis

DNA content and cell cycle analysis were performed by flow cytometry. Cells were centrifuged, and the pellets were gently resuspended in 1.5mL hypotonic propidium iodide solution (PI, 50 µg/mL in 0.1% sodium citrate plus 0.1% Triton X-100; Sigma). The tubes were kept at 4°C in the dark overnight. The PI-fluorescence of individual nuclei was measured by flow cytometry using standard FACScan equipment (Becton Dickinson, Mountain View, CA).

Constructs and cDNA Cloning

A cDNA clone encoding the type 2 EWS-FLI-1 fusion protein was amplified from the SK-N-MC Ewing sarcoma cell line by RT-PCR. The amplified fragment carrying an in frame v5 tag at the 3′extremity was digested with XhoI and Hpa I and inserted into the pMSCV Puro retroviral expression vector (BD Biosciences Clontech, Palo Alto, CA).

The EWS-FLI-1 R340N DBDM was amplified by PCR using the wt sequence as template, with the following primers:

hEWS forward XhoI:

5′CCGCTCGAGCCACCATGGCGTCCACGGATTACAG 3′;

hFLI-1 reverse R340N:

5′ ATCATAGTAATAATTGAGGGCCCGGCTCAGCTTGTC 3′;

hFLI-1 forward R340N:

5′ CGACAAGCTGAGCCGGGCCCTCAATTATTACTATGA 3′;

V5 reverse Hpa I (including a stop codon):

5′ GTTAACTCACGTAGAATCGAGACCGAGGAGAGGGTTAGGGATAGGCTTACC 3′.

The amplified fragment was XhoI-HpaI-digested and inserted into the pMSCVPuro retroviral vector.

EWS-ERG and ERG-1 fragments were amplified from surgically resected Ewing's sarcoma tissue by RT-PCR.

Amplified fragments, carrying a BglII restriction site at the 5′end were inserted, using the pcDNA3.1/V5-HisTOPO TA cloning kit (Invitrogen, Carlsbad, CA), into the corresponding expression vector in frame with the V5 epitope tag. The EWS-ERG-v5 and the ERG-1-v5 fragments were digested with BglII and PmeI and inserted into the BglII/HpaI-digested pMSCV Puro vector. Human FLI-1 cDNA was amplified from hMSC RNA and inserted into EcoRI/XhoI-digested pMSCV Puro. cDNA clones encoding the human FUS-ERG AML and FUS-ERG ESFT fusions were obtained by PCR and inserted into BglII/HpaI-digested pMSCV Puro. Fragments encoding exons 1-7 or exons 1-6 of human TLS/FUS, amplified from the EST clone IMAGE:3345294, and a fragment encoding the C-terminal portion, starting at exon 9, of human ERG, amplified from the EWS-ERG-v5 clone, were used in the PCR reactions.

The primer sequences were:

External ERG-v5-reverse 5′GTTAGGGATAGGCTTACCTTCGAACCGCGGGCCGTAGTAAGTGCCCAG 3′

External BglII FUS forward

5′GGAAGATCTTCCCCACCATGGCCTCAAACGATTATACC 3′

Internal for FUS-ERG AML:

Forward5′GCTTCAATAAATTTGGTGGCAGTGGCCAGATCCAGCTTTGGCAGTTCC3′

Reverse 5′AGCTGGATCTGGCCACTGCCACCAAATTTATTGAAGCCACCAC3′

Internal for FUS-ERG ESFT:

Forward 5′GCCGTGGAGGCAGAGGCAGTGGCCAGATCCAGCTTTG 3′

Reverse 5′CAAAGCTGGATCTGGCCACTGCCTCTGCCTCCACGGCCACCT 3′

A cDNA control encoding the first 262 amino acids of the human TLS/FUS protein followed by a tail of 54 random amino acids was obtained by PCR and inserted into pMSCV Puro. The 54 random amino acid tail displayed no homology in a BLAST search and contained none of the motifs present in the ERG-1 or FLI-1 proteins as assessed by the SMART- Simple Modular Architecture Research Tool [http://smart.embl-heidelberg.de]. All constructs were sequence-verified.

