Fig 1.
Exogenously expressed chimeric RNAs induce the expression of JAZF1-SUZ12 fusion transcripts found in endometrial stromal sarcoma.
(A) Model of RNA-mediated gene fusion in mammalian cells. A chimeric RNA invades chromosomal DNA of gene A and gene B in a sequence-dependent manner to stabilize a transient RNA/DNA hybrid. Resolving such an RNA/DNA hybrid by DNA breaks and rearrangements yields the final fusion gene A-B. Transcription of this fusion gene and subsequent RNA splicing produce a fusion RNA consisting of exons from both genes joined together by annotated splice sites. Note that for most known cancer fusion genes, the genomic break points are located in the introns, which is different from fusion RNA junction that is a splice junction. (B) Upper panel: chromosomal locations of JAZF1 and SUZ12 genes. Lower panel: schematics of the putative three-way junctions formed among JAZF1 and SUZ12 genomic DNA and the designed antisense chimeric RNA. In this model, the sense genomic strands of JAZF1 and SUZ12 genes form an imperfect DNA stem. The RNA/DNA hybrid formed between antisense chimeric RNA (in light and dark blue) and genomic DNA (in black) is indicated by shaded region. The chimeric RNAs are designed with 50 nts targeting JAZF1 gene and 50 nts targeting SUZ12 gene. See supplementary information for chimeric RNA designs. (C) RT-PCR detection of induced JAZF1-SUZ12 fusion transcript. Top panel: Nested primers designed to cross the JAZF1-SUZ12 fusion RNA junction. The JAZF1-SUZ12 fusion RNA found in endometrial stromal sarcoma has an RNA junction consist of JAZF1 exon 3 spliced to SUZ12 exon 2. Lower panels: RT-PCR detection of JAZF1-SUZ12 fusion transcript induced by chimeric RNAs in hESC cells. Thirteen out of fifty-one designed antisense chimeric RNAs led to clear induction of JAZF1-SUZ12 fusion transcripts amplified in a double-band pattern (asJS-1, -2, -8, -11, -14, -31, -36, -38, -39, -44, -45, -51, -53). Notation: the prefix ‘as’ stands for ‘antisense’, ‘JS’ stands for ‘JAZF1-SUZ12’, ‘M’ as DNA markers. As controls, transfection with a parental plasmid expressing mCherry (lane 52), cells without transfection (lane 53), and PCR reaction without cDNA (lane 54) all resulted in the absence of induced fusion transcripts. (D) Sanger sequencing confirmation of the induced JAZF1-SUZ12 fusion transcript. Each of the thirteen positive double bands was sequenced and results revealed that they contain the same JAZF1-SUZ12 RNA fusion junction sequence. An example of the RNA fusion junction sequence between JAZF1 exon-3 and SUZ12 exon-2 is shown. The full-length Sanger sequencing results are shown in S1 and S2 Figs. (E) The intron locations targeted by chimeric RNAs for the thirteen positive cases. These effective locations appeared to be scattered without a clustered hot spot. (F) RNA-driven gene fusion is both sequence-specific and cell type-specific. Chimeric RNA “asJS-53” designed to target JAZF1 and SUZ12 induced JAZF1-SUZ12 but not TMPRSS2-ERG in hESC cells (lane 3 and 4, upper vs. lower panel). Conversely, chimeric RNA “antisense-5” targeting TMPRSS2 and ERG induced TMPRSS2-ERG but not JAZF1-SUZ12 fusion in LNCaP cells (lane 6, upper vs. lower panel). Furthermore, JAZF1-SUZ12, an endometrial cancer fusion gene was induced in endometrial cells (hESC), but not in prostate cells (LNCaP) (lane 3 and 4 vs. lane 7 and 8). Conversely, TMPRSS2-ERG, a prostate cancer fusion gene, was induced in prostate cells, but not in endometrial cells (lane 6 vs. 2).
Fig 2.
Antisense chimeric RNAs but not their corresponding sense chimeric RNAs induce endogenous JAZF1-SUZ12 fusion transcripts.