Retrovirus generation and infection

Expression of EWS-FLI-1, EWS-ERG, FUS-ERG (AML), FUS-ERG (ESFT) , ERG-1 and FLI-1 in MPCs was achieved using the Retroviral Gene Transfer and Expression (BD Biosciences Clontech), according to the manufacturer's recommendations. Infected cells were selected in 1.5 µg/ml puromycin for a maximum of 14 days.

RNA and genomic DNA extraction and Reverse Transcription PCR

Total RNA was extracted using an RNeasy Mini Kit (Qiagen, Hilden, Germany) and genomic DNA was isolated using a DNeasy Tissue Kit (Qiagen) according to the manufacturer's recommendations. RT-PCR was performed using Super Script one step RT-PCR with the platinum Taq kit (Invitrogen, Carlsbad, CA).

Western blot analysis

Cell lysis, SDS-PAGE and blotting were done according to standard procedures and immunodetection was performed using chemoluminescent substrate kits from Amersham according to the manufacturer's recommendations. Primary antibodies used were: monoclonal mouse anti-V5epitope antibody (Invitrogen), mouse anti human FLI-1 monoclonal antibody (G146-222 BD- Pharmingen, San Diego, CA), monoclonal mouse anti-actin (Sigma) and mouse IgG (Sigma). The goat anti-mouse HRP conjugated secondary antibody was from Amersham. Densitometric analysis was performed using the public domain image processing and analysis program NIH image J (http://rsb.info.nih.gov/ij/).

cDNA array hybridization and cDNA array analysis

For each comparison 3 independently isolated MPC populations were used and 3 independent infections were performed. Total RNA was extracted at 36 hours following infection. The quality of RNA was verified by an Agilent RNA 600 nanoassay and by measuring the 260/280 absorbance ratio. Quality-tested total RNA was amplified using the RiboAmp RNA Amplification Kit (Arcturus, Moutainview, CA) and processed using a reverse transcription based method of label incorporation to yield labeled cDNA as previously described [8]. Following hybridization and washing, microarrays were imaged using a DNA microarray scanner (agilent technologies) and array analysis and quality control were performed as described in [8].

Statistical analysis of the expression data

RNA derived from EWS-FLI-1, EWS-ERG, FUS-ERG (AML), FUS-ERG (ESFT), ERG-1 and FLI-1 expressing MPC populations were all subjected to microarray analysis; five m17k microarrays (among which 2 were dye swaps) were performed for each population, comparing cells expressing the protein of interest with the corresponding empty vector control cells. Differentially expressed genes were then identified with the rank product method [38] as implemented in the “RankProd” Bioconductor package [39]. We retained for further analysis the clones identified as differentially expressed with a false discovery rate of 5%. The statistical significance of the overlap between differentially expressed genes in different translocations was evaluated with Fisher's exact test.

Real time quantitative RT-PCR

cDNA was obtained using an M-MLV reverse transcriptase and RNAse H minus (Promega). Typically 500 ng of template total RNA and 250 ng of random hexamers were used per reaction. Real time-PCR amplification was performed in an ABI Prism 7700 instrument (Applied Biosystems), using Taq Man Universal PCR mastermix and Assays-On-Demand Taq Man probes for IGF1, IGFBP5, IGFBP3, SFRP4, DKK2, DCN, MT2 and MMP3.

For real time quantitation of CITED1, DHH, PODXL, KRT1-19, CDH5, ENO3 and DKK3 RNAs, the Universal Probe Library system (Roche Rotkreuz, Switzerland) was used and primers were designed according to the ProbeFinder software (http://www.roche-applied-science.com).