(A) The thirteen positives antisense RNAs (prefix ‘as’) along with their corresponding sense chimeric RNAs (prefix ‘s’) were tested in parallel in hESC cells. RT-PCR results showed that all antisense were able to induce JAZF1-SUZ12 fusion transcript. In contrast, all corresponding sense chimeric RNAs failed to induce fusion transcripts (top panel). The experiment was repeated twice and results were identical. GAPDH serves as internal loading control (bottom panel). (B) The inability of sense chimeric RNAs to induce fusion transcripts is not due to the expression levels. The hESC cells were transfected with different amounts of plasmids (1.00μg, 0.75μg, 0.50μg and 0.25μg) expressing either sense or antisense chimeric RNA (middle panel). To maintain the same transfection protocol, mCherry plasmid was added to each transfection to make the final amount of plasmid to 1.0 μg. Experiment was performed for three antisense/sense pairs: asJS-8/sJS-8, asJS-14/sJS-14 and asJS-53/sJS-53. RT-PCR results showed that even when the sense chimeric RNA was intentionally expressed at a much higher level than the antisense chimeric RNA (middle panel, lane 5 vs. 3, 13 vs. 10, and 21 vs. 19), they failed to induce fusion transcripts (top panel, lane 5, 13 and 21). GAPDH serves as internal loading control (bottom panel). (C) A model that explains the disparity between antisense and sense chimeric RNA as the result of transcriptional conflict. Left panel: The antisense chimeric RNAs are able to form transiently stable DNA/RNA hybrids with sense strands of genomic DNA. Right panel: In contrast, the sense chimeric RNAs forming DNA/RNA hybrids with antisense strands of genomic DNA (the template strand used for transcription) are likely be “bumped” off by RNA polymerase and unable to stabilize the structures required for initiating genomic rearrangements. (D) RT-PCR results that support the transcriptional conflict model. The chimeric RNAs were expressed by U6 (a pol-III promoter) while α-amanitin was used to inhibit pol-II transcription of the parental genes for various time periods (0, 1, 2, and 3 days). α-amanitin was then rinsed off so that the newly induced fusion gene can express the fusion RNA. The induced fusion RNA was then assayed by RT-PCR at day 3. The sense chimeric RNAs that previously failed to induce fusion began to induce JAZF1-SUZ12 (lane 4, 11 and 12) after 3 days of α-amanitin treatment. GAPDH is used as internal loading control.
Fig 3.
Effects of estrogen and progesterone on JAZF1-SUZ12 fusion induction.
(A) Antisense chimeric RNAs were expressed in hESC cells followed by treatment with either estrogen (E2, 1μM) or progesterone (P, 1μM) or both (E2 1μM + P 1μM) for three days. RT-PCR results indicate that treatment with estrogen or progesterone consistently inhibited JAZF1-SUZ12 fusion induction in all the thirteen cases of antisense chimeric RNAs tested (None vs. E2 or P). However, the inhibitory effect was lessened in some cases when both hormones were combined (E2+P vs. E2 or P). (B) Hormone dosage effects on JAZF1-SUZ12 induction. RT-PCR results indicate that treatment with estrogen or progesterone at 10nM had no effect on JAZF1-SUZ12 induction. More effective hormone concentrations at 100nM or 1μM consistently inhibited the JAZF1-SUZ12 fusion transcripts induced by antisense chimeric RNAs tested (asJS-8 and asJS-53). (C) In the absence of antisense chimeric RNAs, estrogen or progesterone has little or no effect on parental JAZF1 gene activity in hESC cells (lane 1 vs, 2, 3, and 4), suggesting that the activity of JAZF1 promoter is not inhibited by estrogen or progesterone. Since the parental JAZF1 gene and the JAZF1-SUZ12 fusion gene have the same promoter, the reduced fusion transcript suggests that it is the process of JAZF1-SUZ12 fusion gene formation that is suppressed by estrogen or progesterone. The primers used to detect the JAZF1 mRNA are shown on the right. Notation: ‘E2’ for estrogen, ‘P’ for progesterone, ‘None’ for no treatment.
Fig 4.
Induced JAZF1-SUZ12 fusion is the result of genomic rearrangements.
(A) RT-PCR shows the transient nature of exogenously expressed chimeric RNA that was degraded and completely absent by day 60 (lower panel, lane 1 vs. 2), and the persistent nature of the induced fusion transcript (upper panel, lane 1 vs. 2) which was continuously expressed up to day 60 in the enriched cell population. This shows that the continuously expressed JAZF1-SUZ12 fusion RNA does not require the presence of chimeric RNAs. See S5 Fig for procedures used to propagate and enrich the induced hESC population. (B) The wild-type alleles with two identified genomic breakpoints marked as ‘x’ and ‘y’, and the primers used to amplify these breakpoints. The intron sizes are not presented in proportion as JAZF1 intron-3 (54kb) is much larger than SUZ12 intron-1 (2.7kb). (C) Schematics of the rearranged allele of the final fusion gene with JAZF1 and SUZ12 joined at the intron breakpoints. (D) The unrearranged wild-type JAZF1 allele near breakpoint ‘x’ was amplified by primer pair A/B (781 bp; lanes 1 and 4) and near breakpoint ‘y’ by primer M/N (877 bp; lanes 7 and 10). The unrearranged wild-type SUZ12 allele near breakpoint ‘x’ was amplified by primer pair C/D (722 bp; lanes 2 and 5) and near breakpoint ‘y’ by primer O/P (825 bp; lanes 8, and 11). The genomic fusion band ‘x’ (951 bp) revealed by fusion-specific primer pair A/D, and fusion band ‘y’ (976 bp) revealed by primer pair M/P, were present only in the enriched hESC population but absent in untransfected hESC cells (lane 6 vs. 3, and lane 12 vs. 9). See S6 Fig for multiplex primer designs used for initial scanning of potential breakpoints. (E) Sanger sequencing of the ‘x’ and ‘y’ fusion band identified the exact genomic breakpoints marked by dash lines. The genomic breakpoint ‘y’ contains a ‘AA’ insertion (marked by dash lines). The full-length Sanger sequences of 951 bp for “x” and 976 bp for “y” are shown in S7 and S8 Figs.