The following primers were designed:

ENO3:

forward5′CCGGGAAATCCTGGACTC3′

reverse 5′CAGCTGCTCGGAATCGAC3′

CDH5:

forward5′TCACCTTCTGTGAGGAGATGG3′

reverse 5′GATGATCAGCAAGGTAATCACTGT3′

PODXL:

forward5′GGATCTCCCAGAGGAAGGAC3′

reverse 5′CAAGGTTGGGTTGTCATGGT3′

DKK3:

forward5′CTCAATGAGATGTTTCGAGAGGT3′

reverse 5′ CTTCTTCCGCCTCCATCTC 3′

CITED1:

forward5′CTAGGTCGCTTCGTCCGTA3′

reverse 5′AGCTGGGCCTGTTGGTCT3′

KRT1-19:

forward5′TGGAGATGCAGATTGAGAGC3′

reverse 5′TCCTCAGGGCAGTAATTTCC3′

For GAPDH amplification, rodent GAPDH control reagent (Applied Biosystems) was used. Relative quantitation of target, normalized with an endogenous control (GAPDH), was performed using a comparative Ct method (Applied Biosystems).

IGF1 promoter analysis and cloning

In silico IGF1 promoter analysis was performed using the Genomatix (http://www.genomatix.de) genome annotation tool “Eldorado”. Specifically, GEMS analysis was used with both MatInspector and ModelInspector tools to search for putative transcription factor binding sites and sequence models [22]. Comparative genomix was also used.

The human IGF1 promoter sequence from position –1682 to +1 (start codon on exon 1) was amplified from hMSC genomic DNA using primers: 5′CCCAAGCTTGGGTGCTTCTGAAGTACAAAGTCTG3′ and 5′GAAGATCTTCAAATGTTGCTGAACATAGTGCACCATTG3′.

The amplified fragment was ligated into the BglII and HindIII cloning sites of the pGL3-basic vector (promega) and sequence-verified. Smaller IGF1 promoter fragments (−1098, −913 and –673) were obtained from the –1682pGL3 construct by digestion and re-ligated into the pGL3-basic vector. Restriction enzyme digestion was as follows: HindIII/XhoI to obtain the −1098 fragment, : HindIII/KpnI to obtain the –913 fragment and HindIII/AvaI to obtain the –673 fragment.

Transient transfection and luciferase assays

2 µg of empty pMSCV plasmid or pMSCV encoding human FLI-1, ERG-1, EWS-FLI-1, EWS-ERG, FUS-ERG, EWS-WT1 and the EWS-FLI-1 R340N DBD mutant were introduced by nucleofection along with 2 µg of a pGL3 basic vector or the pGL3-IGF1 promoter construct (−1682pGL3) into freshly isolated human mesenchymal stem cells using a nucleofector II device and the appropriate solution according to manufacturer's recommendation (Amaxa GmbH, Koeln, Germany). For promoter region analysis, 2 µg of pMSCV encoding human EWS-FLI-1 were introduced by nucleofection along with 2 µg of a pGL3 basic vector or one of the pGL3-IGF1 promoter constructs. General transfection efficiency was controlled using a pMAX-GFP control vector (AMAXA) or a pGL3 promoter vector containing a SV40 promoter. A 75% efficiency was calculated in MSCs using nucleofection. Transfection efficiency for both fusion proteins and luciferase reporter plasmid was more specifically controlled by PCR or RT-PCR performed on RNA or DNA using primers specific for either luciferase or each of the fusion genes. Luciferase activity was measured 48 hours later on cleared cell lysates using the luciferase assay system (Promega) according to manufacturer's recommendations. 3 independent experiments were performed using 2 different batches of hMSCs. Reporter luciferase activity was measured in triplicate, mean values were corrected for total cell number and reported as relative values to vector transfected cells. Total protein normalization was not possible due to the presence in the luciferase kit lysis buffer of components that disrupt common total protein measurement assays. Nevertheless parallel experiments were performed with fractions of the same cell populations lysed in appropriate buffer to control the correspondence between cell number and total proteins.

Statistical significance was determined with a 2 sided t-test on the logarithm of the relative activity.

We thank Carlo Fusco and Marina Bacac for helpful comments, Prof. Pierre Hoffmeyer for fruitful collaboration and Karine Baumer for technical support. All animal experiments were performed in compliance with the regulations on animal experimentation of the Canton de Vaud, permit #1477.1